Orthogonally redundant sensor systems and methods

ORTHOGONALLY REDUNDANT SENSOR SYSTEMS AND METHODS

FIELD OF THE INVENTION

Embodiments of the present invention relate generally to sensor technology, including sensors used for sensing a variety of physiological parameters, e.g., glucose concentration.

More particularly, embodiments of the invention relate to redundant sensors and sensor systems, as well as methods of making and using such sensors and sensor systems. More particularly still, embodiments of the invention relate to orthogonally redundant glucose sensors and sensor systems, to methods of making and using such sensors and sensor systems, 20 to closed-loop insulin-infusion systems that employ orthogonally redundant glucose sensors and sensor systems, and to methods of making and using such closed-loop systems.

BACKGROUND

The pancreas of a normal healthy person produces and releases insulin into the blood stream in response to elevated blood plasma glucose levels. Beta cells (β-cells), which reside 25 in the pancreas, produce and secrete the insulin into the blood stream, as it is needed. If β-cells become incapacitated or die (Type I diabetes mellitus), or in some cases, if β-cells produce

CA 2845804 2018-04-10

WO 2013/036493

PCT7US2012/053707 insufficient quantities of insulin (Type II diabetes), then insulin must be provided to the body from another source.

Traditionally, since insulin cannot be taken orally, insulin has been injected with a syringe. More recently, the use of infusion pump therapy has been increasing, especially for 5 delivering insulin for diabetics, for example, external infusion pumps are worn on a belt, in a pocket, or the like, and deliver insulin into the body via an infusion tube with a percutaneous needle or a cannula placed in the subcutaneous tissue. Physicians have recognized that continuous infusion provides greater control of a diabetic’s condition, and are increasingly prescribing it for patients.

Infusion pump devices and systems are relatively well-known in the medical arts for use in delivering or dispensing a prescribed medication, such as insulin, to a patient. In one form, such devices comprise a relatively compact pump housing adapted to receive a syringe or reservoir carrying a prescribed medication for administration to the patient through infusion tubing and an associated catheter or infusion set. Programmable controls can operate the infusion pump continuously or at periodic intervals to obtain a closely controlled and accurate delivery of the medication over an extended period of time. Such infusion pumps are used to administer insulin and other medications, with exemplary pump constructions being shown and described in U.S. Patent Nos. 4,562,751; 4,678,408; 4,685,903; 5,080,653; and 5,097,122.

There is a baseline insulin need for each body which, in diabetic individuals, may generally be maintained by administration of a basal amount of insulin to the patient on a continual, or continuous, basis using infusion pumps. However, when additional glucose (i.e., beyond the basal level) appears in a diabetic individual’s body, such as, for example, when the individual consumes a meal, the amount and timing of the insulin to be administered must be determined so as to adequately account for the additional glucose while, at the same time, avoiding infusion of too much insulin. Typically, a bolus amount of insulin is administered to

CA 2845804 2018-04-10

WO 2013/036493

PCT/US/2012/053707 compensate for meals (i.e., meal bolus). It is common for diabetics to determine the amount of insulin that they may need to cover an anticipated meal based on carbohydrate content of the meal.

Over the years, a variety of electrochemical glucose sensors have been developed for use in obtaining an indication of blood glucose levels in a diabetic patient. Such readings are useful in monitoring and/or adjusting a treatment regimen which typically includes the regular administration of insulin to the patient. Generally, small and flexible electrochemical sensors can be used to obtain periodic readings over an extended period of time. In one form, flexible subcutaneous sensors are constructed in accordance with thin film mask techniques. Typical thin film sensors are described in commonly assigned U.S. Pat. Nos. 5,390,671; 5,391,250; 5,482,473; and 5,586,553. See also U.S. Pat. No. 5,299,571.

These electrochemical sensors have been applied in a telemetered characteristic monitor system. As described, e.g., in commonly-assigned U.S. Pat. No. 6,809,653, the telemetered system includes a remotely located data receiving device, a sensor for producing signals indicative of a characteristic of a user, and a transmitter device for processing signals received from the sensor and for wirelessly transmitting the processed signals to the remotely located data receiving device. The data receiving device may be a characteristic monitor, a data receiver that provides data to another device, an RF programmer, a medication delivery device (such as an infusion pump), or the like.

Regardless of whether the data receiving device (e.g., a glucose monitor), the transmitter device, and the sensor (e.g., a glucose sensor) communicate wirelessly or via an electrical wire connection, a characteristic monitoring system of the type described above is of practical use only after it has been calibrated based on the unique characteristics of the individual user. Accordingly, the user is required to externally calibrate the sensor. More

CA 2845804 2018-04-10

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 specifically, a diabetic patient is required to utilize a finger-stick blood glucose meter reading an average of two - four times per day for the duration that the characteristic monitor system is used. Each time, blood is drawn from the user’s finger and analyzed by the blood glucose meter to provide a real-time blood sugar level for the user. The user then inputs this data into the glucose monitor as the user’s current blood sugar level which is used to calibrate the glucose monitoring system.

Such external calibrations, however, are disadvantageous for various reasons. For example, blood glucose meters include inherent margins of error and only provide discrete readings at one point in time per use. Moreover, even if completely accurate, blood glucose meters are cumbersome to use (e.g., one should not operate an automobile and take a finger stick meter reading at the same time) and are also susceptible to improper use. Furthermore, there is a cost, not to mention pain and discomfort, associated with each application of the finger stick. Thus, finger stick replacement remains a goal for the next generation of glucose monitoring systems.

As sensor technology improves, there is greater desire to use the sensor values to control the infusion of insulin in a closed-loop system (i.e., an artificial pancreas system). Specifically, a closed-loop system for diabetes includes a glucose sensor and an insulin infusion pump attached to the patient, wherein the delivery of insulin is automatically administered by the controller of the infusion pump--rather than by the user/patient-based on the sensor’s glucose value readings. The benefits of a closed-loop system are several-fold, including tighter glycemic control during the night when the majority of hypoglycemic events occur.

An accurate and reliable sensor has long been identified as a necessity for closed-loop realization. Glucose sensor technology has been evolving in an effort to meet the accuracy required for fingerstick replacement and the reliability needed for consistent closed-loop

CA 02845804 2014-02-18

WO 2013/036493

PCT/US2012/053707 functionality. Several types of technology are available, with two of the most common and developed being electrochemical sensing, as noted above, and optical sensing. See Table 1.

Table 1: Existing glucose sensor technologies, benefits, and drawbacks

<td> Sensing Technology</td><td colspan="2"> Details</td><td> Benefits</td><td> Drawbacks</td>

<td rowspan="4"> Electrochemical</td><td rowspan="2"> Enzymatic</td><td> First generation</td><td> High glucose specificity and sensitivity</td><td> Interferences, high potential, added outer membranes effect response time</td>

<td> Second and third generation</td><td> High specificity, low overpotential prevents interferences</td><td> Toxic mediators, competition between mediators and oxygen, repeatability</td>

<td colspan="2"> Non-GO, based</td><td> No interferences from oxygen</td><td> Can oxidize other substances</td>

<td colspan="2"> Nonenzymatic</td><td> No enzyme degradation</td><td> Not specific to glucose</td>

<td rowspan="8"> Optical</td><td rowspan="3"> Fluoropltorebased</td><td> Fluorescence or FRET intensity</td><td> Highly specific to glucose due to flurophore with glucose binding</td><td> Photobleaching, dependant on skin pigmentation and thickness</td>

<td> FRET lifetime</td><td> Independent of scattering and fluorophore concentration</td><td> Miniaturization of instrumentation difficult</td>

<td> Ocular spectroscopy</td><td> Noninvasive, uses tears to measure glucose visually</td><td> Leaching of chemicals, effected by pH and ionic strength, lag time</td>

<td rowspan="5"> Nonfluorophore based</td><td> Optical coherence tomography</td><td> Not affected by urea, ionic strength, lemiierature, heart rate, and hematocrit</td><td> Affected by motion and tissue heterogeneity</td>

<td> Polarimetry</td><td> Can use visible light, can be miniaturized</td><td> Effected by scattering in the tissue, pH. and temperature, lack of specificity</td>

<td> Thermal infrared spectroscopy</td><td> Can use visible light, can be miniaturized</td><td> Effected by scattering in the tissue, pH. probe position, fever, and temperature</td>

<td> Photoacoustic spectroscopy</td><td> Not affected by ionic strength or albumins</td><td> Effected by scattering in the tissue, miniaturization difficult</td>

<td> Raman spectroscopy</td><td> No interference from luminescence and fluorescence</td><td> Longer stabilization times, effected by tissue density, thickness, hematocrit</td>

<td rowspan="2"> Combinatorial</td><td colspan="2"> 1 mpedance spectroscopy</td><td> Can measure glucose levels in the vascular compartment, no lag time in sensor response</td><td> Temperature, disease state may affect measurements, changes in properties not specific to glucose</td>

<td colspan="2"> Electromagnetic spectroscopy</td><td> Can measure glucose levels In the vascular compartment, no lag time In sensor response</td><td> Body temperature, sweating and motion affect glucose measurements</td>

To offer improved performance, the possibility of redundant electrodes has been explored and shown to provide a benefit. For example, previous studies in the literature have reported using two implanted glucose electrodes to simultaneously monitor glucose levels in rat tissue combined with a signal processing algorithm. These studies demonstrated that the overall glucose measurement accuracy could be improved over that of a single sensor. 10 However, while it may provide for improved accuracy, such simple redundancy may not provide the reliability necessary for closed-loop applications.

Since the closed-loop system replaces the patient as the decision-making element, a reliable system must typically deliver reliable data and have error detecting functionality, enabling the closed-loop system to take action on erroneous data. Such data may be caused by

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 drift, noise, or temporary or permanent malfunction of the sensor, often due to the implanted environment’s effect on sensors. Actions may vary from simply prompting the patient to calibrate the system to terminating the sensor and requesting insertion of a new sensor. With identical sensor configurations, the redundant elements are similarly affected by environmental conditions and therefore could simultaneously present erroneous data.

Thus, although recent advances in continuous glucose monitoring (CGM) technology have offered several benefits for easier and more effective glycemic control in diabetes management, further improvements such as improved sensor accuracy and reliability, reduced number of blood glucose calibrations, improved specificity, and improved comfort during sensor insertion and wear are still desirable.

SUMMARY

In accordance with an embodiment of the invention, a continuous glucose monitoring system includes a hand-held monitor having a display, an external transmitter, an insulin pump, and an orthogonally redundant glucose sensor. The orthogonally redundant glucose sensor includes an optical glucose sensor and a non-optical glucose sensor, which may be an electrochemical glucose sensor. Moreover, the electrochemical sensor may have a distributedelectrode design. Each of the optical and non-optical sensors has a distal portion that is configured to be placed inside a user’s body, and a proximal portion that remains external to the user’s body. The distal portions of the optical and non-optical sensors are deployed simultaneously via an insertion needle, and may be co-located within the user’s body.

