Automated near-fall detector

FIELD AND BACKGROUND OF THE INVENTION

p-0002The present invention, in some embodiments thereof, relates to motion detection, and more particularly, but not exclusively, to a system useful for identifying gait or fall related motion.

p-0003A public health issue of concern is the incidence of falls, in which a person falls to the ground from an upright position while standing or walking The problem of falls affects the elderly in general, and is of particular concern for older persons and others who have a movement disorder or other illness that affects balance and motor control, such as Parkinson's disease.

p-0004The effect of a fall on an elderly person can be particularly serious since many elderly people have weak or brittle bones, and are generally further weakened by other illnesses and the effects of aging. In some cases a fall causes the death of a person, either at the time of the fall or indirectly as a result of the injuries sustained. The type of injuries commonly experienced may include one or more of: a broken or fractured hip and other bones, head injuries, internal and external bleeding, and soft tissue and skin damage. The patient will most likely suffer a great deal of pain and may require hospitalization. In addition, he or she may face the prospect of long term or permanent loss of mobility, since their age and condition may mean that the injuries will take a long time to heal or may never heal completely. The patient may be plagued by fear of a recurrence, so that their mobility and confidence is further compromised. Accordingly, even if death is avoided, the injuries suffered from a fall can be devastating to the person's physical and mental well-being.

p-0005Various systems have been proposed to automatically identify falls, so that an action can be triggered to help alleviate the damage caused by the fall. For example, upon detecting that a fall has occurred, a system could notify a relative or doctor to check up on the patient. Dinh et al. in “A Fall Detection and Near-Fall Data Collection System” (Microsystems and Nanoelectronics Research Conference (MNRC), October 2008) describe a wearable device containing a 3-axis accelerometer, a 2-axis gyroscope, and a heart beat detection circuit. Data collected from the sensors is beamed wirelessly to a receiver connected to a computer. The researchers observed that combining the accelerometer data with the gyroscope data produced good results in identifying whether a fall had occurred.

p-0006Bourke et al. in “Distinguishing Falls from Normal ADL using Vertical Velocity Profiles”, (IEEE Conference on Engineering in Medicine and Biology, August 2007) observe that a single threshold applied to the vertical velocity profile of the trunk may distinguish falls from activities of daily living (ADL).

p-0007In another paper, Wu and Xue in “Portable Preimpact Fall Detector With Inertial Sensors” (IEEE Transactions on Neural Systems and Rehabilitation Engineering, April 2008), describe a portable preimpact fall detector that detects a pending fall at its inception, so that an inflatable hip protector can be triggered in time to break the fall. The detector was equipped with an orientation or inertial sensor that included triaxial accelerometers and triaxial angular rate sensors, and used a detection algorithm based on the inertial frame velocity profile of the body. In particular, the inertial frame vertical velocity magnitude was measured and compared to a threshold value to identify a fall. The system was tested in a variety of activities to determine the threshold level of inertial frame vertical velocity magnitude.

SUMMARY OF THE INVENTION

p-0008An aspect of some embodiments of the invention relates to detection of gait irregularity and/or of near fall.

p-0009In an exemplary embodiment of the invention, a near fall is characterized based on its vertical acceleration profile, for example, the rate of change of vertical acceleration being above a threshold. Optionally, a comparison to a threshold uses inexact methods, for example fuzzy logic. Optionally or alternatively, the comparison is of a function of acceleration to a function of the threshold. Optionally, the threshold is dynamic, for example, as a function of context of the gait and/or of recent movement parameters.

p-0010In some exemplary embodiments of the invention, gait irregularity is characterized based on vertical acceleration. Typically, corresponding to gait's steps movements, movement's acceleration signal exhibits a generally cyclic pattern with peaks. In some embodiments, irregularity is determined when the periods of the cycles (e.g. between peaks) vary above a threshold. In some embodiments, the irregularity is determined when the shape of the cycles vary above a threshold, where the variability of the shape is determined, for example, by variations in cross-correlation between the cycles. In some embodiments, the irregularity is determined by a frequency spread of the acceleration signal, such as obtained with a Fourier transform.

p-0011Optionally, a comparison to a threshold uses inexact methods, for example fuzzy logic. Optionally or alternatively, the comparison is of a function of acceleration to a function of the threshold. Optionally, the threshold is dynamic, for example, as a function of context of the gait and/or of recent movement parameters.

p-0012In some embodiments, a combination of two or more of the methods, i.e. cycles time, cycles shape and frequency spread, is used to determine irregularity.

p-0013In some embodiments, the irregularity is checked along a certain or determined time. Optionally, the irregularity is checked within a moving window of a certain or determined time.

p-0014Alternatively or additionally to evaluation of near fall and/or gait irregularity by parameters or values derived from the acceleration, in some exemplary embodiments of the invention determination of near fall and/or gait irregularity is based on the waveform of the acceleration (or other movement signals).

p-0015In some embodiments, the waveform of gait acceleration over a certain period is evaluated against a reference waveform or library of waveforms of gait acceleration, and near fall and/or gait irregularity is determined or classified according to a degree of matching or mismatching with the reference waveform(s).

p-0016In some embodiments, the waveform of a subject is matched against a reference waveform by methods of pattern matching such as correlation or cross-correlation or wavelet matching or machine learning (e.g. neural networks) or any combination of methods of the art.

p-0017In some exemplary embodiments of the invention, a derivative of the accelerations is used to determine near fall and/or gait irregularity. Optionally, other parameters such as angular velocity or tilt are used.

p-0018An aspect of some embodiments of the invention relates to gait regulation assistance. In some embodiments, irregularity in gait is detected, such as described above. Responsive to a determined gait irregularity of a person, the person is prompted, such as by audio message or tactile incitement, to adjust and/or stabilize the gait (cuing signals).

p-0019An aspect of some embodiments of the invention relates to enhancing a Timed Up and Go (TUG) test to assess the tendency of a person to fall (persons prone to fall). In some embodiments, the enhancement is based on the rate of change of position during sitting or rising (jerks), such as a time derivative of the vertical acceleration. In some embodiments, the tendency to falling is assessed when the rate of change of the acceleration is above a threshold. In some embodiments, the threshold is based on the rate of change of acceleration of healthy person or persons. Optionally or additionally, the threshold is based on the physiological state of the person being assessed, such as neurological disorder.

p-0020There is provided in accordance with an exemplary embodiment of the invention, a method of gait data collection, the method comprising:

p-0021A method of gait data collection, the method comprising: <ul><li id="ul0001-0001" num="0000"><ul><li id="ul0002-0001" num="0021">collecting movement data, and</li><li id="ul0002-0002" num="0022">determining from said data at least one irregularity of the gait.</li></ul></li></ul>

p-0022In some embodiments, an irregularity comprises a near fall.

p-0023In some embodiments, an irregularity comprises a fall.

p-0024In some embodiments, determining comprises determining from said data a movement parameter that includes a third order derivative of position, and counting at least a near fall based on said movement parameter.

p-0025In some embodiments, determining comprises matching the pattern with respect to time of the movement data with a reference pattern.

p-0026In some embodiments, the reference pattern represents proper gait pattern. In some embodiments, the reference pattern represents improper gait pattern.

p-0027In some embodiments, the reference pattern represents a gait pattern exhibiting at least one near fall.

p-0028In some embodiments, the matching classified the data as exhibiting fall, near fall or lack thereof.

p-0029In some embodiments, wherein the matching comprises at least one of correlation, cross-correlation, wavelets matching or neural networks or a combination thereof.

p-0030In an exemplary embodiment of the invention, the method comprises comparing said movement parameter with a threshold value to identify a near fall.

p-0031In an exemplary embodiment of the invention, said movement parameter comprises a difference between a maximum acceleration derivative and a minimum acceleration derivative. Optionally, said movement parameter relates to movement in substantially a vertical direction.

p-0032In an exemplary embodiment of the invention,

p-0033determining from said data further includes determining a second movement parameter,

p-0034comparing said movement parameter further includes comparing said second movement parameter with a second threshold value, and

p-0035counting at least a near fall comprises counting at least a near fall if said movement parameter exceeds said threshold value and said second movement parameter exceeds said second threshold value.

p-0036In an exemplary embodiment of the invention, said second movement parameter includes a second order derivative of position. Optionally or alternatively, said movement parameter and said second movement parameter relate to movement in substantially a vertical direction.

p-0037In an exemplary embodiment of the invention, said threshold value is a predetermined value.

p-0038In an exemplary embodiment of the invention, said threshold value is a continuously updated function of said movement parameter. Optionally, said function is a mean of said movement parameter plus a multiple of a standard deviation of said movement parameter.

p-0039In an exemplary embodiment of the invention, determining a movement parameter comprises collecting acceleration data and taking a derivative of said acceleration data with respect to time.

p-0040In an exemplary embodiment of the invention, determining a movement parameter comprises collecting velocity data and taking a second order derivative of said velocity data with respect to time.

p-0041In an exemplary embodiment of the invention, determining a movement parameter comprises collecting position data and taking a third order derivative of said position data with respect to time.

p-0042In an exemplary embodiment of the invention, said count of at least a near fall provides a quantitative measure of effectiveness of therapeutic interventions.

p-0043There is provided in accordance with an exemplary embodiment of the invention, a method of gait data collection, the method comprising:

p-0044collecting movement data,

p-0045determining from said data a plurality of movement parameters, each of said movement parameters including at least one of a second order derivative of position and a third order derivative of position,

p-0046comparing each of said movement parameters with an associated threshold value, and

p-0047counting at least a near fall if a predetermined combination of movement parameters from said plurality of movement parameters exceeds their associated threshold value.

p-0048There is provided in accordance with an exemplary embodiment of the invention, a method of gait data collection, the method comprising:

p-0049collecting movement data,

p-0050extracting from said movement data an indicator indicating a loss of control,

p-0051counting at least a near fall if said indicator indicates said loss of control.

p-0052There is provided in accordance with an exemplary embodiment of the invention, a device to detect falling body movement, the device comprising:

p-0053a sensor operatively connected to said body and responsive to movement of said body, and

p-0054a processor to receive movement data from said sensor and to process said movement data to identify events that are at least near falls.

p-0055In an exemplary embodiment of the invention, said processor is configured to log a record of events that are at least near falls. Optionally or alternatively, said sensor is responsive to movement of said body in substantially a vertical direction. Optionally or alternatively, said sensor is responsive to acceleration of said body.

