Fall detection and fall risk detection systems and methods

Abstract

The present invention relates to a light-weight, small and portable ambulatory sensor for measuring and monitoring a person's physical activity. Based on these measurements and computations, the invented system quantifies the subject's physical activity, quantifies the subject's gait, determines his or her risk of falling, and automatically detects falls. The invention combines the features of portability, high autonomy, and real-time computational capacity. High autonomy is achieved by using only accelerometers, which have low power consumption rates as compared with gyroscope-based systems. Accelerometer measurements, however, contain significant amounts of noise, which must be removed before further analysis. The invention therefore uses novel time-frequency filters to denoise the measurements, and in conjunction with biomechanical models of human movement, perform the requisite computations, which may also be done in real time.

Claims

1. A fall risk assessment system comprising: a data processing system comprising one or more processor circuits configured to process data generated by a sensor, the data including information representative of at least one signal generated by the sensor in response to movement of an upper part of a body of a person, the data processing system programmed to at least; process said data to identify one or more peaks in the at least one signal by comparing values of the at least one signal with one or more predefined fall thresholds; and for at least one identified peak of the one or more peaks: process said data to classify a non-fall activity performed by the person during a time period based on the one or more predefined fall thresholds, the time period that encompasses said non-fall activity being at least one of before said identified peak or after said identified peak; and process a portion of the data corresponding to the time period that encompasses said identified non-fall activity to compute at least one parameter of the following parameters using said data: a lateral sway of the person; an acceleration of the upper part of the person; or a duration of the non-fall activity of the person; and the data processing system further programmed to: identify a risk of falling of the person at a subsequent time based on an evaluation of the at least one computed parameter with a statistical model, the statistical model to produce a numerical score of the risk of falling based on the at least one computed parameter.

2. The fall risk assessment system of claim 1, wherein the acceleration of the upper part of the person is associated with acceleration in at least one of a frontal direction or a vertical direction.

3. The fall risk assessment system of claim 1, wherein the said non-fall activity is at least one of standing, taking one step, taking multiple consecutive steps forming an episode of walking, transitioning from sitting to standing, or transitioning from standing to sitting.

4. The fall risk assessment system of claim 1, wherein the risk of falling is identified by evaluating at least one of the following: a number of one or more non-fall activities including the identified non-fall activity occurring during the time period; or an average of the at least one computed parameter, the average computed based on said portion of the data corresponding to the time period that encompasses said non-fall activity during said time period.

5. The fall risk assessment system of claim 4, wherein the number of one or more non-fall activities or the average of at least one computed parameter is evaluated over one day.

6. A fall risk assessment system comprising: a data processing system comprising one or more processor circuits configured to process data generated by a sensor, the data including information representative of at least one signal generated by the sensor in response to movement of a part of a body of a person, the data processing system programmed to at least; process said data to identify one or more peaks in the at least one signal by comparing values of the at least one signal with one or more predefined fall thresholds; and for at least one identified peak of the one or more peaks: process said data to classify a non-fall activity performed by the person during a time period based on the one or more predefined fall thresholds, the time period that encompasses said non-fall activity being at least one of before said identified peak or after said identified peak; and process a portion of the data corresponding to the time period that encompasses said identified non-fall activity to compute at least one parameter of the following parameters using said data: an acceleration of the part of the body of the person; or a duration of the non-fall activity of the person; and the data processing system further programmed to: identify a risk of falling of the person at a subsequent time based on an evaluation of the at least one computed parameter with a statistical model, the statistical model to produce a numerical score of the risk of falling based on the at least one computed parameter.

7. The fall risk assessment system of claim 6, wherein the acceleration of the part of the body of the person is associated with acceleration in at least one of a frontal direction or a vertical direction.

8. The fall risk assessment system of claim 6, wherein the said non-fall activity is at least one of standing, taking one step, taking multiple consecutive steps forming an episode of walking.

9. The fall risk assessment system of claim 6, wherein the risk of falling is evaluated by considering an average of the at least one computed parameter, the average computed based on said portion of the data corresponding to the time period that encompasses said non-fall activity during said time period.