In an embodiment of the invention, the optical glucose sensor includes an optical fiber with a glucose-permeable membrane joined to its distal end. The membrane may be, e.g., tube-shaped, such that its a hollow interior defines a compartment for holding an assay. In one aspect of the invention, the assay is a competitive glucose binding affinity assay that includes a glucose receptor, a glucose analog, a first fluorophore labeled onto the glucose receptor, and an acceptor dye labeled onto the glucose analog. In a variation of this aspect of the invention, in

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 addition to the first fluorophore, the assay may include a reference fluorophore which serves, inter alia, as a sensor diagnostic tool.

In embodiments of the invention, the external transmitter has a housing that houses the proximal portion of the optical glucose sensor and the proximal portion of the non-optical glucose sensor. The respective proximal portions of the sensors are operatively coupled to instrumentation for receiving and processing respective signals from the optical and nonoptical sensors, including converting each of the optical and non-optical signals to respective glucose values. The transmitter also includes additional instrumentation for wirelessly communicating the respective glucose values to the hand-held monitor, the insulin pump, etc.

In embodiments of the invention, as part of the instrumentation, the transmitter houses an optical system for lifetime and/or intensity interrogation of the assay contained in the optical sensor. For example, a fluorophore-labeled assay may be interrogated by an optical interrogating system including a light source and a filter substrate having one or more coatings to effect, e.g., an excitation filter and/or an emission filter. In one aspect of the invention, the interrogating system may be manufactured as a wafer-scale stacked planar integrated optical system (SPIOS) and diced into smaller units. In another aspect, the light source may be either a LED, or a red laser diode, with the latter enabling a substantial reduction in the size and volume of the transmitter.

The above features and aspects may also be operationalized in closed-loop systems, with predictive diagnostics and minimal requirements for external calibration.

Other features and advantages of the invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings which illustrate, by way of example, various features of embodiments of the invention.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

BRIEF DESCRIPTION OF THE DRAWINGS

Figs. IA and IB show continuous glucose monitoring systems for orthogonally redundant sensing in accordance with embodiments of the invention.

Fig. 2 shows a system-based approach to targeted electrochemical sensor improvements.

Figs. 3A-3C show a capsule-based optical sensor implanted under the skin in accordance with an embodiment of the invention.

Fig. 4 shows a glucose binding competitive affinity fluorophore-labeled assay, including an internal reference, in accordance with embodiments of the invention.

Fig. 5 shows an optical interrogating system for interrogating a fluorophore-labeled assay with an internal reference used for intensity measurement in accordance with an embodiment of the invention.

Fig. 6A shows the various equilibria and the non-glucose consuming feature of an optical glucose sensor in accordance with embodiments of the invention.

Fig. 6B shows, in accordance with an embodiment of the invention, the use of a reference fluorophore, as a diagnostic tool for an optical sensor, indicating when, e.g., the integrity of the membrane may have been compromised or the optical connection may have been misaligned.

Fig. 7 shows a plurality of sensor electrodes distributed along the length of an electrochemical sensor in accordance with an embodiment of the invention.

Fig. 8A shows, in accordance with an embodiment of the invention, a side view of an optical fiber sensor containing an assay within a membrane coupled to the fiber’s distal end, with excitation light entering, and fluorescence leaving, the fiber.

Fig. 8B shows the optical fiber glucose sensor of Fig. 8A, with the details of the assay shown, in accordance with an embodiment of the invention.

WO 2013/036493

PCT/US2012/053707

Fig. 9A is a sectional view of a transmitter having a dual connector for connecting to both an electrochemical sensor and an optical sensor in accordance with embodiments of the invention.

Fig. 9B is a sectional view of a transmitter, with an optical connection, an electrical contact, and co-located deployment of an electrochemical sensor and an optical sensor in accordance with embodiments of the invention.

Fig. 9C shows a sectional view of an integrated flex circuit in accordance with embodiments of the invention.

Fig. 10 is a side view of a needle for housing and simultaneously deploying both an electrochemical sensor and an optical sensor in accordance with embodiments of the invention.

Fig. 11 shows a graphical illustration of an error-check feature based on a meter value obtained from a hand-held monitor with integrated meter in accordance with embodiments of the invention.

Fig. 12 shows theoretical response functions for an optical equilibrium glucose sensor and an electrochemical glucose sensor in connection with embodiments of the invention.

Figs. 13A and I3B show algorithms for analyzing signals and performing diagnostics to assess reliability of individual signals and assign weights through calibration in accordance with embodiments of the invention.

Fig. 14 shows a two compartment model utilized in algorithms for transforming sensor signals into blood glucose values in accordance with embodiments of the invention.

Fig. ISA is an illustration of improving sensor accuracy through assessing each individual sensor current with its reliability index in accordance with embodiments of the invention.

Fig. I SB is an illustration of improving sensor accuracy through creating a weighted average in accordance with embodiments of the invention.

Fig. 16 shows a comparison of calibrations frequency vs. time between existing systems (a) and embodiments of the present invention (b).

CA 2845804 2018-11-29

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Figs. I7A and 17B show the use of electrochemical impedance spectroscopy in detecting a drop in low frequency Nyquist slope (a), which predicts a drift in sensor signal (b), in accordance with embodiments of the invention.

Fig. 17C illustrates predictive diagnostics proactively identifying sensor anomalies for improved reliability' in accordance with embodiments of the invention.

Fig. 18 shows an optical system having discrete components (left), and a stacked planar integrated optical system (right) in accordance with embodiments of the invention.

Fig. 19 shows illustrative layers of a wafer-scale stacked planar integrated optical system (SPIOS) in accordance with embodiments of the invention.

Fig. 20 illustrates the addition of key optical sensor electronic components to an analog front-end for electrochemical sensing in accordance with embodiments of the invention.

Fig. 21 shows wavelength ranges for three fluorophores which may be used with a laser diode source at 645 nm in accordance with embodiments of the invention.

Fig. 22 shows a care network using various components and methodologies in accordance with embodiments of the invention.

DETAILED DESCRIPTION

In the following description, reference is made to the accompanying drawings which form a part hereof and which illustrate several embodiments of the present invention. It is understood that other embodiments may be utilized and structural and operational changes may be made without departing from the scope of the present invention.

As shown in the drawings for purposes of illustration, embodiments of the invention are directed to sensors that may be introduced and/or lodged transdermally, or may be implanted in and/or through subcutaneous, dermal, sub-dermal, inter-peritoneal, or peritoneal tissue. In the discussion herein, preferred embodiments of the devices, systems, and methods of the invention are described with reference to glucose as the analyte whose level/concentration in the blood and/or bodily fluids of the user is to be determined. However, this is by way of illustration and not limitation, as the principles, devices, systems, and

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 methods of the present invention may be used for sensing and/or determining the level of a variety of other physiological parameters, agents, characteristics, and/or compositions.

In light of the above-noted needs in continuous glucose monitoring, embodiments of the invention are directed to a more robust solution in the form of an orthogonally redundant sensor system. Orthogonal redundancy is defined as two devices employing two different technologies to reach the same goal, where the failure modes of the two devices are completely unique and do not intersect. This can be applied to continuous glucose sensing through the use of unique glucose detection schemes combined into a single body-worn device. The distinctive measurement technology, responses, and failure modes for each sensor provide true redundancy to ensure reliable and safe glucose measurements regardless of the environmental response or sensor anomalies.

In an embodiment of the invention, the above-mentioned orthogonal redundancy may be created by combining the technologies of optical sensing and electrochemical sensing to provide a unique solution to combat the complexities of the implanted environment. The two (i.e., optical and electrochemical) sensors are subject to different types of interferences, failure modes, and body responses, as described in Table 2 below. With this in mind, the reliability of each sensor can be calculated and weighted to provide the most robust and accurate glucose sensor measurement. Thus, as shown in Table 2, the unique and distinctive response to interferents and environmental perturbations by each of the sensors offers an enhanced ability to diagnose and filter environmental response.

Table 2

<td> Perturbation</td><td> Optical Sensor Response</td><td> Electrochemical Sensor Response</td>

<td> Endogen substances, i.e. ascorbate</td><td> No interference</td><td> Elevates glucose levels minimally</td>

<td> Exogenic substances, i.e. acetaminophen</td><td> Decreases glucose levels</td><td> Elevates glucose levels, reduced with membrane</td>

<td> Biofouling</td><td> No interference</td><td> Change in sensitivity</td>

<td> Temperature</td><td> Small change in baseline</td><td> Minimized by design</td>

<td> Oxygen</td><td> No interference</td><td> Minimized by design</td>

<td> With reference to Tab</td><td colspan="2"> e 2 above, it has further been found that the interference profile</td>

of the optical sensor is very different from the interference profile for the electrochemical

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 sensor. Thus, for all three of the primary electrochemical interfering substances-i.e., Acetaminophen, Uric Acid, and Ascorbic Acid--a single fluorophore optical sensor has either no interference or an interference signal that is in the opposite direction to that of the electrochemical sensor.

There are several sources of inaccuracies in glucose sensors. These inaccuracies may cause errors in sensor readings that can be corrected by a calibration, or they may be more serious errors from which the sensor cannot recover. The most common sources of error and the impact on the individual sensors are listed in Table 3 below.

Table 3: Sources of sensor inaccuracy for both optical and electrochemical sensors, as well as benefit of simple and orthogonal redundancy on sensor performance.