p-0056In an exemplary embodiment of the invention, the device includes a user interface to communicate with a user of said device.

p-0057In an exemplary embodiment of the invention, said sensor and said processor are enclosed in a housing.

p-0058In an exemplary embodiment of the invention, said processor is located remote from said sensor.

p-0059In an exemplary embodiment of the invention, the device includes a radio transmitter operatively connected to said sensor and a radio receiver operatively connected to said processor,

p-0060wherein said transmitter and said receiver are configured to enable said processor to receive movement data from said sensor in real time.

p-0061There is provided in accordance with an exemplary embodiment of the invention a method for assisting a person's gait, comprising:

p-0062(a) detecting, based on time derivation of gait movements, irregularity in the gait; and

p-0063(b) providing gait regulating cueing signals.

p-0064There is provided in accordance with an exemplary embodiment of the invention an apparatus for assisting a person's gait, comprising:

p-0065(a) a sensor operatively connected to the person and responsive to movement of said person;

p-0066(b) a processor adapted to receive movement data from said sensor and to process said movement data to detect irregularity in the movement; and

p-0067(c) at least one device operable to provide cuing signals responsive to detected irregularity.

p-0068In some embodiments, the signals are at least one of audible, tactile or visual.

p-0069There is provided in accordance with an exemplary embodiment of the invention a method for augmenting a Timed Up and Go test, comprising:

p-0070(a) determining rate of change of acceleration of movement about at least one of seating or rising; and

p-0071(b) screening, based on the rate of change of the acceleration, a tendency to fall.

p-0072In some embodiments, the screening is determined of a rate larger than that of a healthy person.

p-0073In some embodiments, the screening is determined when the rate of the acceleration of a sitting movement is about 1 g/sec

p-0074In some embodiments, the screening is determined when the rate of the acceleration of a rising movement is about 2 g/sec

p-0075Unless otherwise defined, all technical and/or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention pertains. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of embodiments of the invention, exemplary methods and/or materials are described below. In case of conflict, the patent specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and are not intended to be necessarily limiting.

p-0076Implementation of the method and/or system of embodiments of the invention can involve performing or completing selected tasks manually, automatically, or a combination thereof. Moreover, according to actual instrumentation and equipment of embodiments of the method and/or system of the invention, several selected tasks could be implemented by hardware, by software or by firmware or by a combination thereof using an operating system.

p-0077For example, hardware for performing selected tasks according to embodiments of the invention could be implemented as a chip or a circuit. As software, selected tasks according to embodiments of the invention could be implemented as a plurality of software instructions being executed by a computer using any suitable operating system. In an exemplary embodiment of the invention, one or more tasks according to exemplary embodiments of method and/or system as described herein are performed by a data processor, such as a computing platform for executing a plurality of instructions. Optionally, the data processor includes a volatile memory for storing instructions and/or data and/or a non-volatile storage, for example, a magnetic hard-disk and/or removable media, for storing instructions and/or data. Optionally, a network connection is provided as well. A display and/or a user input device such as a keyboard or mouse are optionally provided as well.

BRIEF DESCRIPTION OF THE DRAWINGS

p-0078Some embodiments of the invention are herein described, by way of example only, with reference to the accompanying drawings. With specific reference now to the drawings in detail, it is stressed that the particulars shown are by way of example and for purposes of illustrative discussion of embodiments of the invention. In this regard, the description taken with the drawings makes apparent to those skilled in the art how embodiments of the invention may be practiced.

p-0079In the drawings:

p-0080<figref idrefs="DRAWINGS">FIGS. 1A</figref>, <b>1</b>B, and <b>1</b>C are schematic views of a person walking, having a near fall, and recovering to resume walking, respectively, while wearing an automated near-fall detector, in accordance with an embodiment of the invention;

p-0081<figref idrefs="DRAWINGS">FIGS. 2A</figref>, <b>2</b>B, and <b>2</b>C are schematic views of the automated near-fall detector of <figref idrefs="DRAWINGS">FIG. 1</figref>, in accordance with several embodiments of the invention;

p-0082<figref idrefs="DRAWINGS">FIGS. 3A and 3B</figref> are flow charts describing a method of gait data collection, in accordance with an embodiment of the invention;

p-0083<figref idrefs="DRAWINGS">FIG. 4</figref> shows graphs of derived parameters Vertical Maximum Acceleration and Vertical Maximum Peak to Peak Derivative, in accordance with an embodiment of the invention;

p-0084<figref idrefs="DRAWINGS">FIG. 5</figref> shows exemplary charts of stride acceleration and frequency spread of a healthy person and a person with Parkinson disease, respectively; and

p-0085<figref idrefs="DRAWINGS">FIG. 6</figref> shows exemplary charts of Timed Up and Go (TUG) of healthy person and a person prone to falling, respectively.

DESCRIPTION OF SPECIFIC EMBODIMENTS OF THE INVENTION

p-0086The present invention, in some embodiments thereof, relates to motion detection, and more particularly, but not exclusively, to a system useful for identifying gait or fall related motion.

p-0087Before explaining at least one embodiment of the invention in detail, it is to be understood that the invention is not necessarily limited in its application to the details of construction and the arrangement of the components and/or methods set forth in the following description and/or illustrated in the drawings and/or the Examples. The invention is capable of other embodiments or of being practiced or carried out in various ways.

p-0088In an exemplary embodiment of the invention, a near fall is characterized based on its vertical acceleration profile, for example, the rate of change of vertical acceleration being above a threshold. Optionally, a comparison to a threshold uses inexact methods, for example fuzzy logic. Optionally or alternatively, the comparison is of a function of acceleration to a function of the threshold. Optionally, the threshold is dynamic, for example, as a function of context of the gait and/or of recent movement parameters.

p-0089In some exemplary embodiments of the invention, gait irregularity is characterized based on vertical acceleration. Typically, corresponding to gait's steps movements, movement's acceleration signal exhibits a generally cyclic pattern with peaks. In some embodiments, irregularity is determined when the periods of the cycles (e.g. between peaks) vary above a threshold. In some embodiments, the irregularity is determined when the shape of the cycles vary above a threshold, where the variability of the shape is determined, for example, by variations in cross-correlation between the cycles. In some embodiments, the irregularity is determined by a frequency spread of the acceleration signal, such as obtained with a Fourier transform.

p-0090Optionally, a comparison to a threshold uses inexact methods, for example fuzzy logic. Optionally or alternatively, the comparison is of a function of acceleration to a function of the threshold. Optionally, the threshold is dynamic, for example, as a function of context of the gait and/or of recent movement parameters.

p-0091In some embodiments, a combination of two or more of the methods, i.e. cycles time, cycles shape and frequency spread, is used to determine irregularity.

p-0092In some embodiments, the irregularity is checked along a certain or determined time. Optionally, the irregularity is checked within a moving window of a certain or determined time.

p-0093Alternatively or additionally to evaluation of near fall and/or gait irregularity by parameters or values related to the size of the acceleration or other movement signal (i.e., above or below a threshold), in some exemplary embodiments of the invention determination of near fall and/or gait irregularity is based on the waveform of the acceleration (or movement signal).

p-0094In some embodiments, the waveform of gait acceleration over a certain period is evaluated against a reference waveform(s) of gait acceleration, and near fall and/or gait irregularity is determined or classified according to a degree of matching or mismatching with the reference waveform. In some embodiments, the classification comprises a fall or near fall event or the lack thereof.

p-0095In some embodiments, the waveform of acceleration of the gait of a subject is matched against a reference waveform by methods of pattern matching such as correlation or cross-correlation or wavelet matching or machine learning (e.g. neural networks) or any combination of methods of the art. Optionally the determination or classification of gait irregularity and/or near fall by matching methods is augments by other methods such as fuzzy logic.

p-0096When a subject's waveform sufficiently deviates from a reference signal representing a proper gait, the subject is determined to exhibit irregular gait. Optionally, by features matching between the waveforms a near fall is determined if characteristic features are different between the waveforms.

p-0097When a subject's waveform sufficiently matches a reference signal representing an improper gait, the subject is determined to exhibit irregular gait. Optionally, when the subject's waveform sufficiently matches a waveform with near fall events, the subject is determined to exhibit near fall behavior. Optionally, by features matching between the waveforms a near fall is determined if characteristic features are similar between the waveforms.

p-0098For example, the acceleration waveform of a subject is matched against a waveform representing a healthy gait, and if the waveforms deviated above a threshold the subject's gait is determined to be irregular. Optionally, features of the waveforms are matched and based on dissimilarities such as missing or different features between the waveforms, the subject's gait is determined to exhibit near fall behavior.

p-0099As another example, the acceleration waveform of a subject is matched against a waveform representing a person having improper gait. If, based on a threshold or other measures, the waveforms are sufficiently close and/or exhibit similar features the subject's gait is determined to be irregular or having near fall characteristics. Optionally, features of the waveforms are matched and according to some measures, such as missing or different features between the waveforms, the subject's gait is determined to exhibit near fall behavior.

p-0100In some embodiments, a ‘ healthy’ or ‘proper’ reference waveform is based on the gait of healthy persons, optionally of about the age of the subject being evaluated. For example, acceleration waveforms of healthy persons are collected and combined, such as by scaling and averaging or by any other methods, to provide a representative waveform of proper or regular gait. Optionally, the representative waveform is based, at least partially, on the gait acceleration of other neurologically diseased while they exhibit regular gait. Optionally, the representative waveform is based, at least partially, on synthetic waveform computed to represent a proper gait.

p-0101In some embodiments, an ‘ill’ or ‘improper’ reference waveform is based on the gait of neurologically diseased persons, optionally of about the age and/or disorder of the subject being evaluated. For example, acceleration waveforms of persons exhibiting irregular or disordered or near fall behavior are collected and combined, such as by scaling and averaging or by any other methods, to provide a representative waveform of improper gait. Optionally, the representative waveform is based, at least partially, on the gait acceleration of other neurologically diseased while they exhibit irregular gait. Optionally, the representative waveform is based, at least partially, on synthetic waveform computed to represent an improper gait.

p-0102In some embodiments, in order to improve or refine the evaluation of a subject's waveform, the waveform is matched against a plurality of reference waveforms, either proper and/or improper waveforms. For example, the subject's waveform is matched against both proper and improper references and the irregularity or near fall characteristics are determined by a combination of the matching results.

p-0103In some embodiments, the representative waveforms are updated from time to time to form a library or repository of reference waveforms.

p-0104In some exemplary embodiments of the invention a derivative of the accelerations are used to determine near fall and/or gait irregularity. Optionally, other parameters such as angular velocity or tilt are used such as to refine the determination of fall and/or gait irregularity.

p-0105In some embodiments, the presence or absence of a near fall or other gait irregularity is made by combining methods based on pattern recognition of the waveforms with those that are based on threshholding of the acceleration jerk or other derived movement parameters.

p-0106In some embodiments, irregularity in gait is detected, such as described above. Responsive to a determined gait irregularity of a person, the person is prompted, such as by audio message or tactile incitement, to adjust and/or stabilize the gait (cuing signals).

p-0107In some embodiments, a Timed Up and Go (TUG) test to assess the tendency of a person to fall (persons prone to fall) is enhanced. In some embodiments, the enhancement is based on the rate of change of position during sitting or rising (jerks), such as a time derivative of the vertical acceleration. In some embodiments, the tendency to falling is assessed when the rate of change of the acceleration is above a threshold. In some embodiments, the threshold is based on the rate of change of acceleration of healthy person or persons. Optionally or additionally, the threshold is based on the physiological state of the person being assessed, such as neurological disorder.