10. The fall risk assessment system of claim 9, wherein the risk of falling is identified by evaluating a number of one or more non-fall activities including the identified non-fall activity occurring during the time period, wherein the number of one or more non-fall activities or the average of at least one parameter is evaluated during one day.

11. The fall risk assessment system of claim 6, wherein the sensor is attached to an upper part of the body.

12. The fall risk assessment system of claim 6, wherein the sensor is attached to a torso of the person or a shoulder of the person.

13. The fall risk assessment system of claim 6, wherein the sensor is attached to an arm of the person.

14. The fall risk assessment system of claim 6, wherein said one or more processor circuits of the data processing system are further programmed to identify said risk of falling using a history of one or more fall events of the person.

15. The fall risk assessment system of claim 14, wherein the one or more processor circuits of the data processing system are further programmed to use at least one algorithm to detect one or more falls of the person, and wherein said history of one or more fall events includes a history of the detected one or more falls.

16. The fall risk assessment system of claim 6, wherein said risk of falling is identified based at least in part on a linear combination of two or more of said at least one computed parameter.

17. The fall risk assessment system of claim 6, further comprising a communications system configured to receive said data from said sensor.

18. The fall risk assessment system of claim 6, further comprising the sensor.

19. The fall risk assessment system of claim 6, wherein said one or more processor circuits of the data processing system are further programmed to identify said risk of falling based at least in part on one or more of: a percentage of time the person is standing or walking; a number of postural transitions the person attempts; and a ratio of a duration of time the person standing to a duration of time the person walking.

20. The fall risk assessment system of claim 6, wherein said movement of the part of the body of the person is associated with activities of daily living.

21. A method of evaluating a risk of falling of a person, the method comprising: electronically receiving data generated by a sensor, the data representative of at least one signal generated by the sensor in response to movement of an upper part of a body of a person; processing said data to identify one or more peaks in the at least one signal by comparing values of the at least one signal with one or more predefined fall thresholds; for at least one identified peak of the one or more peaks; processing said data to classify a non-fall activity performed by the person during a first time period based on the one or more predefined fall thresholds, the first time period that encompasses said non-fall activity being at least one of before said identified peak or after said identified peak; processing a portion of the data corresponding to a the first time period that encompasses said identified non-fall activity to compute at least one parameter of the following parameters using said data: an acceleration of the upper part of the person; or a duration of the non-fall activity of the person; and identifying a risk of falling of the person at a subsequent time based on an evaluation of the at least one computed parameter with a statistical model, the statistical model to produce a numerical score of the risk of falling based on the at least one computed parameter.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The foregoing and other objects, features, and advantages of the invention will be apparent from the following detailed description of the invention, as illustrated in the accompanying drawings, in which like reference numerals designate like parts throughout the figures thereof and wherein:

(2) FIG. 1a illustrates how an elderly subject may wear the sensory module, and also shows the three components of acceleration measured by the sensory unit;

(3) FIG. 1b is a two-dimensional schematic of a subject wearing the sensory unit, and shows the subject's trunk lean angle , the direction of gravity, as well as the frontal and vertical acceleration components;

(4) FIG. 2 is a flowchart of the algorithms used to determine the time, time and duration of the subject's postural transitions;

(5) FIGS. 3a-f demonstrate the operation of the algorithms in determining the time, type and duration of the subject's postural transitions;

(6) FIG. 4 is a flowchart of the algorithms used to identify the walking periods, and to compute the subject's spatiotemporal parameters of gait;

(7) FIGS. 5a-c demonstrate the operation of the algorithms in identifying the walking periods, and in computing the subject's spatio-temporal parameters of gait;

(8) FIG. 6 is a flowchart of the algorithms used to detect and classify the lying posture;

(9) FIG. 7 is a flowchart of the algorithm used to compute the subject's risk of falling, and the quality of the subject's physical activity; and

(10) FIG. 8 is a flowchart of the algorithm used to automatically detect the subject's falls.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

(11) The present invention consists of a system and method for performing the following tasks during the user's everyday life: (1) monitoring the user's physical activity; (2) automatically detecting the user's falls; and (3) assessing the user's risk of falling. The second and third tasks are based on the results obtained from the first.