___“NB” = No Benefit, “PB” = Potential Benefit, and “CB” = Clear Benefit.________

<td> Sources of inaccuracy</td><td> Electrochemical sensor effect</td><td> Optical sensor effect</td><td> Benefit with simple electrochemical redundancy</td><td> Benefit with orthogonal redundancy</td><td> Benefit with orthogonal redundancy + predictive diagnostics</td><td> Orthogonally redundant sensor mitigation</td>

<td> Insufficient hydration</td><td> Low signal at Startup</td><td> High signal at startup</td><td> PB</td><td> CB</td><td> CB</td><td> Weighted average to estimate glucose</td>

<td> Connection issue</td><td> Loss of signal</td><td> Shift in reference signal</td><td> PB</td><td> CB</td><td> CB</td><td> Diagnose connection issue and advise patient</td>

<td> Partial pull-out</td><td> Decrease in signal</td><td> No effect to output</td><td> NB</td><td> CB</td><td> CB</td><td> Use optical sensor signal, advise to replace sensor</td>

<td> Low local glucose concentration*</td><td> Dip in signal</td><td> Dip in signal</td><td> NB</td><td> NB</td><td> PB</td><td> Predictive diagnostics on electrochemical sensor wam of local change</td>

<td> Low local oxygen concentration</td><td> Drilldown in signal</td><td> No effect</td><td> NB</td><td> CB</td><td> CB</td><td> Weight optical sensor signal until re-calibration</td>

<td> Interference of electroaclive species (te, acetaminophen)</td><td> Increase in signal</td><td> Slight decrease in signal</td><td> NB</td><td> CB</td><td> CB</td><td> Use magnitude of differing responses to diagnose acetaminophen</td>

<td> Interference of saccharides**</td><td> No effect</td><td> Increase in signal</td><td> CB</td><td> PB</td><td> PB</td><td> Request recalibration based on differing signals</td>

<td> Biofouling</td><td> Decrease in signal</td><td> No effect</td><td> NB</td><td> CB</td><td> CB</td><td> Weight optical sensor signal until re-calibration</td>

<td> Compromised membrane</td><td> Decrease in signal</td><td> Reference signal gradually decreases</td><td> PB</td><td> CB</td><td> CB</td><td> Sensors will likely not experience compromised membranes at same time</td>

* Failure mode of decreased glucose around sensor hypothesized to be due to attenuated perfusion to the sensor implant site ·♦ Interference potential with maltose, used in certain hospital procedures such as peritoneal dialysis, is under evaluation to determine the extent to which this affects optical sensor response in the concentrations available in interstitial fluid.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

It is known that acquiring signals from multiple electrochemical sensors can provide improved performance in the form of simple redundancy, accomplished through either multiple electrodes on the same probe, or by utilizing spatial separation and two separate probes. For example, Medtronic, Inc. sells hospital glucose sensors that include two probes, with two working electrodes on each probe, resulting in four independent glucose signals.

Systems utilizing multiple electrochemical sensors are also being developed by Medtronic, Inc. However, these systems still do not provide true redundancy through alternate sensing technologies with separate and distinct failure modes. As an example, studies have shown that, as the electrochemical sensor is pulled from the subcutaneous region into the dermal layers, the sensor signal goes to zero. In comparison, optical sensors perform well in both the dermis and the subcutaneous region, which allows the optical sensor to maintain functionality even as the sensor is partially explanted, providing the patient with a measurement until the patient is able to replace the sensor. Simple redundancy with electrochemical sensors would result in inaccurate data from both sensors in the event of partial explantation. See fable 3 above.

In short, in order to achieve the reliability required of continuous glucose monitoring systems, including closed loop, orthogonal redundancy is necessary. With orthogonally redundant sensing, the advantages of simple redundancy are maintained, with the additional benefit of having different susceptibilities and interferes between optical and electrochemical sensors. Thus, in embodiments of the instant invention, an orthogonally redundant sensor may include an optical sensor and an electrochemical sensor, wherein the latter may include up to, e.g., 5 independent sensing electrodes.

Figures 1A and IB show components of a continuous glucose monitoring system for orthogonally redundant sensing in accordance with an embodiment of the invention. With reference to Table 4 below, in developing sensor systems, the role of the entire system on

CA 02845804 2014-02-18

WO 2013/036493

PCT/US2012/053707 accuracy is considered, and a system-based approach to design is employed. Thus, as detailed in Table 4, each subsystem or component plays an integral role in contributing to the accuracy.

Table 4

<td> SUBSYSTEM Transmitter</td><td> DESCRIPTION Powers device end samples sensor response</td>

<td> Algorithm</td><td> Software contained either within the transmitter or monitor</td>

<td> Monitor</td><td> Receives data from transmitter and (.oniinuiiiLatev to patient. Link to the cloud. Hauses a BG meter</td>

<td> Sensor</td><td> Implanted unit and Ore base that adheres to patient skin.</td>

<td> Chemistry</td><td> Enzyme and membrane deposited on top of the implanted sensor circuit.</td>

<td> Accessories</td><td> Additional components, such as serte· and overtape and patch adhesive.</td>

ACCURACY DETERMINER initializes the sensor with a mociflet! pulse sequence to improve initial run in time for the sensor to reach stability

Calibrates the sensor and performs fault detection diagnostics, reducing occurrence of erroneous data through additional meter points or data exclusion.

Performs error check to eliminate influence of bad meter points on accuracy and communicates reference factor/ cal values to aid transmitter diagnostics.

Optimized electrode placement improves startup arid removes local effects.

□imination of solvents and chemical reucticrs in processes improve accuracy Thickness and layers optimized to improve Oay 1, dynamic range ana durability.

Sorter reduces trauma due to insertion. Overtape and patch adhesive prevent migration of the sensor that would reduce accuracy or result in early end of life.

As will be described in more detail below, one goal of embodiments of the present invention is to continue to simultaneously improve both performance and usability. Thus, within each of the sub-systems described in Table 4 above, electrochemical sensor performance advancements have focused on reduction of variation through targeted improvements. These targeted improvements are designed to improve day 1 performance, 10 durability, and hypo- and hyper glycemic performance and are detailed in Figure 2. Targeted improvements drive the electrochemical sensor to a predictable sensitivity across sensors, glucose ranges and over time. The sensor anomalies that remain as outliers can be reduced through predictive sensor diagnostics, which proactively detect faults or failures and recalibrate or shut down the sensor before it results in inaccurate glucose measurements.

It is understood that, for a given sensor or sensing system, the lower the Mean Absolute

Relative Difference/Deviation (MARD) value, the higher the accuracy of the sensor or sensing system. As noted in Figure 2, the system-based approach (to targeted sensor improvements) of the present invention reduces the MARD value for an electrochemical sensor from about 16%

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 to about 9%. and preferably less. For example, with respect to the transmitter 10, MARD is reduced by 0.5% by improving responsiveness and reducing lag time (reference numeral 11). Similarly, with regard to the design of the electrochemical sensor 200, MARD is reduced by an additional 0.5% by effecting a distributed-electrode design in order to reduce local effects (reference numeral 201).

With the above in mind, embodiments of the present invention are directed to an orthogonally redundant glucose sensor that includes an optical based sensor and a non-optical sensor. Thus, within the context of the present invention, in an orthogonally redundant glucose sensor, the above-mentioned electrochemical (i.e., non-optical) glucose sensor may be complemented with an optical based glucose sensor. In an embodiment of the invention shown in Figures 3A-3C, the optical sensor may be a sensor capsule 80 that is inserted under the skin 81 in the dermal layer, with a reader device 82 positioned above the skin. Light is transmitted between the reader device 82 and sensor 80 through the dermal layer in order to excite the sensing element under the skin, and the resultant fluorescence is measured in the reader device. Figure 3C shows the relative size of an exemplary optical sensor capsule 80.

In an alternative embodiment, shown in Figures 4 and 5, the optical sensor may be implemented by employing a transcutaneous optical fiber. Here, the fiber serves as a light guide with the sensing element attached to the distal tip of the fiber. The fiber extends through the skin where it is aligned with the reader device. Light is transmitted between the reader device and the sensing element through the optical fiber.

In a preferred embodiment, the sensing element includes a glucose binding competitive affinity assay surrounded by a glucose-permeable membrane, allowing the glucose within the assay to equilibrate with the glucose present in the surrounding tissue. The assay, in tum, includes a glucose analog (e.g., dextran) and a glucose receptor (e.g., Mannan Binding Lectin (“MBL”)) which is fluorophore-labeled to impart fluorescence. The equilibrium between MBL bound to glucose and dextran, respectively, determines the fluorescence intensity' in

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 response to illumination of the assay. A non-glucose sensing macromolecule labeled with another fluorophore serves as an internal reference (i.e., a reference fluorophore), wherein the latter emits its own fluorescence in response to illumination. The ratio of the assayfluorescence and reference-fluorescence intensities is converted into a glucose concentration.

An optical glucose sensor having an assay compartment may be formed, e.g., by including a glucose permeable membrane containing the assay at the distal end of an optical fiber. The optical fiber may then be inserted transdermally into the user’s body, thereby situating the assay compartment in the user’s tissue, while leaving at least a part of the optical fiber outside the body such that it can be accessed by (i.e., optically coupled to, or aligned with) an interrogating system. Alternatively, the optical sensor may be implantable, e.g., as part of an implantable glucose monitor including an interrogating optoelectronic system and a power source. The assay compartment may be formed between a glucose permeable membrane and an optical interface to the optoelectronic system. The glucose-permeable membrane may preferably be biodegradable.

As noted above and shown in Figure 4, an optical glucose sensor may be based on a competitive glucose binding affinity assay including a glucose receptor (e.g., MBL) and glucose analog/ligand (e.g., I lOkDa dextran) contained in an assay compartment. The binding between MBL and glucose-like molecules (e.g., dextran) is reversible. When no glucose is present, MBL and dextran will predominantly be bound together. When glucose is added to the assay, it will compete off a part of the dextran population, such that the assay enters a new equilibrium state. The equilibrium state at all times conesponds to the glucose concentration. In order to determine this equilibrium state, MBL is labeled with a donor fluorophore (e.g., Alexa Fluor 594, or AF594), and the dextran is labeled with an acceptor dye (e.g., hexamethoxy crystal violet-1 (HMCV1) - a proprietary crystal violet derivative manufactured by Medtronic, Inc.). The donor fluorophore and the acceptor dye together form a FOrster

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Resonance Energy Transfer (FRET) pair—i.e., the emission spectrum of the fluorophore and the absorption spectrum of the dye overlap.

The occurrence of FRET affects the lifetime of the excited state and the intensity of the emitted fluorescence and can only occur when the fluorophore and the corresponding dye are in close proximity (i.e., in the range of about 50Â). Thus, the FRET mechanism permits intenogation of the equilibrium state optically by illuminating the assay and measuring either the lifetime of the excited state (“lifetime interrogation’*), and/or the intensity of the emitted fluorescence from the donor fluorophore (intensity interrogation). In embodiments of the invention, the latter approach is preferred, as it exposes the assay to 25 times less light than with the lifetime interrogation.

The FRET mechanism offers several advantages. First, it works transdermally, within an appropriate wavelength range, so that interference from the skin is minimized. Second, FRET fluorescence lifetime measurements are generally insensitive to the relative position of the sensor and the reader unit as long as they are within optical reach of each other, and are also insensitive to changes in the environment, which helps make the system virtually calibration free. Lastly, FRET it considered very sensitive if the appropriate donor-acceptor ratio and suitable donor-acceptor geometry are obtained.

In selecting the FRET pair, the donor fluorophore and the acceptor dye are preferably water soluble, as they are to function in an aqueous environment. In addition, since the sensor is implanted or resident in the body, both FRET components should be non-toxic, as well as stable at 37°C for at least 2 weeks in the interstitial fluid (ISF). Moreover, fluorescence emission from the FRET pair should be in the red/far-red spectrum to minimize interference from substances in the skin and/or tissue auto-fluorescence.