1. Overview

p-0108<figref idrefs="DRAWINGS">FIGS. 1A</figref>, <b>1</b>B, and <b>1</b>C show a near fall detector device <b>20</b>, according to an embodiment of the invention, in a typical application being used by a walking person <b>22</b>. Person <b>22</b> may be a man or woman of any age and of any physical condition. In this example near fall detector <b>20</b> is a device attached to a belt <b>24</b> worn by person <b>22</b>. As will be discussed in greater detail below, near fall detector <b>20</b> optionally uses signal processing methods to monitor the quality of a walking person's gait or ambulatory movement, and responds or records in some fashion in the event that the person's walk is interrupted by a near fall or a real fall. Optionally, detector <b>20</b> is also capable of detecting a fall or near fall that may be experienced by a person that is standing or sitting.

p-0109In <figref idrefs="DRAWINGS">FIG. 1A</figref> person <b>22</b> is shown walking in a normal fashion. At some later point in time, as shown in <figref idrefs="DRAWINGS">FIG. 1B</figref>, person <b>22</b> experiences a near fall. The near fall, also called a stumble or misstep, is a momentary loss of balance by the person from which the person recovers. By contrast, in a real or actual fall (or just “fall”) the person does not recover and continues to fall until he or she comes to rest on the ground, floor, or other lower level. <figref idrefs="DRAWINGS">FIG. 1B</figref> illustrates some characteristics of an example of a near fall. As may be seen, the person's legs have slipped so they are no longer directly underneath, and accordingly the person's center of gravity <b>26</b> has moved off center so that the person experiences a sensation of loss of balance. As most people may relate, the person's arms thrust out to compensate in an effort to recover balance and avoid falling. In this example person <b>22</b> is successful at avoiding the fall, and is shown in <figref idrefs="DRAWINGS">FIG. 1C</figref> at a later point in time resuming his or her walk. Near fall detector <b>20</b> however has detected the incident shown in <figref idrefs="DRAWINGS">FIG. 1B</figref>. This is indicated, by way of example, in the enlarged representation of the detector in inset <b>28</b> in <figref idrefs="DRAWINGS">FIG. 1C</figref>, which shows detector <b>20</b> displaying the words “Near Fall”. In other examples, detector <b>20</b> might log a record of the incident and not display anything, or detector <b>20</b> might query the user to confirm the near fall.

p-0110The near fall illustrated in <figref idrefs="DRAWINGS">FIG. 1B</figref> could occur in any direction, have a degree of magnitude or force behind it, and be due to any cause. For example, the near fall could be in a forward direction (as shown in the figure), as might occur due to tripping. Other types of near falls include, for example, a backward near fall caused by a slippery floor, or a sideways near fall caused by a misstep. The person might also have a near fall directed straight down, for example due to fainting. Near falls may be caused by external circumstances, such as an unexpected obstacle or slippery surface, or by circumstances internal to the person, such as by fainting, general weakness, or a movement disorder. In many cases the near fall is caused by a combination of the two. For example, an obstacle may be encountered that a healthy person would easily avoid, but that precipitates a near fall in an older person with poor eyesight and a slow reaction time. In an exemplary embodiment of the invention, near fall detection is practiced in clinical/diagnostic settings, where a patient is given a task, such as an obstacle course, and his performance thereon monitored.

p-0111In addition to detecting the incident of a near fall, near fall detector <b>20</b>, in some embodiments of the invention, may also detect the magnitude and/or direction of the near fall. Further, as will be discussed in greater detail below, near fall detector <b>20</b> in some embodiments of the invention performs gait data collection and/or includes an algorithm configured to detect the occurrence of actual falls as well as near falls.

p-0112The inventors have realized that many people who have experienced actual falls, or that are prone to falling, may in fact only fall a relatively small number of times. This does not detract from the seriousness of the problem, since all it takes is one bad fall to seriously injure a person. However, it does suggest that for such people it can be difficult to collect meaningful data to prevent future falls, especially if their memory is faulty and/or if interrogation occurs at a time after such an event. The inventors have further observed that people at risk of falling often have multiple near falls for every actual fall that they experience, and also prior to their falling for the first time. In an exemplary embodiment of the invention, the detection of near falls provides insight into a person's condition that may assist in the diagnosis and prevention of subsequent real falls by that person.

p-0113As discussed in greater detail below, near fall data can provide quantifiable parameters whose value can be used to better assess the person at risk. Additionally, when combined with data on a person's actual falls (which as noted can also be obtained from detector <b>20</b> of some embodiments of the present invention), a diagnostician can obtain a ratio (or other relationship) of actual falls to near falls and acquire a more complete picture of the person's condition. Through a review of the pattern of near fall frequency and optionally other parameters such as magnitude and direction of the near falls that might precede a full fall, near fall detector <b>20</b>, in an embodiment of the invention, may be useful to alert a person and/or the person's physician that the person is at risk of falling. The person may then respond by wearing protective padding r other safety equipment, for example, or by taking other suitable precautions that prevent a fall from happening that would otherwise have occurred. Near fall data may also provide a quantitative measure that can be used to evaluate the effectiveness of therapeutic interventions.

p-0114Near fall detector <b>20</b> of the present invention, in some embodiments, can be configured to automatically record and/or report the number of instances of near falls, as well as details of each near fall such as one or more of the date and time at which it occurred, its magnitude, direction, its location and/or movements before or after the fall (e.g., indicating stair climbing, fast walking or other gait, task and/or physiological characteristics). This feature of automatic self-reporting represents an improvement in accuracy over self-reporting of near fall instances by the person. Self-reporting can be highly unreliable because it is subjective in nature, relying on the patient's memory and motivation, and/or lacks sensitivity, in that a patient might not recognize that an experience was in fact a near fall (particularly if its magnitude was low). Self-reporting also usually requires a long observation period, such as six months or a year. Optionally, the systems described herein, while usable for long periods, can be used for short periods, such as 1-10 hours, 1-10 days or 1-10 weeks, or intermediate periods.

p-0115As will be discussed in greater detail below, near fall and actual fall detection in some embodiments of the invention is measured based on acceleration of person <b>22</b>. Some embodiments are based on the inventive realization that whereas regular walking is a controlled form of movement that involves a consistent level of acceleration, when there is a fall there is a loss of control resulting in a much higher level of acceleration. Movement data obtained for near falls and other parameters can be used to construct a “gait acceleration profile” that is particularly configured to the movement or gait characteristics of the person. It is hypothesized, without being limited to any particular hypothesis, that one or more parameters of the person's gait acceleration profile constitute a useful measure or indicator of loss of control by person <b>22</b>. Alternatively, one or more parameters of the gait acceleration profile may be viewed as an indicator of over-control by person <b>22</b>, since in recovering from a near fall and avoiding a real fall, person <b>22</b> has made a successful attempt to regain control.

p-0116In some embodiments of the invention, detector <b>20</b> counts both near falls and actual falls. Data relating to both experiences comprise the gait acceleration profile of the person. The two types of events may be lumped together, or alternatively, upon further analysis of the data, instances of near falls may be separated from instances of actual falls. Optionally, falls are detected based on the sudden deceleration at the end of a fall, or based on the time of the fall and/or a time integral of velocity or acceleration which indicates vertical distance moved of the sensors.

p-0117In some exemplary embodiments of the invention, detector <b>20</b> is configured to detect gait irregularity. Optionally, detector <b>20</b> is configured to detect gait irregularity in addition to near fall detection. Optionally or alternatively, detector <b>20</b>, or a variation thereof, is configured to detect gait irregularity irrespective or instead of near fall.

p-0118In some embodiments, gait irregularity detection is based on vertical acceleration. Typically, corresponding to gait's steps, the acceleration signal exhibits a generally cyclic pattern with characteristic peaks. In some embodiments, irregularity is determined when the periods of the cycles (e.g. between peaks) vary above a threshold. Optionally or additionally, the irregularity is determined when the shape of the cycles vary above a threshold, where the variability of the shape is determined, for example, by cross-correlation. Optionally or additionally, the irregularity is determined by a frequency spread of the acceleration signal, obtained for example, with a Fourier transform.

p-0119In some embodiments, detector <b>20</b> is configured to assist in regulating a person's gait. Responsive to a detected gait irregularity of a person, the person is prompted by cuing signals, such as audio message or vibration, to adjust and/or stabilize the gait.

p-0120In some embodiments, detector <b>20</b> is configured to enhance a Timed Up and Go (TUG) test to assess the tendency of a person to fall. In some embodiments, the enhancement is based on time derivative of the vertical acceleration. In some embodiments, a tendency to falling is detected when the rate of change of the acceleration is above a threshold. In some embodiments, the threshold is based on the rate of change of acceleration of healthy person or persons. Optionally or additionally, the threshold is based on the physiological state of the person being assessed, such as neurological disorder.