(12) As shown by FIG. 1a, the system includes a sensing module (SM) 101 for sensing, filtering and analyzing the user's 100 body movements. The SM 101 is positioned on the user's 100 upper body (typically, on the user's chest or torso), and is comprised of one to three accelerometers, each of which may be mono-axial or multi-axial. The only constraints on the accelerometer configuration are that (1) accelerations in three perpendicular directions must be measured; and (2) the accelerometer(s) is(are) configured to record accelerations in the frontal (F), vertical (V) and lateral (L) directions, which directions are relative to the user 100 (see FIG. 1a). In this document, all acceleration quantities are expressed in units of g (i.e., as multiples or fractions of g), where g is the gravitational constant equaling 9.81 m/s.sup.2: for example, by this convention an acceleration magnitude of 9.81 m/s.sup.2 (in SI units) will be expressed 1.

(13) The SM 101 may also include a data-storage system for storing the measured accelerations. An optional on-board communications system provides the SM 101 the capability to transmit the collected data and/or analyzed signals through either wired or wireless links for storage and/or for further offline analysis.

(14) Analysis of the measured acceleration signals may be carried out (1) entirely on-board the SM 101, (2) partially on-board the SM 101 and partially at other location(s), or (3) entirely at other location(s). In case some or all of the analysis is (are) carried out on-board the SM 101, a data processing circuit will be included on-board the SM to carry out the required computations according to software-based algorithms developed as part of the present invention. In case some or all of the analysis is carried at location(s) separate from the SM 101, the required data processing circuits performing the analysis may be ordinary or special-purpose computers, and are integrated with software-based algorithms developed as part of the present invention.

(15) A. Monitoring the User's Physical Activity

(16) Monitoring the user's physical activity consists of monitoring and assessing the user's postures, movements, trunk tilt, as well as fall-related task parameters. To this end, the system computes various parameters associated with the subject's movement from the data recorded by the SM 101. These parameters consist of: (a) the subject's trunk tilt (specified in degree, measuring the angle between the subject's trunk axis, and the axis aligned with the gravitational force-see FIG. 1b); (b) the type of the subject's postural transitions (PT); (c) the time of the subject's postural transitions; (d) the duration of the subject's postural transitions; (e) the duration of the subject's locomotion; (f) characterization of the subject's locomotion (gait analysis); and (g) the type of subject's postures (e.g., sitting, standing, lying).

(17) Use of accelerometers in place of gyroscopes by the present invention allows for long-term autonomous operability of the system. The associated challenges introduced by this replacement, however, consist of processing the resulting noisy accelerometer signals during everyday living activities.

(18) I. Identifying the Types of Postural Transitions, and Computing their Durations and Occurrences:

(19) The flowchart in FIG. 2 and FIGS. 3a-3f demonstrate the operation of the algorithms, developed as part of the present invention, used to continuously determine the type, time, and duration of the subject's postural transitions (in this case, SI-ST and ST-SI) during everyday movements. The algorithms use the frontal and vertical accelerometer signalsa.sub.F(t) and a.sub.V(t) respectively in FIG. 1awhere their time-varying nature is explicitly shown by including the time variable t in the notation used for these signals. In implementing the algorithms, the time variable t is by necessity discrete.

(20) FIG. 3a shows an example of the acceleration patterns recorded by the vertical and frontal accelerometers from an elderly subject with a high risk of falling (a.sub.V(t): gray line 301; a.sub.F(t): black line). As identified on the plot, the pattern consists of a sit-to-stand (SI-ST) postural transition followed by a period of walking and turning, followed by another postural transition (stand-to-sit; ST-SI).

(21) As shown in FIG. 2, the algorithm performs the following steps on the frontal accelerometer signal to determine the occurrence, duration and type of the postural transitions: 1) segmenting, followed by wavelet filtering (box 201 in FIG. 2) to remove signal artifacts induced by locomotion (e.g., walking, climbing or descending the stairs, etc.)see also the white trace 305 in FIG. 3b, an example of the resulting filtered signal a.sub.F-filt(t); 2) locating the local maximum peaks (denoted by a.sub.F-p 306 in FIG. 3b) in the filtered signal a.sub.F-filt(t) 305 through a peak-detection algorithmthis step corresponds to box 202 in FIG. 2; 3) for each postural transition, corresponding to a particular a.sub.F-p 306, computing an initial estimate of the postural transition duration (T.sub.1) by (boxes 203 and 204): (i) determining whether a.sub.F-p 306 is greater than a pre-defined threshold Th.sub.1; (ii) if yes, locating the local minima 307 in a.sub.F-filt(t) 305, within a specified time window, that precede and follow the particular maximum peak a.sub.F-p 306see FIG. 3b; (iii) computing T.sub.1 310 as the duration of the resulting time interval I.sub.1 separating the local minima computed above.