Resistance to photo-bleaching, i.e., the photostability of both the dyes and the MBL and dextran, is also important. The photostability of the protein originates from its resistance towards Radical Oxygen Species (ROS) generated by the excited dyes, and is an important

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 feature in the stability of the assay. As will be discussed further hereinbelow, this is also a reason why MBL is relatively more resistant to e-beam radiation (wet or dry) than other proteins.

Finally, the donor fluorophore and the acceptor dye must work with a coupling chemistry suitable for protein (preferably amine) conjugation. As discussed above, in an embodiment of the invention, the MBL molecule is labeled with a donor fluorophore via the εamino group on lysine residues using N-hydroxy succinimide (NHS) derivatives of the fluorophore, since this chemistry generates a very stable amide bond between the protein and the fluorophore, and works well in aqueous buffers at pH values that do not compromise the protein.

From an optical point of view, a number of different fluorophores, such as, e.g., Alexa Fluor fluorophores, Texas Red, and Cy5 may be used as fluorophores. However, it has been found that the Alexa Fluor fluorophores work best as they exhibit and/or facilitate several practical advantages, e.g., coupling chemistry, water solubility, photo stability, and quantum yield. Alexa Fluor 594 (AF594), in particular, works well in the conjugation process with MBL; it is commercially available as an NHS derivative and, as such, is ready to be coupled to lysine residues on the MBL molecule.

The single MBL polypeptide has 19 lysine residues which are all potential conjugation sites. The polypeptide organizes in triplexes, each having 3 carbohydrate recognition domains (CRD), that again form higher complexes, usually with 9, 12, or 15 CRDs. For embodiments of the invention, it has been found that a degree of labeling (DOL) with AF594 of about 0.8-1 AF594/CRD gives optimal dose-response, with dextran labeled with HMCV1 as ligand. A DOL value that is too high would lead to self-quenching, while a DOL value that is too low would compromise the signal magnitude. It should be noted that, when using NHS as conjugation chemistry, AF594 will be more or less randomly coupled to the 19 lysine residues per polypeptide chain. This means that AF594 sitting on lysine residues in the collagen like

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 domain of MBL, distant to the CRD, may not participate in the FRET, unless the dextran molecule (size 110.000 Da), due to its linear conformation, is able to reach, with an HMCVI dye, into the Forster space of such an AF594.

As noted, the ligand in the sensor is preferably dextran supplied with amino groups in order to be able to use NHS coupling chemistry for labeling with the acceptor dye. For the latter acceptor dye, hexamethoxy crystalviolet-1 (HMCVI) is preferred over commerciallyavailable acceptor dyes because it is “non-fluorescent”--i.e., it has an absorption spectrum overlapping AF594’s emission spectrum, without overlapping AF594’s absorption spectrum too much-and works with NHS, i.e., it has a carboxylic group. The above-mentioned nonfluorescence is important, as it helps reduce not only the amount of optical interference with the donor emission, but also the amount of optics instrumentation that is required. In addition, HMCVI is versatile, such that it can also be used with other fluorophores, e.g., AF647, which is discussed more fully below in connection with use of a red laser diode as a light source.

For embodiments of the invention, it has been found that approximately 5 HMCV1 molecules per dextran molecule produce optimal dose-response, with the fluorophore-labeled glucose receptor MBL-AF594. Here, a DOL value that is too low would result in inefficient quenching, which would compromise the magnitude of dose-response, while a DOL value that it too high would compromise excitation of AF594, since HMCVI also absorbs at AF594’s excitation wavelengths.

With reference to Figure 6A, it is noted that there are actually three separate equilibria involved in the operation of the optical sensor described above. The first equilibrium is the one between glucose in the interstitial fluid and glucose inside the sensor compartment, which is regulated by osmotic pressure, i.e., the difference in glucose concentration in the 1SF and inside the sensor compartment. The second equilibrium is the one between the glucose interacting with MBL and free glucose, which is mainly regulated by the affinity between glucose and MBL. The third equilibrium is the one between MBL and dextran, which is

CA 02845804 2014-02-18

WO 2013/036493

PCT/US2012/053707 regulated by the affinity between dextran and MBL and the concentration of glucose inside the sensor compartment.

All three equilibria are dynamic and reversible. What this means is that the same glucose molecule may at one moment interact with a MBL molecule, and in the next moment be non-interacting with MBL, and in a third moment cross the sensor membrane, leaving the sensor compartment and entering into the ISF. The interaction between the assay chemistry components (MBL-AF594 and dextran-HMCVI) reflects at any time the concentration of glucose in the sensor compartment. Fouling of the sensor-which may potentially compromise the permeability of the sensor-may extend the response time to changes in glucose concentration in the ISF, but does not interfere with the glucose measurement in the sensor. That is, the assay chemistry always measures the correct glucose concentration inside the sensor compartment. In short, fouling of the sensor has no influence on the equilibria inside the sensor. Moreover, all equilibria that involve glucose are fully reversible and, as such, glucose is not consumed in the measuring process.

In contrast with optical glucose sensors, electrochemical glucose sensors are glucose consuming enzyme kinetics based systems. Since the latter reactions consume glucose, sensor response is dependent on glucose diffusion across the outer membrane of the sensor. This can be described by the following mass transfer equation:

<img file="CA2845804C_D0001.tif" />

Eq.(l) where j is the glucose flux, D is the diffusion constant, C = [Glu], and X is distance. Biofouling changes the thickness of the sensor membrane (dX), thus reducing the glucose flux and measured sensor response. Hence, a sensor re-calibration would be required.

However, since optical glucose sensor technology is not glucose consuming, i.e., it is based on reversible glucose binding to a glucose receptor protein, as detailed above, sensor response depends on the concentration of glucose inside the sensor (assay) compartment. The

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 glucose levels inside the compartment will always be in equilibrium with glucose levels outside the membrane regardless of the thickness of the outer membrane and/or bio-film, because glucose is not being consumed. This equilibrium system can be described by the following equation:

K = ([MBL - Dex][Glu])/([MBL - G/w][to]) Eq. (2)

Since MBL and Dextran concentration is fixed inside the sensor, K. is only dependent on glucose concentration. Since bio-fouling occurs outside the membrane, the equilibrium of the reaction is not affected. Empirical data confirm the above-noted outcome.

Returning to Figure 5, the optical system used to interrogate the above-described sensing element (assay) is essentially a modified epi-fluorescence set-up with one light source to excite (i.e., illuminate) the assay and two detectors to detect the fluorescence emitted from the assay and the internal reference, respectively. As noted, the intensity of the emitted fluorescence correlates to the glucose concentration. Here, the measured intensity of the emitted fluorescence is affected by the intensity of the light source and the coupling between the assay and the optical system. Therefore, the intensity measurement requires an internal reference fluorophore to be incorporated into the assay.

The reference fluorophore must differ from the assay fluorophore in a way that the emitted fluorescence from the assay and that from the reference may be separated from one another, e.g., by having different absorption spectra or emission spectra. The reference fluorophore may be, e.g., Alexa Fluor 700 (AF700) labeled onto Human Serum Albumin (HAS) or another macro molecule, which largely does not bind to the glucose receptor. Alexa Fluor 700 may be excited simultaneously with the Alexa Fluor 594 as their absorption spectra spectrally overlap. The emission spectrum from Alexa Fluor 700 is slightly red shifted with respect to Alexa Fluor 594, which makes it possible to detect their respective fluorescence emissions in separate wavelength regions.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

The excitation, as well as the detection, of the emitted fluorescence for the assay and the reference follow the same optical path from the optical system to the assay. As such, the detected signal from the reference serves as a measure for the optical coupling between the optical interrogating system and the assay. Any effect originating from changes in the optical coupling such as alignment may be cancelled out.

With reference to Figure 5, in an embodiment of the invention, a driver circuit 1310 modulates a LED 1320 at a low frequency-solely with the purpose of eliminating the 1/f noise and canceling out ambient light—with a wavelength range capable of simultaneously exciting the assay and reference fluorophores. The LED output is filtered using a multilayer dielectrical filter 1330 to select a distinct wavelength region. The filtered LED output is reflected by a first dichroic beam splitter 1340 and focused onto the sensor 1300, which includes the assay and the reference, by a lens 1350.

The assay and the reference emit fluorescence. The emitted fluorescence 1301 and the reflected excitation light 1323 are picked up and collimated by the lens 1350. The first dichroic beam splitter 1340 transmits the fluorescence 1301. However, it reflects the majority of the back reflected excitation light 1323. A second beam splitter 1344 reflects the reference fluorescence at a 90° angle 1307, but it transmits the assay fluorescence 1309. A first emission filter 1360 with a distinct wavelength region red shifted with respect to, and not overlapping, the pass band of the excitation filter and matching the desired part of the assay fluorescence spectrum then blocks the remaining part of the excitation light and transmits the assay fluorescence.

Similarly, a second emission filter 1364 with a distinct wavelength region red shifted with respect to, and not overlapping, the pass band of the excitation filter and matching the desired part of the assay fluorescence blocks the remaining part of the excitation light and transmits the reference fluorescence 1307. Thus, in effect, only the fluorescence from the assay and the fluorescence from the reference are focused onto their respective photo detectors

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

1380, 1384 using respective lenses 1370, 1374. The ratio between the detected assay fluorescence and the detected reference fluorescence correlates with the glucose concentration in the assay.

T he above-described optical sensor technology offers several advantages over other available technologies. For example, as noted previously, due to the non-consuming and stable nature of the assay, the measurement technique is insensitive to bio-fouling. As such, it offers the possibility of one single point calibration throughout the entire lifetime of the sensor. Furthermore, the assay contains a reference dye, which remains stable with changing glucose concentrations, but is affected by many non-glucose induced changes. Therefore, it serves as a sensor diagnostic tool for the optical sensor, indicating when the integrity of the membrane has been compromised or the optical connection is misaligned. See, e.g.. Figure 6B. In addition, as will be described further below, the assay may comprise a protective formulation, which is suitable for radiation sterilization, a common sterilization technique for glucose sensors.

Moreover, the glucose receptor, MBL, is a human derived protein. As such, there is no immune response. Moreover, MBL may be derived from plasma or produced rccombinantly. In addition, compared to other proteins that may be used for equilibrium-based glucose sensing, MBL has proven biocompatibility and is used clinically for pharmaceutical purposes. Table 5 shows the known differences between MBL and other glucose binders employed for equilibrium-based glucose sensors.