2. Exemplary Structure

p-0121<figref idrefs="DRAWINGS">FIGS. 2A</figref>, <b>2</b>B, and <b>2</b>C shows the component elements of three exemplary embodiments of near fall detector <b>20</b>.

p-0122The embodiment of <figref idrefs="DRAWINGS">FIG. 2A</figref> is a self-contained device, in which all of the elements are contained in a common housing or casing <b>30</b>. As discussed in greater detail below, this embodiment includes features that provide real-time feedback to the user. Accordingly, this embodiment could be used as near fall detector <b>20</b> in the example of <figref idrefs="DRAWINGS">FIG. 1</figref>.

p-0123As shown in <figref idrefs="DRAWINGS">FIG. 2A</figref>, near fall detector <b>20</b> includes a sensor <b>32</b>. This component may be any sensor configured to measure an aspect of movement such as a change of acceleration, velocity, or position. An accelerometer, which is a type of sensor that measures acceleration directly relative to freefall, may be used for sensor <b>32</b> in some embodiments. Accelerometers are convenient to use because they are widely available and inexpensive relative to specialized acceleration measuring devices. In addition, as will be discussed in greater detail below, measuring acceleration directly provides the benefit of reducing the processing burden on the device, as compared with a sensor that measures position or velocity. The tolerance or sensitivity of sensor <b>32</b> should be about 800 mV/g or better. The sampling frequency of sensor <b>32</b> may be about 100 Hz, and optionally is not less than about 60 Hz in order to obtain adequate results. A sampling rate that is too low may adversely affect sensing quality.

p-0124A parameter of sensor <b>32</b> is the number of axes in space in which the sensor takes its measurements. Tri-axial sensors <b>32</b> are configured to measure in all three orthogonal orientations in space, specifically the vertical, medio-lateral, and anterior-posterior directions. A single axis sensor may measure along one axis only, such as in the vertical direction, and a bi-axial sensor measures in two directions. Sensor <b>32</b> of the present invention may be a tri-axial sensor in all embodiments, but may also be a bi-axial or single axis sensor in some embodiments, as long as one of the axes of measurement is the vertical axis. In some cases a bi-axial or single axis sensor may be less expensive than a tri-axial sensor. However, the use of tri-axial sensors may enhance detection accuracy and reliability, and may also provide the monitoring physician with additional information about the direction and nature of any near falls. An example of an accelerometer that may be used for sensor <b>32</b> is the Dynaport, manufactured by the McRoberts company of the Netherlands. If a single axis accelerometer is used, detector <b>20</b> optionally includes an indicator (e.g., an arrow) to show which part of detector should be aimed in a certain direction (e.g., up).

p-0125In this embodiment sensor <b>32</b> transmits the measured movement data to a processor <b>34</b>. As shown, the transmission is made through a sensor output port <b>31</b> on sensor <b>32</b>, which connects directly to a processor input port <b>33</b> of processor <b>34</b>.

p-0126Processor <b>34</b> may be a numeric processor, computer, or related electronic component such as an application specific integrated circuit (ASIC), electronic circuit, micro-controller, or microprocessor capable of processing the raw movement data measured by sensor <b>32</b>. Optionally, the speed of processing, such as a speed of a computation cycle of measurement or measurements of sensor <b>32</b>, is at least that of the sampling frequency of sensor <b>32</b>. In some embodiments processor <b>34</b> records acceleration values and calculates derivatives or other parameters of acceleration. Processor <b>34</b> further includes and/or is coupled with software (not shown) that directs operation of the processor. Internal memory (not shown) may optionally be included in and/or is coupled with processor <b>34</b> to store logged and derived acceleration values, and/or other numerical values calculated by the software. Alternatively or additionally, processor <b>34</b> may connect with a separate memory module <b>36</b> to store these values. In some embodiments, processor <b>34</b> is further configured to control some or all aspects of a user interface <b>38</b> and/or a radio transmitter or receiver or combined transmitter/receiver (“transceiver”) <b>40</b>. Connection with these elements may be made in some embodiments through a processor output port <b>42</b> and a user interface input port <b>43</b>.

p-0127Processor <b>34</b> may also connect with an external device such as a computer through an optional external interface port <b>44</b>. This connection may enable processor <b>34</b> to transfer data to the external computer and/or to receive a software program, software updates, or other inputs, for example, by a physical connection (e.g. wired) and/or wirelessly such as using a Bluetooth or a Wifi or cellular connection. In some embodiments, external port <b>44</b> may be a USB port or other industry standard connection. For additional flexibility, external port <b>44</b> may comprise two or more such ports rather than just one.

p-0128After person <b>22</b> has used the device for a given period of time, a record of the person's near fall and other gait related data is optionally stored in the device (optionally as it occurs). This processed data may be provided to the person's doctor by connecting device <b>20</b> through port <b>44</b> into a corresponding port, such as a USB port, of a computer. The data may then be transferred between devices in the manner well known in the art. In practice, person <b>22</b> may hand device <b>20</b> to the doctor or doctor's staff when visiting the doctor for an appointment, and the information may then be transferred to the doctor's computer directly. Alternatively, person <b>22</b> might transfer the information to his or her own computer and then email it to the doctor. Alternatively, the information might be sent wirelessly directly or indirectly from device <b>20</b> to the doctor's computer or another location, for example, by email. Another embodiment includes real-time transfer of the data as it is processed for online monitoring. In some embodiments, the memory is or comprises a removable card such as an SD card. Data on the card can be read by a card reader, and the data is optionally transferred to a computer and/or for archiving such as on hard disk or CD or DVD.

p-0129User interface <b>38</b> is an element of near fall detector <b>20</b> configured, in some embodiments, to provide information to the user or person <b>22</b> and/or to receive information from the user. The information may be in any convenient format such as visual, audio, and/or touch, and may be configured to meet the particular needs of the user. For example, in some embodiments user interface <b>38</b> may emphasize audio-based elements rather than visual elements, to better meet the needs of elderly users whose sight is weak.

p-0130User interface <b>38</b> may optionally include information output elements such as a visual display screen <b>46</b> capable of displaying alphanumeric and/or graphical messages, a speaker <b>48</b>, and/or alarm lights <b>50</b>. Optional user input elements include a touchscreen <b>52</b>, microphone <b>54</b>, keypad, and touchpad (not shown). In some embodiments, user interface <b>38</b> may include a camera and/or a video recorder.

p-0131In some embodiments, visual display screen <b>46</b> may also include the functionality of touchscreen <b>52</b>, and accordingly comprise a means for both displaying information to the user and receiving information from the user. Visual display screens <b>46</b> based on liquid crystal technology (LCD) may be used due to their readability and low power requirements, but other types of display and/or touchscreen technologies may also be used.

p-0132As noted, near fall detector <b>20</b> optionally includes wireless transceiver <b>40</b>. In a handheld device, transceiver <b>40</b> in some embodiments will operate at relatively high frequencies such as from about 100 MHz to 2 GHz, this may allow a device to be made smaller. Transceiver <b>40</b> optionally connects to processor <b>34</b> through processor output port <b>42</b>, and may include a transponder (not shown), antenna <b>41</b>, and other radio frequency components required to maintain wireless communication. In some embodiments transceiver <b>40</b> may comprise a radio and antenna such as that used in a cellular telephone or, in other embodiments, components of the type used in a computer standard Bluetooth interface.

p-0133In order to power the elements of near fall detector <b>20</b>, an energy source such as a battery <b>56</b> may be used. In some embodiments battery <b>56</b> is a light weight battery that provides power for an extended number of hours, or even several days or weeks. In this way, near fall detector may be used for the greater part of a day, and enable a meaningful amount of data to be gathered. In some embodiments battery <b>56</b> is a lithium ion battery, but other battery types, for example, rechargeable or one-time may be used as well.

p-0134The various optional elements of user interface <b>38</b>, along with transceiver <b>40</b>, may be combined to provide a range of responses that assist person <b>22</b> in the event of a near fall or a fall. For example, upon detecting a near fall or fall, speaker <b>48</b> could emit an audible beep and then deliver a message in the form of a human voice asking if the person is ok, and requesting person <b>22</b> to press a button on the device or screen for confirmation. Alternatively, the message could be a visual one on display screen <b>46</b>. If the user signals that he or she is ok no further action need be taken. If the user suggests otherwise or does not respond within a predetermined time, near fall detector <b>20</b> may be programmed to automatically send an email, page, or text message to a family member or doctor to alert them that person <b>22</b> fell or has almost fallen and needs assistance. An optional geographical position system (gps) in near fall detector <b>20</b> may automatically inform the doctor of the location of the person. In some embodiments, the device could automatically dial the doctor's phone number to enable direct voice communication.

p-0135In some embodiments, near fall detector <b>20</b> could be programmed to engage person <b>22</b> in a dialogue, to obtain more precise information. Person <b>22</b> could respond in a variety of ways, such as by keyboard, touchscreen, or by speaking into microphone <b>54</b>. Sample questions from such a dialogue may be, for example, “did you fall?”, “are you ok?”, “where are you?”, “do you need help?”, and “would you like to call your doctor/spouse?”. The device might also be used to record a voice or video message by person <b>22</b> and forward the message to the assisting party.

p-0136Housing or casing <b>30</b> is optionally sized and/or shaped sufficiently large to enclose the various components. Internal elements such as sensor <b>32</b> and processor <b>34</b> are optionally shielded from the elements, and/or user interface elements such as a keyboard, visual display screen <b>46</b>, if present, are optionally easy to access. Housing <b>30</b> is optionally made of a rigid and durable plastic, but other materials that are light and strong, such as aluminum, may also be used. Optionally, housing <b>30</b> includes a clip (not shown) for convenient attachment to belt <b>24</b> or other article of clothing. If sensor <b>32</b> requires a particular orientation when the device is mounted on belt <b>24</b> in order to operate effectively, visual or audio feedback may be provided by the appropriate elements of user interface <b>38</b> to assist person <b>22</b>.

p-0137Near fall detector <b>20</b> in some embodiments of the invention may comprise a dedicated device having as its only or primary function the detection of near falls and actual falls. In some embodiments, near fall detector <b>20</b> may be incorporated into other types of electronic devices used primarily for other purposes unrelated to fall detection. Examples of such devices include cellphones, pagers, portable media players, mobile Internet devices, and the like. This configuration may be more convenient for the user as it reduces the number of devices to be carried, and may also reduce the risk that the user will forget to take near fall detector <b>20</b>.