CA 02845804 2014-02-18

WO 2013/036493

PCT/US2012/053707

Table 5

<td></td><td> Mannan Binding Lectin (MBL)</td><td> Glucose Galactose Binding Protein (GGBP)</td><td> Conamavalin A (Con A)</td><td> Antibodies (Ab's)</td><td> Boronic Acids</td>

<td> Description of the glucose receptor</td><td> Natural occurring human Lectin. All humans carry as a part of their innate immune system' ’</td><td> PerlpUsmic protein naturally occurmg m e.g. E.Coti. Takes part in chemotaxis (cel movement) and metabolism</td><td> Chemical defense compound in Jack Bean. Helps to protect the bean from being eaten</td><td> Part of any higher organism's immune response</td><td> Synthetic receptor with very simple structure, only two primary binding sites and little secondary staMamo or selectivity creating binding</td>

<td> Specificity and affinity</td><td> Natural selected for mannose and glucose binding in the clinical range (recognizes foreign glycosylation on intruder cells e.g. E.Cofi)</td><td> Selective towards galactose and glucose Wild type protein genetically modified to fit clinical range</td><td> Selective towards mamose and glucose. Native and modified types fits the clinical range well</td><td> Ab's can be made to fit the clinical range by screening libraries. Ab variants created by small changes of the hyper variable regens m the Ab's¹*</td><td> Selectivity between stereoisomeric sugars (mannose and glucose) is difficult to obtain.</td>

<td> Stability</td><td> Natural selected tor stability at 37C and human ton (Ca²’«1.25mM) and metabolite (mM) concentrations</td><td> Naturally selected for body temperature^{1 1} (E.cdi is colon bacteria). Wild type optimized for bw Ca^J‘ (0.1 μΜ) and low metabolite (μΜ) cuncentration^,)</td><td> Poor at 37C¹’</td><td> Depends on the Ab, No generally information</td><td></td>

<td> Regulatory</td><td> Phase l^w and Phase 2^Wclinical studies have been made using h&gt;gh MBL concentration and found no adverse effects. Should ease regulatory route</td><td> No literature found. Genetically modified non-human protein has no known regulatory route (but msuhn is recombinant produced)</td><td> Known to cause agglutination or human cells.”</td><td> Ab variants have been approved as drugs, but a specific Ab is a new chemical entity and needs it own regulatory route”</td><td> New chemical entity needs testing</td>

<td> Sterility</td><td> wet and dry e-beam*¹</td><td> Aseptic assembly³* GGBP cannot be exposed to e-beam Or yradtaimn (iddical formation)</td><td> No information found</td><td> No info» mation found</td><td> Aromatic boronic acid reacts with Reactive oxygen Species (ROS) during sterilization and is eliminated and hence loses binding capability¹*</td>

<td> Sourcing / Production</td><td> Serum derived (hMBl)^slor recombinant (ihMBl). Recombinant comes from mammalian ceil lines to get the right glycosylation·¹</td><td> Recombinant from E.cok. A simple bacteria as E.coli cannot prowde glycosylation</td><td> Extracted from Jack Bean</td><td> Expressed from E.QMi or mammalian cel) (mes</td><td> Chemically synthesized</td>

<td> Development time</td><td> Since 2003</td><td> Since before 2002°</td><td> First publication 1962”</td><td> 2002 to 2005 with very limited success”</td><td> Since before 1998</td>

<td> References</td><td> 1) Mol Immunol 40(2003)423 2) Sea nd J Immunol 2004 51(1)97 3) Eur J Cancer 2009 45(4)505 4) MOT Results 5) Vox Sang 2007 92(4) 338-50 6) Biochem Soc Trans. 2003 31(4):763-7</td><td> 1) Biochem J. 2004 381(1) 97- 103 2) JBniChem 262(1987)12570 3) Patent W02007/022485 4) B»osens Boetectron. 19(2004)653 (first menus 2002)</td><td> 1) PreciSense results 2) Nature 262(1979)738 3) Diabetes Care. 5(1982)3245</td><td> 1) Front Biosci. 1X2008)1117 2) Nat Bfotechnol. 2X2005)1105 3) PreciSense attempts both immunizing mice and screening libraries</td><td> l)USPatUS2011/00ei727</td>

Returning to the continuous glucose monitoring system for orthogonally redundant sensing, the several elements/components shown in Figures 1A and IB will now be described 5 in more detail.

The electrochemical sensor 200 is a state-of-the-art electrochemical sensor, such as, e.g., Enlite3 (third generation Enlite sensor, Medtronic, Inc.). As shown in Figure 7, the Enlite3 implanted sensor features a distributed sensing electrode design, wherein the sensing electrodes 210 are distributed along the length of the sensor to reduce local tissue effects on 10 sensor performance, as well as optimized solvent-free chemistry to improve consistency. In

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 embodiments of the invention, the electrochemical sensor may consist of a flexible polyimide material with no plastic tubing.

As described previously, and shown in Figures 8A and 8B, the orthogonally redundant sensor includes a fiber optical sensor 100. The fiber optical sensor 100 has a fiber 110 with a glucose-permeable membrane 120 attached at/proximate the fiber’s distal end 115. In embodiments of the invention, the optical fiber 110 is made of plastic having tensile and fatigue properties that ensure robustness. The glucose permeable-membrane 120 may, e.g., be heat sealed on the distal end 115 of the fiber. In embodiments of the invention, the membrane 120 may preferably be made of a biocompatible, biodegradable polymer such as, e.g., PolyActive™ (Integra Orthobiologics, Irvine, CA).

The glucose permeable-membrane 120 houses the assay chemistry 125. The size of the optical fiber 110 is optimized so as to improve hydration and response time, as well as to reduce the size of the implant and needle that is used to introduce the fiber into the patient’s body. As is also shown in Figures 8A and 8B, excitation light 130 travels from the proximal end 117 of the fiber to the assay chemistry 125, and the fluorescence response 140 travels back up the fiber to an optical interrogating system that is located, e.g., in the transmitter 10 shown, e.g., in Figures IA and IB.

The transmitter 10 includes instrumentation for the optical sensor 100 and the electrochemical sensor 200. For the optical sensor, such instrumentation may include, e.g., a light source, detector(s), optical drive electronics, and other elements/components of an optical interrogation system (discrete or integrated). For the electrochemical sensor, the instrumentation may include, e.g., a potentiostat and other related components (also discrete or integrated). As shown in Figures 9A and 9B, the transmitter 10 also includes a dual connector 20 that allows the two sensor elements 100, 200 to separately connect to the required instrumentation. Within the dual connection, the electrochemical connection may allow for up to four isolated contacts, and may be watertight. Similarly, the optical connection may be

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 watertight and/or provide for consistent index matching between optical surfaces. Here, while direct contact may not be needed, the light path must be clear.

In addition, the transmitter houses diagnostics, one or more microprocessors and/or digital signal processors (DSPs), memory’, a RF communication chip (using, e.g., 2.4 GHz TelD protocol), and a battery to support the measurement functionality of the sensors, the conversion of signals received from the sensors to glucose values, and wireless communication, including transmission of the glucose values (or an averaged, weighted, or otherwise modified version thereof) to, e.g., a monitor 300, an infusion pump 400, a display device, etc.

The transmitter 10 also houses the algorithm that utilizes predictive diagnostics and signal comparison to assess signal reliability. The algorithm features intelligent startup and calibration schemes so that the sensor performance dictates when calibrations are needed. Additionally, the algorithm operationalizes the conversion of the individual signals into a calculated glucose number, which is communicated to one or more of the devices noted above.

The transmitter 10 is a durable device and, as such, the associated battery may be rechargeable. In these embodiments, the transmitter may require intermittent recharging of the contained battery. Therefore, preferred embodiments of the invention include a charger for use in conjunction with the transmitter (battery). Additionally, the charger may test the transmitter for proper functionality' when required. It is noted that, in embodiments of the invention, some or all of the elements/components that are housed in the transmitter 10 may be integrated in order to miniaturize the device. In this regard, a printed circuit board assembly (PCBA) may be used.

An insertion device 500 is used to implant the sensors 100, 200 in such a way as to minimize trauma and maximize patient comfort and consistency- of sensor delivery. See Figure

10. The insertion device relies on a disposable, automatically retracting needle 510 that is designed with the sensor base to deliver the sensors 100, 200 through the user’s skin.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Specifically, the optical sensor 100 and the electrochemical sensor 200 arc co-located inside the needle and, as such, are inserted simultaneously.

The electrochemical sensor 200 generally comprises a thin and wide flex substrate. As such, it may be located between the opening of the needle 510 and the optical fiber sensor 100 to aid in retention. The diameter of the fiber sensor may be as large as about 500pm, but is preferably less than 200pm. It is noted that, in Figure 10, the needle 510 is shown at 0° (i.e., horizontally). However, in practice, the needle 510 is inserted at 90°.

As is clear from Figures 9A, 9B, and 10, in embodiments of the invention, the substrates for the electrochemical sensor and the optical sensor may be fabricated separately and assembled individually into a single base of a single sensor housing (e.g., the transmitter 10). The two sensors are then inserted within a single insertion device 500. However, although the insertion device deploys both sensor substrates together, the substrates are not connected in the implant area.

The electrochemical sensor (probe) and the optical sensor (probe) may, nevertheless, be co-located in vivo. In this regard, it has been discovered that the performance of one of the sensors is not affected by the presence of the other sensor within close proximity. For example, the presence of an optical sensor probe does not shadow or prevent glucose from reaching the electrochemical sensor (probe). Similarly, peroxide, which is produced as a byproduct of the electrochemical sensor reaction with glucose, does not affect performance of the optical sensor. Even at high concentrations of peroxide, such as 12ppm (i.e., equivalent to a 400 mg/dL glucose response for an electrochemical sensor), peroxide has been found to have no effect on the optical sensor response.

Figure 9C shows an alternative embodiment, where the substrates for the electrochemical sensor and the optical sensor are integrated so as to form an integrated flex circuit.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

The handheld monitor 300, which may also be referred to as “the On Body Controller” or “the On Body Communicator” (OBC), may include an integrated blood glucose meter 320 utilized for calibration. Algorithms within the handheld monitor 300 provide an error check to ensure that inaccurate blood glucose readings arc not communicated. Inclusion of this error check has the potential to decrease MARD-and, therefore, increase accuracy-significantly as an incorrect meter point used for calibration can falsely raise or lower calculated glucose levels. See, e.g. Figure 11.

Accuracy

In the continuous glucose monitoring (CGM) system described above, orthogonal redundancy using two unique sensing technologies provides for increased accuracy and reliability while enabling environmental etfects to be accounted for. Specifically, with respect to accuracy, embodiments of the present invention enable a MARD of about 13%. In this regard, it is understood that existing blood glucose meters (i.e., finger-stick) in-home use models are expected to have generally high accuracy; that is, a MARD approximating 9%, with 95% of all points expected to be accurate in terms of ISO 15197:2003. Under the latter standard, a meter is deemed accurate if it meets the following criteria for at least 95% of samples tested: (I) For blood glucose levels below 75mg/dl, the monitor reading must be within 15mg/dl of the reference; and (2) for readings of 75mg/dl or higher, the monitor reading must be within 20% of the reference reading.

For closed-loop ready sensing systems, meter equivalency is not a necessity. Here, the literature has suggested a much looser system accuracy requirement with a MARD of 15% (see, e.g., Hovorka R., “Continuous glucose monitoring and closed-loop systems,” Diabetic Medicine 2005(23)). In fact, current-generation CGM systems have published accuracies meeting the 15% requirement, but are accompanied by a large reduction in percentage of samples considered accurate according to the ISO 15197 standard noted above. This deviation in system accuracy may be attributed to multiple factors (e.g., calibrating meter inaccuracy,

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 sensor delay, etc.); however, it is noted that the requirement treats blood samples as independent, discrete events. Contextual (trending, historical) data provided by CGM systems should allow for a relaxation of what is deemed an “accurate” reading.

Reliability

The orthogonal redundancy of the inventive system also allows for a combined reliability that far exceeds the individual reliability of either sensing component. Specifically, as will be discussed further below, two orthogonal sensors with an ISO accuracy of 75% would theoretically be accurate 93.75% of the time when combined. The redundancy increases both accuracy and percent of time data is displayed.

A reliable system requires (I) data to be displayed as often as possible while (2) only displaying data when it is accurate. It is noted that, with improvements to sensor technology and failure detection algorithms, the accuracy of sensor systems will improve significantly. However, failure detection algorithms that are too sensitive might reduce the amount of displayed data to an extent that is unacceptable to the user. In this respect, two components make up the reliability of the sensing platform described herein: (1) data display (% of time); and (2) accuracy (% of time).

The system of the present invention meets the following reliability requirements for 94% of sensors: (1) It displays sensor data 90% of sensor wear “calibrated” time; and (2) it meets ISO 15197:2003 requirements on 93.75% of displayed sampled points. It is noted that some existing sensor technologies may currently meet the first criterion above, but, with regard to the second requirement, significant improvements would be needed in order to achieve nearmeter equivalency in terms of ISO 15197:2003.

Existing sensor technology has published accuracy roughly on the order of 70%, meaning that 70% of all evaluated CGM points are deemed accurate according to the ISO 15197:2003 standard. Therefore, assuming two sensing components of roughly equivalent accuracy with random distributions of sensor error occurrence (i.e., assuming that both sensing

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 components will not always be reading inaccurate at the same time), significant gains in accuracy may be realized provided that the system is able to quickly identify possible faults in one or the other sensing component.

Probabilistically, this may be shown as follows:

Let:

be the set of all evaluation points for sensing component I (e.g., an optical sensor).

be the set of all evaluation points for sensing component 2 (e.g., a non- optical sensor).

SI and S2 be independent, normally distributed variables (due to sensor orthogonality).

Then, the probability that for any sample in time either SI or S2 will be accurate is derived from the additive rule for non-mutually exclusive events:

P (a OR b) = P(a) + P(b) - P(a) x P(b) Eq. (3)

Where a, b represent whether a point in SI, S2 is accurate (as defined by ISO 15197:2003); and

P(a), P(b) represent the probability that any such point is considered to be accurate.

Using two sensors with P(a) = P(b) = 0.7, P(a OR b) = 0.7 + 0.7 - (0.7 x 0.7) = 0.91 (i.e., accurate on 91% of points). Thus, any increase in accuracy performance of either sensing component over this baseline increases the accuracy of the overall system as well. Table 6 shows individual accuracy effect on an orthogonally redundant system, assuming true independence between the two sensing components.

Table 6: Individual accuracy effects on an orthogonally redundant system

<td> P(U)</td><td> P(b)</td><td> Pis OR b)</td>

<td> 70%</td><td> 70%</td><td> 91.00%</td>

<td> 70%</td><td> 75%</td><td> 92.50%</td>

<td> 75%</td><td> 75%</td><td> 93.75%</td>

<td> 75%</td><td> 80%</td><td> 95.00%</td>

<td> 80%</td><td> 80%</td><td> 96.00%</td>

<td> 90%</td><td> 90%</td><td> 99.00%</td>

As noted, the expected combined accuracy is based on anticipated improvements in accuracy to one or both sensing components in order to achieve 93.75% accuracy without

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 sacrificing usable sensor lifetime, and assuming complete independence. In a preferred embodiment of the present invention, where one of the two sensor components is an optical glucose sensor, and the non-optical sensor is an electrochemical glucose sensor, some of the factors that may influence complete independence of the optical and electrochemical sensing technologies include, e.g., the following: (1) sensor co-location within a single implant does not account for physiological effects (i.e., decreased interstitial fluid glucose concentration as a result of increased pressure on the insertion site); and (2) simultaneous calibration of both sensing components relies on an expectation of accuracy from the reference point (e.g., meter finger-sticks) such that, if not correctly identified by the system, a sizeable error from the reference point may propagate into sensor glucose calculation, resulting in distortions of sensor accuracy for both sensing components.

Hypoglycemia Performance

Combining the optical sensor and the electrochemical sensor yields a sensing system with high precision both in the hypoglycemic and the hyperglycemic range due to the individual dose responses. Figure 12 shows dose response functions (i.e., the correlation between sensor output and glucose dose) for an optical equilibrium glucose sensor and an electrochemical glucose sensor. The optical sensor features a steeper slope 133 in the hypoglycemic region, leading to higher precision, while the electrochemical sensor has a linear slope 233, resulting in higher precision in the hyperglycemic region.

The established accuracy standards for glucose monitoring devices allow for higher percentage error in the hypoglycemic regions because the clinical treatment decision remains the same regardless of hypoglycemic severity. In closed-loop systems, sensor performance in regions of glycemic excursion (either hypo- or hyper-glycemic ranges) becomes increasingly important, as such systems rely not only on excursion accuracy, but also on contextual trending data as crucial feedback input for control algorithms.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

The orthogonally redundant sensor according to embodiments of the invention offers benefits in terms of hypo- and hyper-glycemic performance. The two glucose sensors have different dose response curves that may improve hypoglycemia and hyperglycemia performance. Equilibrium sensors’ dose response function is not a linear function, but a curved shaped function with the steepest slope when approaching a glucose concentration of 0 mg/dL. The steeper the slope in dose response, the higher the precision of the sensor is. Therefore, the affinity-based glucose sensors generally have better hypo sensitivity than hyper sensitivity as opposed to electrochemical sensors, where the dose response function is a linear function resulting in equivalent hypo and hyper sensitivity. Combining the optical sensor and the electrochemical sensor, therefore, yields a sensing system with precision both in the hypo range and in the hyper range.

As noted previously, Hovorka has suggested that, for closed-loop applications, a MARD between 10-15% would be desirable with a preference toward underestimation rather than overestimation. Moreover, the Clinical and Laboratory Standard institute (POCT05-P, “Performance Metrics for Continuous Glucose Monitoring; Proposed Guideline,” CLSI) has proposed definitions for home-use hypoglycemic sensitivity, specificity, and false alert rates (for continuous interstitial glucose monitoring) as follows: (1) Sensitivity: for any meter reading below 70mg/dl, a sensitive CGM system shall also read 70mg/dl or below within +/30 minutes of the reference sample; (2) specificity: for any euglycemic meter reading (not hypo- or hyperglycemic), a CGM reading also within this range is considered a true negative; and (3) false alert: for any meter reading above 85mg/dl, any CGM reading which at that time reads at 70mg/dl or below will be considered a false alert. The sensitivity/specificity metric allows for consideration of the contextual data provided by the CGM system most relevant to closed-loop control.

In embodiments of the invention, the orthogonally redundant sensing system meets a hypoglycemic MARD of 13% with sensitivity and specificity of at least 95% and false alert

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 occurrence rate below 10%. The independent accuracy of each sensor in the orthogonally redundant system meets this requirement in the majority of situations, especially given that orthogonal redundancy allows for elimination of signals that are on the edge, further improving sensitivity/specificity and false alerts.

Reduced Wann Up

The orthogonally redundant sensing system in accordance with embodiments of the invention also provides reductions in warm-up time through optimization of individual sensor wann-up time. The overall system start-up time, which is defined as the time until sensor signal is stable enough for performing the first calibration, is reduced by utilizing predictable run-in behavior and start-up diagnostics as inputs to the algorithm to create an adaptive warm up. Reducing sensor start-up time is important for accuracy and reliability of the system, as well as the user’s convenience, as it allows the patient to complete finger-stick calibration soon after inserting the sensor.

With respect to minimization of the individual sensor start-up times, the chemistry layers for the electrochemical sensor may be optimized, and new initialization schemes may be employed in the orthogonally redundant sensor. For the optical sensor, the hydration of the (assay) chemistry may be sped up, and the design may be optimized for a maximized surface area to volume ratio. Hygroscopic agent(s) or chemical(s)—such as, e.g., sugar, honey, and certain salts, which attract and retain water molecules from the atmosphere-may also be added to the assay.

One of the major obstacles to obtaining a fast startup time is to remove air from inside the optical fiber sensor. In this regard, it has been discovered that adding a combination of sugars, bicarbonate, and an enzyme to (the assay of) the sensor gets about 90% of the air out of the sensor within about 30 minutes. Further reduction of start-up time may be possible by optimizing the proportional make-up of the above-identified combination.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Similarly, it has been discovered that smaller-diameter optical fiber sensors provide a reduction in run-in time. For example, replacement of a 500pm-diameter fiber with a 250pmdiameter fiber has been shown to reduce run-in times from about 3-4 hours to about 2 hours.

In addition to optimizing the individual sensors, the combined operation of both sensors in one system may also facilitate faster start-up. Predictable run-in characteristics are incorporated in the algorithm, which helps lower the perceived start-up time, thereby also reducing the number of finger-stick calibrations during this time. Also, as will be discussed further below, intelligent algorithms could compensate for the startup characteristics of each sensor element and any sensor anomalies through a reliability index approach.

In fact, the initial profile of sensors is an important input to early-life sensor diagnostic algorithms. The post-initialized behavior is evaluated by the system to (1) determine the times at which sensors will be ready for initial calibration (adaptive warm up) and (2) identify sensors that are not adequately sensitive to glucose fluctuations (non-critical fault detection).

Advanced Algorithms

In embodiments of the invention, an advanced algorithm combines reliability information from each sensor and exploits features of the orthogonally redundant sensors to reduce lag, improve start-up time, and improve accuracy. By comparing signals, faults can be confirmed and self-calibrations can be performed, thereby reducing the number of glucose meter calibrations required.