p-0138In some of these embodiments all or most of the hardware elements may already be available as part of the function of the device. For example, some cellphones known as “smartphones” and even some “regular” cellular telephones and PDAs include relatively powerful computer processors, accelerometers, visual display screens and speakers, wireless telephone and data communication hardware, and the like. Accordingly, some smartphones may only require the addition of specialized software to become configured as near fall detector <b>20</b>, according to some embodiments of the invention. In some instances the smartphones may need other modification such as the addition of memory module <b>36</b> and/or adding of a sensors, optionally with wired or wireless linking to the smartphone.

p-0139In some embodiments of the invention, near fall detector <b>20</b> may be incorporated into a medical device implanted in (or carried by) the user's body for medical purposes, such as a brain pacemaker for example. Other examples of such implanted devices include heart pacemakers, prosthetic hips, and implanted pumps for chronic pain. Similar to smartphones, some of these devices may already include a processor or accelerometer and accordingly may only require software to function as near fall detector <b>20</b>, according to some embodiments of the invention.

p-0140Turning now to <figref idrefs="DRAWINGS">FIG. 2B</figref>, in this embodiment processor <b>34</b> is separated from the portable part of device <b>20</b> contained in housing <b>30</b> and placed at a remote location. In an exemplary embodiment of the invention, remote processor <b>34</b> receives movement data from sensor <b>32</b> in real time (e.g. sufficiently fast to detect a near fall) through transceiver <b>40</b>, and communicates with and controls user interface <b>38</b> through wireless communication. Remote processor <b>34</b> otherwise functions similarly to integrated processor <b>34</b> of the embodiment of <figref idrefs="DRAWINGS">FIG. 2A</figref>, in that it processes and monitors near falls and communicates with person <b>22</b> and doctors or other assisting parties. Since this embodiment performs data analysis in real time, it could be used as near fall detector <b>20</b> in the example of <figref idrefs="DRAWINGS">FIG. 1</figref>.

p-0141In this embodiment, a local processor <b>35</b> may be included in housing <b>30</b>, for example, primarily to manage operation of the portable device <b>20</b>. Local processor <b>35</b> may accordingly be relatively less powerful than remote processor <b>34</b> (e.g. lower requires less power). In some embodiments local processor <b>35</b> may perform a portion of the data processing to ease the burden on remote processor <b>34</b> and/or reduce transmission volume e.g. to reduce power and/or required bandwidth. In this embodiment processor <b>34</b> may be stationary and placed at a fixed location within the range of transmission of mobile device <b>20</b>. Additionally, memory module <b>36</b> may also be remotely located and connected to processor <b>34</b>. Processor <b>34</b> in this embodiment is conveniently a general purpose computer such as a personal computer rather than an electronic component such as an ASIC or microprocessor, and memory module <b>36</b> may be the hard disk drive of computer <b>34</b>.

p-0142The distance at which mobile device <b>20</b> may travel from stationary remote processor <b>34</b> will vary depending on the type of wireless technology used by transceiver <b>40</b> and the power available in battery <b>56</b>. In some embodiments the wireless technology may be Bluetooth, which has a range of several meters. In some embodiments cellular telephone technology may be used, which has a much larger range, potentially in the kilometers. However, as the distance increases the potential for disruption in communication that would adversely affect real time feedback increases. Accordingly, this embodiment may be particularly suitable in a closed environment in which a multiple number of persons need to be monitored, such as a nursing home or a hospital. The aspect of multiple patients each having a mobile device <b>20</b> and sharing remote processor <b>34</b> is represented in <figref idrefs="DRAWINGS">FIG. 2B</figref> by multiple dashed rectangles <b>20</b>.

p-0143<figref idrefs="DRAWINGS">FIG. 2C</figref> shows another embodiment of near fall detector <b>20</b>. This embodiment is similar to the embodiment of <figref idrefs="DRAWINGS">FIG. 2B</figref> in that processor <b>34</b> is remote and housing <b>30</b> includes local processor <b>35</b>. However, in this embodiment there is no transceiver <b>40</b> or wireless communication between mobile device <b>20</b> and remote processor <b>34</b>, and memory module <b>36</b> is connected to local processor <b>35</b> inside mobile device <b>20</b>. In operation, mobile device <b>20</b> accumulates near fall data and stores the data in memory module <b>36</b> for later offline processing by remote processor <b>34</b>. The data may be transferred to remote processor <b>34</b> through external interface port <b>44</b> in the manner described previously. In this embodiment user interface <b>38</b> is optional. In some embodiments there is no user interface <b>38</b> other than in some embodiments, on/off switch. In other embodiments user interface <b>38</b> may be a single element such as display screen <b>46</b>, to guide the user in setting up the device. If this embodiment does not provide real time analysis and feedback, it is optionally not used as near fall detector <b>20</b> in the example of <figref idrefs="DRAWINGS">FIG. 1</figref>.

p-0144In some embodiments, detector <b>20</b> is configured to detect gait irregularity based acceleration measurement. In some embodiments, detector <b>20</b> is further configured to generate cuing signals responsive to detection of a gait irregularity. For example, using speaker <b>48</b> to sound messages such as ‘step . . . step . . . ’, and/or generate audible ‘ticks’ akin to a metronome, or any sound to indicate a regular pace. As another example, a vibrator is attached to the person arm and/or or optionally comprised in detector <b>20</b>, and vibrations are generated to indicate a regular pace. In some embodiments, other methods are used to indicate or prompt a regular pace, such as sending an audible prompt to a earphone or hearing aid by a Bluetooth connection or a wire connection.

p-0145Optionally or alternatively, detector <b>20</b> is configured to assist in detecting tendency to fall in during a Timed Up and Go (TUG). For example, detecting rate of change of acceleration during sitting or rising movements and determining if the person is prone to fall according to threshold criterion of the rate of change. In some embodiments, the determined tendency to fall (and/or lack thereof) is reported on display screen <b>46</b>. Optionally, rate of change and, optimally, the criterion that was used is reported on display screen <b>46</b>. In some embodiments, the rates of change and criterion used are stored in detector <b>20</b> for further study. In some embodiments, the rates of change and criterion used are transferred to other devices as described above.

p-0146In some embodiments, determination of gait irregularity and/or tendency to fall is based on the measurement or an accelerometer such as sensor <b>32</b>. Optionally or alternatively, additional or different accelerometers or sensors are used.

p-0147In some embodiments, configuring detector <b>20</b> is carried out by modifying the software program and/or electronic circuitry (e.g. re-programming an FPGA). Optionally, in case sensors other than senor <b>32</b> are used, the program and/or electronic circuitry are adapted to the other sensors. In some embodiments, in configuring detector <b>20</b>, processor <b>34</b> may be changed and/or an additional processor is incorporated in detector <b>20</b>.

p-0148Referring to detector <b>20</b> implies, without limiting, also variations thereof or similar devices that use one or more accelerometers.

3. Exemplary Operation

p-0149<figref idrefs="DRAWINGS">FIGS. 3A and 3B</figref> are flow charts that illustrate exemplary operation of near fall detector <b>20</b>, according to an embodiment of the invention. <figref idrefs="DRAWINGS">FIG. 3A</figref> provides a broad overview and <figref idrefs="DRAWINGS">FIG. 3B</figref> provides a more detailed view of the method of gait data collection of the invention. The modules shown in the flow charts represent or correspond to processes and methods that can be carried out in software and executed by processor <b>34</b>.

p-0150In these figures, the illustrated processes are based on an embodiment of near fall detector <b>20</b> that uses an accelerometer or other sensor <b>32</b> that measures acceleration directly or other movement parameters such as angular velocity or tilt. Embodiments of the invention that use sensors that measure different aspects of movement, such as velocity or position, may include extra steps that involve taking derivatives of velocity and/or position in order to obtain an estimate of acceleration and/or may measure movement parameters other than acceleration. It may be advantageous to use an acceleration based sensor <b>32</b> in some embodiments, since it is more accurate and enables processing with fewer steps and accordingly provides a faster overall processing time.

p-0151In exemplary embodiments of the invention, beginning with <figref idrefs="DRAWINGS">FIG. 3A</figref>, upon starting and calibrating the device, sensor <b>32</b> begins to measure acceleration for the current gait segment of time T_n(module <b>110</b>). The gait segment T_nis simply the inverse of the sampling frequency, e.g. 0.01 seconds for a sampling frequency of 100 Hz. Acceleration is measured in the axes for which sensor <b>32</b> is configured, i.e. vertical, medio-lateral, and anterior-posterior for a tri-axial sensor.

p-0152In some embodiments, the raw acceleration data is then passed to processor <b>34</b> (module <b>120</b>), through sensor output port <b>31</b> and processor input port <b>33</b>. Processor <b>34</b> performs one or more calculations to obtain certain parameters that are used to obtain a gait acceleration profile of person <b>22</b>. These parameters may be called “derived parameters” since they are derived from the raw movement data provided by sensor <b>32</b>. Processor <b>34</b> optionally calculates a dynamic threshold for each derived parameter. The threshold is optionally “dynamic” because it is based on and updated from the stream of acceleration values received for each period T_n.

p-0153Upon calculation of these values, processor <b>34</b> optionally determines whether a near fall has occurred (module <b>130</b>). In making this decision, processor <b>34</b> compares each derived parameter with an associated threshold value. The threshold value may be the dynamic threshold calculated earlier, or a predetermined “static” threshold. A near fall is indicated if a particular parameter exceeds its threshold. In addition to comparing individual derived parameters with their threshold, processor <b>34</b> may optionally also combine any two or more individual parameter results using logical operators such as OR and AND.

p-0154In some embodiments, upon completing a plurality of comparisons, processor <b>34</b> will make an overall determination of whether a near fall has occurred. If every comparison indicates a near fall (or optionally a subset such as a majority the number of comparisons indicate a near fall), then the determination of decision module <b>130</b> will be “Yes”, a near fall has occurred. If none of the comparisons indicate a near fall, the determination will be “No”, a near fall has not occurred. In most cases the results lie somewhere in between, with some comparisons indicating a near fall and some indicating no near fall. Processor <b>34</b>, in some embodiments of the invention, may be programmed to assign a likelihood of a near fall according to a predetermined sensitivity set by the doctor in accordance with the particular medical profile and fall risk of the patient. For example, the physician may set near fall detector <b>20</b> to determine that a near fall has occurred if half or more of the comparisons indicate a near fall, and to determine no near fall otherwise.