As shown in Figure 13A, the algorithm takes the signals and fault detection of each algorithm into account, and then determines the reliability of each signal and weighs them appropriately. The algorithm also takes advantage of the specific benefits of each sensor. For example, the optical sensor generally has a more stable signal compared to the electrochemical sensor, which is known to have a gradual change in sensitivity over time, requiring recalibrations. With Electrochemical Impedance Spectroscopy (EIS) measurements, or by comparing large recent periods of the electrochemical sensor’s signal, instances can be

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 identified where the sensitivity of the electrochemical sensor has changed. The optical sensor will then allow an immediate confirmation of possible sensitivity changes and, if the signal is deemed reliable enough, the electrochemical sensor can be re-calibrated based on the optical sensor. This self-calibration feature reduces the required number of external calibrations, which are typically necessary to maintain high accuracy. In the optimal scenario, calibrations will be needed to maintain confidence in the signal.

While the optical sensor is generally more stable, the electrochemical sensor has other advantages. For example, during the first few hours of start-up, the electrochemical sensor is expected to reach a semi-stable point more quickly, but have a slight increase in sensitivity over the next few hours. As previously described, predictable run-in characteristics can be incorporated in the algorithm.

Figure 13B shows an embodiment in which diagnostics are used to determine the reliability of individual signals, which signals are then weighted accordingly. The individual weighted signals are then added and multiplied by a calibration factor to determine a calculated glucose value. The term “calibration factor”, or “cal factor”, as used herein refers to the ratio of blood glucose to sensor signal.

In another aspect, the algorithm includes a model for transformation of the sensor signal to match blood glucose concentration. See Figure 14. This is done by a twocompartment model, which presumes the sensor is in a different compartment than the calibration measurements. The model accounts for the diffusion of glucose between blood, where calibration measurements take place, and the interstitial fluid space, where the sensor is located. The model also accounts for glucose uptake by cells.

It is expected that the optical sensor may have a slightly longer response time than the electrochemical sensor. The advanced algorithm can compensate for this lag by examining each signal’s rate of change, and comparing the two signals. Depending on various factors, the electrochemical sensor may detect changes more rapidly. The algorithm needs to detect the

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 change, and if it is unable to compensate for the change, the system may weigh the electrochemical sensor more. Thus, while certain current sensors may perform better when calibrations are taken during more stable periods, incorporation of the two compartment model enables the use of calibrations taken at all times.

As noted previously and shown in Figure 13A, a sensor in accordance with embodiments of the present invention incorporates the benefits of redundancy and sensor weighting using a reliability index. In an exemplary embodiment of the system, multiple electrochemical sensors are evaluated individually, and a reliability index is created for each. In Figure 15, three sensors are sending data. Individually, each of these sensors would result in an accuracy of about 8%. However, when combined, the accuracy improves to about 4.4%. Thus, sensor accuracy is improved through assessing each individual sensor current with its reliability index (Figure 15A), and creating a weighted average (Figure 15B). It is noted that the inventive sensor, sensing system, and associated algorithms herein may be adapted for use at home and/or in a hospital setting.

Calibration

As has been noted, the orthogonally redundant system includes several features which result in a reduction in calibration frequency using an “on-demand” protocol to limit calibrations to 2-4/week (down from, e.g., 2 calibrations per day). These features include: (I) Sensor accuracy/durability improvements of electrochemical glucose sensors; (2) physiological model-based calibration algorithm; (3) redundant and orthogonal sensing technology which allows for internal self-calibration after individual components have reached stable-state; and (4) “Smart” diagnostics which allow for transition from timing-based to need-based calibration requests.

Historically, CGM systems have relied on “minimum scheduled sample time” for sensor calibration as a way to adjust for inaccuracies characteristic to the sensing component. Thus, existing calibration algorithms rely on a minimum of 1 calibration point for every 12

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 hours of sensor operation (ES9I99, ES9573, ES9966). Based on this standard, the DexCom B SEVEN*.· PLUS product, e.g., requires 2 at startup and every 12 hours afterward, and the FreeStyle Navigator*' requires calibration at 10, 12,24, and 72 hours post insertion.

As sensing technology has improved, sampling requirements have decreased, but at the expense of system accuracy. In contrast, the inventive orthogonally redundant sensing system allows for a significant reduction in calibration frequency compared to existing sensor technologies, while maintaining expectations of sensor accuracy throughout its lifetime.

The implementation of a diagnostic algorithm with the ability to verify sensor performance allows for a shift from “in-time” to “on-demand” calibration protocols. In this regard, Figure 16(a) shows a simulated calibration scheme based on current generation singlesensor technology, and Figure 16(b) shows an alternative made possible by measurement redundancy of the type disclosed herein. Pursuant to the latter calibration scheme, initial calibration(s) 331 are still necessary; however, the twice-daily (time-scheduled) calibration requests are no longer required as part of the calibration algorithm. Instead, a combination of infrequent scheduled requests 333 (i.e., once every 72 hours) and on-demand requests 335 ensures that sensor calibration will only be required when the system identifies a need to confirm sensor health. As system performance using this scheme relies on accurate and frequent diagnostic information, failure detection and other advanced algorithms will be critical to reducing the number of calibrations requested on a consistent basis.

It is noted that current prototype electrochemical sensors in development have internal targets of 13% MARD with signal drift less than 10% / day. Likewise, a calibration algorithm based on a two-compartmental fluid-flow model of glucose transfer within the body will reduce the blood-to-subcutaneous concentration gradient effect (delay) as well as eliminate artifacts from the signal that are deemed to be physiologically unlikely.

The above-mentioned sensor drift and failure detection will now be discussed.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Sensor Drift

In one aspect, the orthogonally redundant system of the instant invention increases confidence in drift detection by providing an internal reference from which the system is able to verify suspected drifts and confirm sensor deviations without the need for action from the 5 user.

Sensor drift is a characteristic of all sensing systems, and occurs over time or in response to other environmental conditions such as temperature, bio-fouling, etc. Such improvements in sensor design as, e.g., thermal stabilizers, membrane changes, and electrode treatments may be shown to reduce signal drift to levels on the order of 5-10% per day. While 10 a relatively small drift represents an improvement over existing sensors, system requirements for calibration frequency and accuracy must allow the system to account for these deviations.

The inventive system and related algorithms herein identify cases of significant sensor drift (in both sensors), and either account for the detected drift or halt glucose display to the user until the potential fault is resolved, e.g., by calibration. In this way, drift detection is 15 realized through signal analysis and is one parameter that is fed into the system reliability index (see Figure I3A).

Independently, the electrochemical and optical glucose sensing systems are able to do some amount of self-diagnosis of sensor drift simply by evaluating periodic sensor behavior and how it changes over the course of sensor life. As discussed previously, the non-glucose 20 consuming nature of the optical sensor chemistry offers the benefit of being insensitive to biofouling. Since the glucose sensitivity is not dependent on diffusion rate across the membrane, sensor drift through bio-fouling is generally not a concern.

Failure Detection

The state of the art in failure detection has been steadily moving towards predictive diagnostics that are designed to proactively identify sensor issues before they affect the glucose reading. The orthogonally redundant system of the present invention implements a three-tiered

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 approach to failure detection, i.e., failure detection solely with the electrochemical electrode, solely with the optical sensor, and then with information from the combined signal.

With the electrochemical sensor, the most sophisticated failure detection uses electrochemical impedance spectroscopy (EIS). EIS offers a quick on-line method to diagnose the sensor and sensor membrane status. An important advantage to EIS is that it can be done during sensor operation, without turning the sensor off or changing the electrode state. EIS is performed by passing a small AC voltage signal at a fixed frequency along with the sensor operating voltage (Vset). The current is measured and the impedance is calculated. This measurement is repeated across a range of frequencies, and the impedance output is then examined to look for specific frequency dependent membrane characteristics.

EIS can identify poorly performing sensors and instances where the electrode has been partially pulled out of the tissue (and therefore is no longer sensing correctly). This is particularly useful as it can be difficult for a patient to know when sensor pull-out occurs when wearing miniaturized components. More importantly, EIS may be used as a predictive diagnostic tool, alerting the system to issues before the sensor signal changes drastically.

In the example shown in Figures I7A and 17B, e.g., EIS detects a drop in low frequency Nyquist slope (Figure 17A), which predicts a drift in sensor signal (sensor anomaly) shown in Figure 17B. In Figure 17C, electrochemical sensors are periodically interrogated and analyzed using EIS, and the response is used to proactively identify potential faults or failures, such that the sensor may be recalibrated or shut down before it results in inaccurate glucose measurements. In short, such predictive diagnosis provides the system the opportunity to mitigate the issue through suspended data or calibration request, thereby minimizing the effect(s) on the patient.

Other methods-not involving EIS measurements-for detecting signal anomalies include short periods where the calculated glucose would not be correct, periods where the

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 signal needs stronger filtering, or instances where the sensor’s glucose sensitivity has changed (in this case, require a new calibration).

For the optical sensors, the glucose value is calculated from the ratio between the assay signal and the reference signal, as detailed previously. These two signals are independently interrogated and are used to detect failures during use. Both the reference and the assay signal must be within a certain interval (dynamic range), and if outside these intervals, the sensor’s performance is not to be trusted. Additionally, if the rate of change exhibited by either the reference or the assay signal is outside the given limits, then this behavior will cause a failure alarm. An example is detecting a misalignment between the reader and the sensor. This will cause both signals to drop to a very low value in a very short period of time and hence cause an alarm based on the signal gradient control function.

The orthogonally redundant system allows comparison of signals. Based on the signal characteristics of each sensor, a reliability index is created for each signal. Comparing the reliability index of each sensor and the signals themselves allows confirmation of suspected faults, or provides assurance for the algorithm that both signals are accurate. For situations when the reliability of the combined signal is under a threshold, a finger-stick confirmation may be necessary. In other regions, the system could give a range of values, such as an expected minimum glucose value to be used for bolusing purposes. Micro-environmental aspects, such as drugs or temperature changes, have the potential to influence the system, but the optical sensor does not necessarily respond in the same way as the electrochemical sensor. For example, electro-active species can cause an increased current in the electrochemical sensor, but the optical sensor is not affected the same way or possibly unaffected due to this.

Failure detection in the system of the instant invention is quire robust, as a multi-sensor system has an added benefit of being able to confirm failures. Orthogonally redundant sensors increase this benefit, since the optical sensor and electrochemical sensor have different failure modes and different responses to interfering compounds.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Duration of Wear

The orthogonally redundant sensor system increases duration of wear and reliability of data through the use of redundancy, fault detection, and advanced algorithms to ensure at least one sensor is providing reliable measurements. In addition, the sensor lifetime is limited to the specified duration of wear to ensure reliability of data.