p-0155As indicated in the flow chart of <figref idrefs="DRAWINGS">FIG. 3A</figref>, if it is determined that a near fall has not occurred the system moves on to the next time period T_n, for n=n+1 (module <b>140</b>), and the process is repeated with a new sensor measurement (module <b>110</b>). However, if it is determined in module <b>130</b> that a near fall has occurred, near fall detector <b>20</b> may then respond in some manner (module <b>160</b>). As described earlier, the response could, for example, take the form of any one or combination of logging a record of the near fall event, prompting the user by display or audio, querying the user to obtain more information, and/or communicating with another party for assistance.

p-0156Decision module <b>170</b> asks whether near fall monitoring should continue. This will depend on the seriousness of the near fall. If the near fall was a relatively minor event that did not overly stress the user then control passes to module <b>140</b>, “n” is incremented, and the process repeats at module <b>110</b>. Otherwise near fall monitoring may stop (module <b>180</b>) as the user recovers from the effects of the near fall or fall. Optionally, the stop is for a limited period of time and/or until a rest is performed.

p-0157Turning now to the flow chart of <figref idrefs="DRAWINGS">FIG. 3B</figref>, the processes performed by near fall detector <b>20</b> may now be reviewed in greater detail. Again, upon startup and calibration (module <b>100</b>), sensor <b>32</b> measures acceleration for the current gait segment of time T_n(module <b>110</b>). In module <b>120</b>, as noted, processor <b>34</b> calculates derived parameters of acceleration (and/or other movement signals).

p-0158The derived parameters in some embodiments may include, for example, any one or combination of the following six example parameter types:

p-01591) “Max” is the maximum measured acceleration value. For example, a measurement of acceleration along the vertical (“y”) axis that is the maximum such value for a period of time may be referred to as “Vertical Max”.

p-01602) “Maxp2p” is the maximum peak-to-peak value (positive peak to negative peak within a single cycle) of the measured acceleration over a period of time.

p-01613) “SVM” is the signal vector magnitude. This is calculated as the square root of the sum of the squares of the measured acceleration, for each axis measured. For example, using a tri-axial sensor <b>32</b>, SVM is the square root of the sum of (x²+y²+z²), where x, y, and z are the measured acceleration values in the medio-lateral (“x”), vertical (“y”), and anterior-posterior (“z”) directions.

p-01624) “SMA” is the normalized signal magnitude area. This is calculated as the sum of the absolute values of the acceleration along each measured axis, integrated over time “t”. The sum is divided by “t” to obtain the normalized value.

p-01635) “Maxdiff” is the maximum acceleration derivative. This is obtained by taking the derivative of the measured acceleration (sometimes called the “jolt”), and is the maximum of this value.

p-01646) “Maxp2pdiff” is the maximum peak-to-peak acceleration derivative. Like Maxdiff this is also based on the acceleration derivative or jolt rather than the raw acceleration value. This parameter is the maximum value between positive peak and negative peak of the acceleration derivative within a single cycle over a period of time.

p-0165The inventors have observed that use of the above six parameters, and even a small subset of the six including as few as one or two parameters, have provided adequate results in some embodiments. In some embodiments, additional derived parameters other than the six described above may also be calculated by processor <b>34</b> and used to determine near falls, optionally in a more robust manner.

p-0166In some embodiments, the “Vertical Max” parameter is included, solely and/or in combination with other parameters, in the determination of a near fall.

p-0167Returning to the flow chart of <figref idrefs="DRAWINGS">FIG. 3B</figref>, in module <b>120</b> processor <b>34</b> calculates or updates an incremental value for a particular derived parameter. In module <b>122</b>, processor <b>34</b> updates a dynamic threshold value for this parameter. In module <b>124</b> the system queries whether there are any other derived parameters to be calculated. If the answer is “Yes” control is returned to module <b>120</b> and the process repeats. Accordingly, if for example the system is programmed to use three derived parameters, then modules <b>120</b> to <b>124</b> will loop three times before proceeding to module <b>130</b>. Alternatively, a flow process in which processor <b>34</b> calculates all of the derived parameters first, and then calculates all of the associated thresholds is also comprehended by the present invention. In embodiments that use a static threshold instead of a dynamic threshold, module <b>122</b> may be bypassed or its results ignored. In embodiments that use only one derived parameter, decision module <b>124</b> may be bypassed.

p-0168The calculation of dynamic threshold for each derived parameter in module <b>122</b> may be performed in a variety of ways. In some embodiments, a mean and standard deviation of the parameter may be calculated and updated with each successive measurement. The threshold may then comprise the mean value plus some multiple of the standard deviation. For example, a “usual-walk” period of time may be identified, based perhaps on measures of rhythmicity and regularity, and one or more derived parameters and their mean and standard deviations estimated based on this usual-walk episode. If in any subsequent window of time the value of one of these derived parameters exceeds the mean plus three times the standard deviation of that observed during the usual-walk, the algorithm will record this parameter as detecting a near fall. For other activities, such as stair climbing (e.g., similarly identified from the gait signals, or based on displacement as a function of time), other thresholds may be applied. Optionally, a user can indicate, for example, during a calibration stage, if a recent event was a near fall or not. This may be, for example, initiated by the user, or by the system asking regarding a specific event.

p-0169Unlike the dynamic threshold, the calculation of the static threshold is optionally performed offline, at some time prior to operation of the near fall detector <b>20</b>. Parameter data may be obtained for a time period in which a person's walk is directly observed (or recorded for later observation). From this, two groups of time periods or intervals may be defined, comprising “near fall” groups and “non-near fall” groups. Since the near fall groups have been directly observed and are known to be accurate, they comprise a “gold standard” of known near falls that may be correlated with the signal processing data.

p-0170In some embodiments, the static threshold may be calculated as an optimization of sensitivity and specificity with respect to a single or multiple number of derived parameters. The algorithms used may be non-linear and advanced. Some examples of the types of discriminant functions that may be employed include linear, diaglinear, quadratic, diagquadratic, and mahalanobis. Algorithm performance may then be measured in terms of sensitivity (true positive/(true positive+false negative)) and specificity (true negative/(true negative+false positive)).

p-0171In decision module <b>130</b> processor <b>34</b> determines whether a near fall has occurred in time period “n” based on the updated derived parameter values. As noted above, the determination may be made by subtracting (or comparing in another way) from the derived parameter value the value of its associated threshold. In embodiments that use dynamic thresholds, the threshold values calculated in module <b>122</b> are used. In embodiments that use static thresholds, the threshold values will have been pre-loaded in memory and may be retrieved at the time of the calculation. Also as noted, in some embodiments a plurality of such comparisons are made involving individual parameters and combinations of parameters.

p-0172The inventors have discovered that, in some embodiments, adequate detection of near falls may be obtained through the calculation of a single derived parameter, Maxp2pdiff, based on acceleration along the vertical axis. The inventors observed that vertical Maxp2pdiff identified near falls with a sensitivity of 85.7% and a specificity of 88.0%. It may be noted that in this case, decision module <b>130</b> would only need to review a single comparison of Maxp2pdiff with its associated threshold, as no other comparisons need to be considered.

p-0173The inventors have also discovered that, in some embodiments, adequate detection of near falls may be obtained through the calculation of two derived parameters, Maxp2pdiff and Max, both based on acceleration along the vertical axis, and by performing a logical AND operation on the individual results. Accordingly, this method will find a near fall only in the event that both parameters exceed their respective thresholds. The inventors observed that this method of detection identified near falls with a sensitivity of 85.7% and a specificity of 90.1%.

p-0174An illustration of the results using the above methods of detection is shown in <figref idrefs="DRAWINGS">FIG. 4</figref>. As indicated, in the time period recorded in the graphs, person <b>22</b> had three near falls or missteps. In <figref idrefs="DRAWINGS">FIG. 4</figref>, the lower graph shows Maxp2pdiff and the upper graph shows Vertical Max over this time period. It may be seen that at or about the time of each misstep, both derived parameters display distinct increases in value relative to their average values over the balance of the time period. Accordingly, a gait acceleration profile based on the derived parameter Maxp2pdiff, or one based on the logical combination of Maxp2pdiff “AND” Vertical Max, may be used to detect near falls with adequate results.

p-0175As noted, the present invention comprehends many other selections of specific derived parameters and combinations of derived parameters to determine near falls. In another example, all six derived parameter examples may be calculated, and near falls could be determined if any three or more confirm a near fall. Through logical combinations of individual parameters many more comparisons may be made and considered in determining a near fall to enhance robustness of the decision tree. The various comparisons could be listed in a hierarchy and determination of a near fall could be made along a gradient that corresponds with the results of the plurality of comparisons. The output could be binary (i.e. “yes/no” a near fall has likely occurred) and/or a continuous measure related to the likelihood that a near fall has occurred, based on the number of parameters exceeding thresholds. For example, an embodiment may have <b>100</b> comparisons involving the different parameters individually and in various logical combinations. The comparisons could represent <b>100</b> “levels” over which the near fall is graded, ranging from a sure near fall at one end to a sure non-near fall at the other end. Similarly, output scores could be graded based on the percent of steps in which a misstep occurred.

p-0176Optionally (e.g., as discussed above), the determination of a person's near fall experience may be used to prepare or modify that person's gait acceleration profile. Near fall detector <b>20</b> may also be used in some embodiments to determine other aspects of a person's gait that enhance the gait acceleration profile. For example, if a near fall has occurred, it is useful to know the magnitude and direction of the near fall. It is also useful to know (e.g. detect or note) if the incident has resulted in an actual fall. Other useful gait parameters arise from a study of the person's walking motion, and include, for example one or more of, step width, step or stride regularity, and symmetry between steps.

p-0177The calculation of gait parameters that arise from the near fall may be seen in flow chart of <figref idrefs="DRAWINGS">FIG. 3B</figref> in the series of modules that follow a “Yes” determination of module <b>130</b>.

p-0178In module <b>142</b>, the magnitude of the near fall is optionally determined. Magnitude may be obtained from the peak of the acceleration, i.e. the derived parameter Max. Alternatively, in some embodiments that calculate the derived parameters SVM and/or SMA, these parameters may be used individually or in combination to obtain a better quality of the magnitude of the near fall. A magnitude value derived from SVM and/or SMA is considered to be more robust and stronger than a value derived from acceleration data alone. The magnitude value may be converted and presented on a number on a scale, for example between 1 and 100. In reviewing a person's gait acceleration profile, it is helpful to know that the person's near falls had an average magnitude of 70, for example, as opposed to an average magnitude of 20.