Duration of wear can be classified in two ways: (1) the overall lifetime of the sensor; and (2) the percent of time during wear that the sensor is displaying accurate data. The sensor lifetime is limited through loss of sensitivity and drift in-vivo that may be caused by environmental influences. The orthogonally redundant sensor system decreases the frequency of early sensor termination through the use of redundancy and dual sensing technologies, ensuring at least one sensor is providing reliable measurements for an increased duration and safeguarding against environmental influences. Additionally, body worn devices must be safeguarded against sensor pull-outs that result in early termination. As such, custom adhesives for both patch and overtape may be implemented for the combination device.

As mentioned previously (see above section on “Accuracy”), failure detection algorithms limit the inaccurate data that is visible to the patient but, as a result, may limit the data to such an extent that the continuous sensing benefits are not realized. Utilizing a redundant sensing system improves the percent of time the sensor displays data because the frequency of anomalies simultaneously in both sensors is significantly less than in a single sensor.

Additionally, sensors may also stay implanted beyond seven days. Sensors implanted beyond the labeled lifetime may be more likely to provide erroneous data. Therefore, to ensure reliability, it is important that the system limit sensor lifetime to the labeled time period. This is accomplished through the system design utilizing embedded firmware timers in the instrumentation coupled with diagnostics methods that can detect whether a sensor has been

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 previously used. By combining embedded timers and intelligent diagnostics, the system ensures that sensors are not used beyond the period of optimal reliability and accuracy.

Form Factors

While combining two sensor systems into a single device requires more instrumentation and battery capacity, miniaturization and integration methods may be used to ensure that the transmitter device 10 is similar in size to other CGM devices.

Device size, form factor, and use model play a significant role in therapy adoption. When placing the device on the body, a larger, simply-shaped device tends to be easier to handle, whereas a smaller, organically-shaped device tends to be more preferable to wear. In preferred embodiments of the invention, a well-balanced design based on the foregoing factors is adopted.

In order to avoid unsightly distortions when the device is worn under clothing, patients generally prefer a larger device footprint over added height. Because the device in accordance with embodiments of the instant invention contains more complex and substantial internal components than other CGMS products currently available, it is understood that the footprint of the assembly is slightly larger than what is currently available. Thus, the device is as slim and sleek as possible, with minimal sacrifice in the way of volumetric efficiency.

Wafer-level design and production methods are used in a novel way to minimize the size of the optoelectronic (or optical) interrogating system. A Stacked Planar Integrated Optical System (SPIOS) may be created by fixing one multi-functional filter layer between two injection molded layers of optical components. The SPIOS forms a solid block, which is selfsupporting. The SPIOS is shown in the right-hand side of Figure 18, with the left-hand side showing an example of an optical system built from discrete components.

More specifically, in an embodiment of the invention shown in Figure 19, the inventive optical interrogating system may be designed to be manufactured as a SPIOS (also referred to as a “Wafer Scale Optical System” or a “Wafer Level Optical System”). As shown in Figure

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

19, the SPÏOS includes various layers that are stacked and aligned. In the wafer layer 1610, one or more light sources (e.g., LEDs and photodiodes) and detectors may be laid out on a wafer. Alternatively, they may be naked chips (e.g., sold by Avago Technologies or Hamamatsu), which are individually aligned and laminated onto the SP1OS units.

One or more optical layers 1620 may include mirrors, absorbers, and/or other optical components laid out on a wafer-sized injection molded disk. Mold inserts defining optical surfaces are made by a diamond tuming/milling company (e.g., Kaleido Technology in Denmark). Gold or protected silver is applied to mirror surfaces, e.g., by sputtering, while any absorbers are masked off during the process.

The optical filter layer 1630 includes a wafer-sized glass substrate with optional (e.g., dielectrical) coatings. Specifically, multilayer optical coatings may be applied on both sides of the glass substrate using ion-assisted sputtering to form durable coatings. The technique is similar to that used in manufacturing fluorescence filters by, e.g., Semrock in the United States and Delta in Denmark. Thus, in one example, diclcctrical coatings applied on both sides of the substrate operate to filler excitation light, as well as the resulting fluorescence.

As shown in Figure 19, in one embodiment, a wafer layer 1610 may be followed by an optical layer 1620, an optical filter layer 1630, and another optical layer 1620. The entire stack is then thoroughly aligned and laminated, e.g., by gluing, and the connections are bonded onto the chips. The stack is then diced 1640 using, e.g., a diamond saw to form multiple assembled SPIOS units 1670, which can then be mounted and connected to electronics.

The above-described system may be made small and is suitable for large-scale production. The system may be used for interrogating a sensor in a light scattering environment, such as a sensor implanted into the skin, as well as a fiber sensor. Packaging may be used to block out ambient light. Moreover, as shown in Figure 20. to save board space, a LED driver, two amplifier chains, and a temperature sensor specific to the optical sensor may be integrated into a custom chip and added to the analog front-end (AFE) for the

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707 electrochemical sensor, e.g., the AFE designed for use with the MiniLink® transmitter (MiniLink® available from Medtronic, Inc.).

In embodiments of the invention, the LED light source 1320 shown in Figure 5 may be replaced with a red laser diode for illumination of the assay chemistry. The nature of a laser 5 diode (smaller source diameter emission angle compared to an LED) provides for reduction of the size of the optical system relating to the excitation of the fiber sensor, as well as enhanced coupling efficiency from the laser diode to the fiber sensor. The latter, in turn, leads to a higher signal to noise ratio, which again leads to shorter measurement times and a smaller battery size. Battery capacity may be reduced by as much as 75%, which also significantly 10 reduces the size of the transmitter 10.

Moreover, the higher excitation efficiency and narrower wavelength range of the laser diode reduce stray light problems, such that a lower light pickup may be accepted at the detector side. As a result, the part of the optical system relating to fluorescence detection is reduced. All in all, the use of a laser diode may reduce the size of the optical system to about 15 75% of the size of an optical system using LED excitation. Thus, e.g., a transmitter device 10 employing a laser diode as the illumination source of its optical interrogating system may have a volume of about 15 cm³ and a weight of about 10g.

To use a red laser diode, the (assay) chemistry must be red-shifted, meaning that new fluorophores operating at higher wavelengths must be used, in order to operate in a range 20 where the laser diode is able to excite the chemistry. In this regard, it has been found that several fluorophores, including AF647, QSY 21, and AF750 may be used in conjunction with a laser diode source at 645 nm. See Figure 21.

To further miniaturize the optical system and thus reduce the size of the transmitter 10, it is beneficial to incorporate the laser diode into the stacked planar integrated optical system 25 (SP1OS) format discussed above, ft has been found that such an implementation further decreases the transmitter size to about 11 cm³.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Sterilization, Storage, and Shelf-Life Stability

A typical electrochemical sensor-e.g., the Enlite® sensor-may normally be stored at room temperature and ambient atmospheric relative humidity levels. To enable storage of the orthogonally redundant sensor (which may include such an electrochemical sensor) under these same conditions and, at the same time, maintain desired usability, embodiments of the invention include a dry version of the assay for the optical sensor. The term “dry chemistry” as used in this context refers to the dry form of the assay as compared to the original wet composition. The dry chemistry may, for example, be in the form of a freeze dried powder or suspended in a polymer, and not only enables dry packaging and dry storage, but also improves shelf life stability. The assay chemistry may, e.g., be dried via a lyophilization step, which includes freezing the assay and sublimation of liquid media through rapid vacuum drying.

Moreover, as noted previously, a typical electrochemical sensor is usually sterilized through a (c-beam) radiation sterilization process. Application of the same sterilization process to an optical sensor, or to an orthogonally redundant sensor that includes an optical sensor, however, presents practical challenges, as e-beam radiation may detrimentally affect the assay chemistry' and, as such, result in loss of (optical) sensor response. In this regard, in embodiments of the invention, a protective formulation may be included in the assay to counteract the harmful effects of e-beam on, e.g., MBL and fluorescent dyes. The protective formulation includes protective chemical agents that, in addition to withstanding radiation sterilization effects, also facilitate sensor hydration and startup.

With regard to the above-described dry' chemistry and protective formulation, it has also been discovered that, even without the protective formulation, optical sensors using the dry chemistry described above show little change in sensor response when exposed to e-beam radiation. In addition, the dry chemistry in fiber sensors has been shown to retain its stability in the dry state for three months at 5°C.

CA 02845804 2014-02-18

WO 2013/036493 PCT/US2012/053707

Connectivity and Data Warehousing

Connectivity and data warehousing are integrated with the orthogonally redundant sensor system through communication with networking products available, e.g., from Medtronic, Inc., including a handheld monitor (such as, e.g., MySentry^IM Glucose Monitor) and CareLink® therapy management software.

In one embodiment, the Medtronic system provides data transfer capability between the Medtronic Patient Network (MPN) and internet-based Medtronic CareLink® therapy management software system. This system is designed to efficiently provide data downloading, warehousing, and reports for patients and their healthcare providers (HCPs). Patients and HCPs use CareLink® reports in many ways, including reviewing data, understanding behavior, and optimizing therapy. Additional reports provide decision support in a ‘professional” version of the CareLink® system (available to HCPs) that streamlines data analysis in the clinical setting and highlights opportunities for therapy modifications that can drive improved outcomes.

In a further embodiment, a Connected Care system includes an On Body Communicator (OBC) utilizing currently available mobile networks technology. The system provides the Patient, a Loved One, and a Physician access to information from the Patient's MPN in near real-time. See Figure 22.

The primary function of the OBC is to provide mobile ambulatory MPN connectivity and data processing. The OBC communicates with the Medtronic proprietary RF protocol to establish communications with the MPN and deliver them to “the cloud” through a cellular network capability. Data can then be retrieved from the cloud and sent to the CareLink® Personal internet-based system. When a cellular signal is unavailable, the OBC continues to maintain operations required to collect and process data from the MPN until the cellular signal is re-established. Once data in the cloud is available in a near real-time, the CareLink® system

WO 2013/036493

PCT/US2012/053707 can deliver features designed for commercially available web enabled electronics devices such as smart phones and tablets.

As noted previously in connection with Figures I and 11, in a preferred embodiment, the OBC may be in the form of a handheld controller or monitor with integrated blood glucose 5 meter used for calibration. The handheld monitor is designed to work in conjunction with the orthogonally redundant sensor system. In addition to sending data to the cloud, the handheld monitor improves accuracy through the use of algorithms to provide an error check, ensuring that inaccurate blood glucose readings are not communicated.

While the description above refers to particular embodiments of the present invention, it will be understood that many modifications may be made without departing from the spirit thereof. The accompanying claims are intended to cover such modifications as would fall within the true scope and spirit of the present invention.

The presently disclosed embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the invention as described herein, and all changes which come within the meaning and range of equivalency of the invention as described herein are therefore intended to be embraced therein.