p-0179In module <b>144</b>, the direction of the near fall is optionally determined. This parameter can enhance the ability to extract meaning, at least as an estimation, and interpret the gait acceleration profile by providing the direction of a near fall relative to vectors along the vertical, medio-lateral, and anterior-posterior axes. It may be noted that in order to obtain directional near fall information sensor <b>32</b> is optionally configured to obtain measurements along all three axes.

p-0180The directional information provided by this parameter may be useful in aiding diagnosis by a physician. For example, falls that occur to the side are more likely to result in a broken hip, which are particularly troublesome and dangerous to elderly persons. Accordingly, the awareness of such data may trigger preventive action that could prevent a disabling fall that might otherwise occur. In another example, a persistent trend to near falls in a particular direction might indicate a structural weakness or postural problem, which may lead to preventive physiotherapy, adoption of a walking aid, or wearing asymmetrical protection such as a pad on one hip.

p-0181Optional modules <b>146</b> and <b>148</b> optionally provide information that assists in determining if a real fall has occurred. After a real fall, there is often a silent period since the person is not moving. Accordingly, module <b>146</b> collects sensor information for a period T_xafter the near fall. If the measured values are zero or close to zero (or reflect a vertical location that is near the floor and/or a small range of motion (e.g., by integrating acceleration over time)), it would suggest that a real fall has occurred. Module <b>148</b> estimates the height or position of the person's center of mass after the near fall. The center of mass may be estimated from a gyroscope, if that instrument is provided in near fall detector <b>20</b>. In some embodiments, an accelerometer such as that used for sensor <b>32</b> may be used to estimate the height of the center of mass.

p-0182Decision module <b>150</b> optionally considers the above information in determining whether a real fall has occurred. This module may also consider the magnitude value obtained in module <b>142</b>, since in a real fall the magnitude value tends to be higher than for a near fall. If a real fall is determined, module <b>152</b> logs data relating to the incident, such as the time and day, magnitude, and direction. In module <b>160</b>, near fall detector responds in the manner described earlier, by interacting with the person and possibly contacting an outside party. If module <b>150</b> determines that a real fall has not occurred, the event is logged as a near fall (module <b>154</b>). An optional module <b>156</b> may consider the parameters of the near fall in deciding whether to respond (module <b>160</b>), or whether to proceed to module <b>140</b> to increment “n” and repeat the sequence at module <b>110</b>.

p-0183Returning to decision module <b>130</b>, if processor <b>34</b> determines that a near fall has not occurred, gait parameters such as step width, step or stride regularity, and symmetry between steps may optionally be determined. These parameters can provide additional information about the patient's balance and gait that can not be obtained simply by observational analysis or self-report. These parameters are also independent of one another, and accordingly provide complementary, objective data that enhances the quality of the patient's gait acceleration profile.

p-0184Beginning with optional module <b>132</b>, the step width parameter may be determined as the distance in the horizontal or medio-lateral direction between the subject's feet, orthogonal to the direction of movement. It may be noted that to calculate step width sensor <b>32</b> is optionally configured to measure along the medio-lateral axis. It may also be noted that step width is a distance value. Accordingly, if sensor <b>32</b> is an accelerometer that measures acceleration directly, the measured value would generally have to be further processed, such as by double integration, to obtain an estimate of the step width distance.

p-0185The step width parameter may be useful for the gait acceleration profile of a patient in that if it is found to be wide, it may be an indication that the patient is over compensating. A step width that is not consistent and is too variable is considered to be unhealthy, and accordingly may prompt further diagnostic testing by the doctor.

p-0186Optional module <b>134</b> may be used to calculate step or stride regularity. This parameter is a measure of the repeatability, regularity, or consistency of the person's gait, and can refer to the length or the timing of the step. Useful information may be obtained along a single axis or from all three axes. This parameter is typically calculated by an autocorrelation of the raw acceleration data.

p-0187A stride of walking is the time to complete one walking cycle, for example from the left foot touching the ground to the subsequent instance of the left foot touching the ground. One stride equals two steps. Accordingly the terms “step regularity” and “stride regularity” mean essentially the same thing, with the only difference being the portion of the gait cycle over which they are measured.

p-0188Measures of regularity can be used to define the degree to which the person's walking pattern is rhythmic. In medical terms, the greater the regularity and “rhythmicity”, the healthier the motor control system is considered to be in the patient.

p-0189Optional module <b>136</b> may be used to calculate symmetry between steps. This parameter measures the degree of equality between steps taken by the left foot relative to steps taken with the right foot. It may be calculated by the formula:

p-0190<br />Gait Asymmetry=<b>100</b>×|1<i>n</i>(SSWT/LSWT)|.

p-0191In the formula, “SSWT” and “LSWT” stand for the mean values of the Short and Long Swing Time, respectively, as determined from the vertical axis.

p-0192Other measures of asymmetry, such as one based on step times, for example, may also be used in some embodiments to provide a more complete estimate of asymmetry patterns.

p-0193For example, identifying cycles periods in accelerometer signal or signals (e.g. peak to peak) and determining the variability (the irregularity) of the cycles' periods.

p-0194In some embodiments, an asymmetry measure such as difference between the longest and shortest cycles may be used. Optionally or alternatively, the standard deviation of the cycles' periods may be used. Optionally or alternatively, some other statistics such as the median of the period may be used.

p-0195In some embodiments, a measure of regularity of asymmetry may be obtained in a frequency domain, optionally within locomotion band (stride) such as 0.5-3.0 Hz. A narrow frequency spread (e.g. standard deviation) indicates regular stride and, vice versa, wide spread indicates irregularity and possibly a sign of physiological or neurological disorder.

p-0196<figref idrefs="DRAWINGS">FIG. 5</figref> shows exemplary charts of stride acceleration and frequency spread of a healthy person and a person with Parkinson disease, respectively.

p-0197Charts <b>501</b> and <b>503</b> are of a healthy and Parkinson diseased persons, respectively, illustrating the acceleration in the anterior-posterior direction, and charts <b>502</b> and <b>504</b> illustrate the respective frequency range. Vertical axis of charts <b>501</b> and <b>503</b> is acceleration (in g) and the horizontal axis is in seconds; horizontal axis of charts <b>502</b> and <b>504</b> is in Hertz and the vertical axis is the frequency amplitude.

p-0198The sharper and narrower peak of chart <b>502</b> with respect to chart <b>504</b> reflects a more consistent gait pattern, i.e., reduced gait variability and lower stride-to-stride fluctuations of a healthy person relative to a Parkinson diseased person.

p-0199In some embodiments, a measure of stride regularity or asymmetry is determined by combining (e.g. averaging) two or more of the methods described above. Optionally, the combination assigns different weights to the various measures obtained by the methods described above. In some embodiments, measures that indicate larger asymmetry are assigned larger weights relative to measures that indicate smaller asymmetry.

p-0200Upon completion of the calculation of the various gait parameters, module <b>140</b> increments “n” and the process is repeated with a new sensor measurement for time period T_nin module <b>110</b>.

p-0201A further aspect of operation of some embodiments of near fall detector <b>20</b> concerns calibration of the device. Calibration initializes the device so that the sensor recognizes and accurately responds to movement along the appropriate axes. In this way, near falls and other gait parameters can more accurately be measured. Calibration is helped by measuring along all three axes, as this enables the device to find the direction of gravity and to orient itself to align with it.

p-0202In an exemplary embodiment of the invention, calibration involves performing procedures recommended or instructed by the sensor or accelerometer manufacturer. In some embodiments of the invention, such as for example where near fall detector <b>20</b> is a dedicated device worn on the person's belt, the orientation of the device in space is relatively fixed. Accordingly, calibration in these cases may be a relatively simple matter. In other embodiments of the invention, such as when near fall detector <b>20</b> is incorporated in another device such as a cell phone, the orientation of the device in space is not fixed and will vary widely in the course of daily use. For example, a cell phone may be vertical when in use by a standing person, but may be horizontal if the person is lying down. Further, when put in a coat pocket or carrying bag the cell phone may be upside down or adopt any other orientation at random. In these cases the device may self-calibrate to ensure that near fall detector <b>20</b> works properly.

p-0203In some embodiments of the invention, calibration and operation of the device may be independent of the weight of the person whose movement is being monitored. For example, near fall detector <b>20</b> will be calibrated and operate in the same manner whether the user is a heavier person or a lighter person.

4. Exemplary Applications of Gait Acceleration Profile

p-0204As discussed, the gait acceleration profile of a person comprises that person's observed or recorded gait parameters over one or more periods of time. For example, a sample gait acceleration profile of a particular person might be: patient experienced three near falls over a two day period. The near falls had magnitudes of 60, 23, and 47 (arbitrary units) and were primarily in the medio-lateral/left direction. During this period, step width was 0.31 meters, stride variability (inversely related to regularity) was 6%, and gait asymmetry was 17. After an intervention consisting of physiotherapy and prescribed medication, in an evaluation over a similar two day period, near falls dropped to one with a magnitude of 14. Step width narrowed to 0.26 meters and gait asymmetry also improved by a reduction to a value of 11.

p-0205Some embodiments of the invention may enable the benefits of a detailed patient gait acceleration profile to become available at greater convenience to both doctors and their patients. An example of this may be in the area of remote exercise monitoring. There is a growing push in the medical field for at-home interventions to improve mobility. A doctor may encourage an older adult or patient with Parkinson's disease to walk for thirty minutes, five times a week, with three sessions outside and two sessions indoors on a treadmill, the latter perhaps having more complex instructions. Near fall detector <b>20</b> in some embodiments may be used for real-time monitoring as the patient carries out the prescribed exercises. If a near fall occurs, an alarm can sound or assistance provided immediately. In this way the safety and usability of such “tele-rehabilitation” approaches are improved, while at the same time enabling patient progress to be closely and precisely monitored. Alternatively, the near fall detector can be used to assess the efficacy of the prescribed therapy.

p-0206In some embodiments, detector <b>20</b>, or a variation thereof, may be used or adapted (e.g. by software and/or circuitry modification) to enhance common screening of subject prone to falls or to near-falls, as described below.

p-0207The Timed Up and Go (TUG) test is a widely used clinical test of fall risk. Subjects are asked to start in a seated position, stand up and walk 3 meters, turn around, and return to the seated position. In older adults and other populations such as patients with Parkinson's disease (PD) or stroke, longer TUG times have been associated with impaired mobility and an increased fall risk (for example, Balash Y, Peretz C, Leibovich G et al. Falls in outpatients with Parkinson's disease: frequency, impact and identifying factors. <i>J Neurol </i>2005;252:1310-1315; Najafi B, Aminian K, Loew F et al. Measurement of stand-sit and sit-stand transitions using a miniature gyroscope and its application in fall risk evaluation in the elderly. <i>IEEE Trans Biomed Eng </i>2002;49:843-851; Podsiadlo D, Richardson S. The timed “Up &amp; Go”: a test of basic functional mobility for frail elderly persons. <i>J Am Geriatr Soc </i>1991;39:142-148).

p-0208However, the TUG does not always successfully identify those with a high fall risk, especially among relatively well-functioning, healthy older adults (for example, Buatois S, Gueguen R, Gauchard GC et al. Posturography and risk of recurrent falls in healthy non-institutionalized persons aged over 65. <i>Gerontology </i>2006;52:345-352; Marschollek M, Nemitz G, Gietzelt M et al. Predicting in-patient falls in a geriatric clinic: a clinical study combining assessment data and simple sensory gait measurements. <i>Z Gerontol Geriatr </i>2009;42:317-321).

p-0209It was observed by the inventors, at least in representative cases, that extracted accelerometer-based measures such as by device <b>20</b> or similar ones can distinguish or be adapted to distinguish (e.g. by software adaptation) between elderly fallers and elderly non-fallers when they perform the TUG, even if TUG duration times are not significantly different in the two groups. It was observed that the rate of change of the acceleration during sitting movement from standing position and during movement of rising from a seated position is different between healthy persons and fallers (persons prone to fall, having a tendency to fall). Healthy persons exhibit a significantly larger rate of change of the acceleration relative to fallers, at least as observed for elderly subjects.

p-0210<figref idrefs="DRAWINGS">FIG. 6</figref> shows exemplary chart <b>601</b> of Timed Up and Go (TUG) of healthy person and <b>602</b> of a person prone to falling ('faller'). Charts <b>601</b> and <b>602</b> illustrate anterior-posterior accelerations measured with an accelerometer, where the horizontal axis is in seconds and the vertical axis is in −g. The acceleration signals of charts <b>601</b> and <b>602</b> are generally divided, respectively, into three zones, namely, <b>604</b> and <b>614</b> are when the persons sit from a standing position, <b>610</b> and <b>612</b> are walking periods, and <b>606</b> and <b>616</b> are when the persons stand from a seated position.

p-0211In the regions of up or down movements <b>604</b>, <b>606</b>, <b>614</b> and <b>616</b> the rate of change of the acceleration was determined as a time derivative of the measured accelerations (in g/sec), indicated in <figref idrefs="DRAWINGS">FIG. 6</figref> as jerk'.

p-0212As illustrated in <figref idrefs="DRAWINGS">FIG. 6</figref>, the rate of change of acceleration of the jerks of the healthy person and the faller person are considerably different. The rate of the acceleration of the healthy person is higher than that of the faller person. For example, as illustrated, the upward jerk of the healthy person is about 2 g/sec and the downward jerk (at <b>606</b>) is about 1 g/sec, wherein the respective jerks of the faller are about 0.5 g/sec, (at <b>614</b> and <b>616</b>, respectively).

p-0213As detector <b>20</b> comprises accelerator and measures acceleration and rate of change of acceleration, in some embodiments detector <b>20</b> is modified or adapted to distinguish (screen) fallers from non-fallers based on the amount of the rate of change of the acceleration in the jerks zones. Thus, in some embodiments, detector <b>20</b> can augment the TUG test by providing indication for differentiation between healthy persons and persons prone to fall, at least in some cases.

p-0214In some embodiments, modifying of adapting detector <b>20</b> comprises modifying the software program and/or circuitry of the detector (e.g. different gate array layout). In some embodiments, the modified or adapted detector <b>20</b> provides control (e.g. by touchscreen or button) to indicate when to measure the jerks. Optionally or alternatively, the program is adapted to recognize jerk zones according relative long generally monotonic acceleration with respect to walking.

p-0215In some embodiments, detector <b>20</b> is capable to determine gait irregularity and asymmetry, as described above. As such, further to a diagnostic tool, in some embodiments detector <b>20</b> can be used as a therapeutic or an assisting device for regulating the gait of a subject having a neurological disease or another subject having a tendency to fall.

p-0216For example, with ongoing assessment of the pattern and regularity of a gait of a subject, a signal could be automatically generated responsive detection of deviation from sufficiently regular or expected gait pattern.

p-0217In some embodiments, the signal indicates that the gait is irregular or that the subject is about to fall (near fall), prompting the subject to recover a proper gait. Optionally or additionally, the signal indicates suggested gait pace that the subject can follow in order to stabilize the gait (cueing signals).

p-0218In some embodiments, the signal indicates suggested pace irrespective of irregularities, providing continuous training to the subject, at least for certain time periods. Optionally the training may, in some cases at least, enhance functional mobility of the subject.

p-0219In some embodiments, upon detection of a near fall situation or irregular pace, detector <b>20</b> generates an alarm message such as by speaker <b>48</b>, notifying the subject of the situation.

p-0220In some embodiments, upon detection of irregular pace, detector <b>20</b> generates audible messages guiding the subject pace, such as ‘step . . . step . . . ’, thereby assisting the subject to regulate and stabilize the gait. In some embodiments, the guided pace is within a determined variability, avoiding too ‘mechanical’ gait. In some embodiments, the guided pace is adapted and/or synchronized with the subject's pace.

p-0221In some embodiments, the pace of the cueing signals are based on behavior detected or assessed in healthy persons, optionally of about the same age. Optionally or additionally, the pace of the cuing signals are based on intervals where the subject's gait is determined to be regular, at least to some extent.

p-0222In some embodiments, one or more other signals are generated in addition or instead the audible messages described above. For example, rhythmic auditory stimulation by tone such as or similar to a metronome, or rhythmic visual stimulation by one or more of alarm lights <b>50</b> or indications on display <b>46</b>.

p-0223In some embodiments, detector <b>20</b> is augmented to comprise a vibrator (e.g. akin to some pagers or cellular phones) and vibrations are generated to indicate the gait situation or provide pace guiding signals.

p-0224The amount of gait acceleration information that may be made available for analysis may be greatly increased due to the convenience provided by near fall detector <b>20</b> in this application. This in turn may lead to improvements in patient cognitive and motor functioning, particularly since available data suggest that interventions are more effective when they take place over longer time periods, are individually tailored, and include exercise in the home environment.

p-0225Near fall detector <b>20</b>, in some embodiments of the invention, may even be incorporated into treadmills or other exercise equipment, or provided as an add-on accessory. The device could be in the form of a “smart-box” that contains the software, processor <b>34</b>, communication hardware, and other elements. When using this type of exercise equipment, the user could indicate that he or she is doing a special activity for monitoring for near falls. In some embodiments the device may adjust the parameter threshold values to account for planned variations in exercise stimulation, such as increases in treadmill speed designed to challenge the patient.

p-0226The information provided by the gait acceleration profile may also provide insight into a person's neurological state related to the diagnosis of other types of medical conditions besides the predilection to fall.

p-0227It is hypothesized that a gait profile based on acceleration and other measures of movement (e.g., gyroscopes, tilt sensors) that includes such information as near falls, step and stride regularity, and symmetry may be tracked as part of a patient's medical record, and used as a tool for therapeutic use.

p-0228For example, in many cases prior to falling, there is an instant or moment in time when the person's brain fails to operate properly. In most cases this aspect of the person's medical condition may not be detectable until the symptoms become more pronounced and the underlying disease becomes more severe. However, in some cases the reduced brain activity may be observable indirectly, through the person's motor output or gait. By monitoring gait with near fall detector <b>20</b> of the present invention, the person's quality of movement may provide an early warning indicator of the onset of Parkinson's disease, for example, or other movement disorders.

p-0229In another example, a physician may have an array of possible treatments available for a patient diagnosed with a particular illness. One of the possible treatments may be a drug that is known to be effective with some patients but not with others, but for which there is no methodology to discern beforehand whether a particular patient will benefit. Upon further research using the gait profile, it may be found that the gait profile provides the missing neurological information to assist the physician in determining whether the drug will be effective in that case. Used in this way, the gait profile may lead to better and more cost effective medical care. Further, the efficacy of treatment may be verified by continuing to monitor the gait profile, and by analyzing subsequent near fall data to confirm that the number of instances of near falls and/or their magnitude has declined.

p-0230As used herein the term “about” refers to ±10%.

p-0231The terms “comprises”, “comprising”, “includes”, “including”, “having” and their conjugates mean “including but not limited to”.

p-0232As used herein, the singular form “a”, “an” and “the” include plural references unless the context clearly dictates otherwise.

p-0233Throughout this application, various embodiments of this invention may be presented in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the invention. Accordingly, the description of a range should be considered to have specifically disclosed all the possible subranges as well as individual numerical values within that range. For example, description of a range such as from 1 to 6 should be considered to have specifically disclosed subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6 etc., as well as individual numbers within that range, for example, 1, 2, 3, 4, 5, and 6. This applies regardless of the breadth of the range.

p-0234Whenever a numerical range is indicated herein, it is meant to include any cited numeral (fractional or integral) within the indicated range. The phrases “ranging/ranges between” a first indicate number and a second indicate number and “ranging/ranges from” a first indicate number “to” a second indicate number are used herein interchangeably and are meant to include the first and second indicated numbers and all the fractional and integral numerals therebetween.

p-0235It is appreciated that certain features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable subcombination or as suitable in any other described embodiment of the invention. Certain features described in the context of various embodiments are not to be considered essential features of those embodiments, unless the embodiment is inoperative without those elements.

p-0236Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims.

p-0237All publications, patents and patent applications mentioned in this specification are herein incorporated in their entirety by reference into the specification, to the same extent as if each individual publication, patent or patent application was specifically and individually indicated to be incorporated herein by reference. In addition, citation or identification of any reference in this application shall not be construed as an admission that such reference is available as prior art to the present invention. To the extent that section headings are used, they should not be construed as necessarily limiting.