(22) The above steps suppress and remove signal artifacts, such as noisy peaks, associated with shocks or other locomotion activities.

(23) Following the initial determination of the postural transition duration (T.sub.1), the system computes a more accurate estimate of the postural transition duration, T.sub.2, by applying additional filters to the frontal acceleration signal only within a time interval that is centered at I.sub.1, but that is typically 10% to 30% longer in duration than T.sub.1 310. Such filtering of the frontal acceleration signal significantly decreases the requisite calculation costs, therefore enabling real-time implementation of the algorithm.

(24) If the value T.sub.1 310 surpasses a defined threshold, Th.sub.2 (box 205 in FIG. 2), the following steps are performed on the frontal accelerometer signal a.sub.F(t) only during a time interval that is centered at I.sub.1 but that is typically 10% to 30% longer in duration: 1) as represented by box 206 in FIG. 2, low-pass filtering the a.sub.F(t) signal during the time interval I.sub.1 by a wavelet; 2) as represented by box 207 in FIG. 2, locating the maximum peak (a.sub.F-p2 309) in the resulting filtered signal a.sub.F-filt2(t) 308 during time interval I.sub.1 (see FIG. 3c); 3) within a specified time window, locating a local minimum in a.sub.F-filt2(t) closest to, and preceding, the particular maximum peak a.sub.F-p2 (box 207 in FIG. 2); 4) within a specified time window, locating a local minimum in a.sub.F-filt2(t) closest to, and following the same maximum peak (box 207 in FIG. 2); 5) computing T.sub.2 311 (see FIG. 3c) as the duration of the resulting time interval I.sub.2 separating the local minima computed above (box 207 in FIG. 2);

(25) The time of the maximum peak a.sub.F-p2 represents the time of the postural transition, and the parameter T.sub.2 311 represents the estimate of the duration of the postural transition.

(26) For each postural transition, following the computation of its time of occurrence and its duration, the system uses the step-by-step algorithm below to identify its type (e.g., ST-SI or ST-SI): 1) as represented by boxes 209 and 210 in FIG. 2, for each postural transition if T.sub.2 exceeds a predefined threshold Th.sub.3, estimate the trunk tilt angle in the sagittal plane, , using a low-pass filtering of the a.sub.F(t) signal during the corresponding time interval I.sub.2since a.sub.F(t) consists of a -dependent gravitational component as well as a higher frequency, pure frontal-acceleration component, low-pass filtering removes the pure frontal-acceleration component, leading to a quantity proportional to the sin(); 2) estimate the time-varying inertial frontal and vertical accelerations a.sub.F-interial(t) and a.sub.V-interial(t) through the following coordinate transformation (see box 211 in FIG. 2):

(27) $\left[\begin{array}{c}{a}_{Finertial}\left(t\right)\\ a_{vinertial}\left(t\right)\end{array}\right]=\left[\begin{array}{cc}\cos \left(\left(t\right)\right)& -\sin \left(\left(t\right)\right)\\ \sin \left(\left(t\right)\right)& \cos \left(\left(t\right)\right)\end{array}\right]\left[\begin{array}{c}{a}_F\left(t\right)\\ a_V\left(t\right)\end{array}\right]-\left[\begin{array}{c}0\\ 1\end{array}\right],$ where, as mentioned before, the acceleration signal is expressed in units of g (g represents the gravitational constant (9.81 m/s.sup.2))see also FIG. 1b for a free-body diagram showing the inertial acceleration components; 3) in parallel, apply an adequate, cascaded low-pass filter to remove the artifacts from a.sub.V(t), where the low-pass filter functions as follows: (i) removal of the gravitational component of a.sub.V(t) 312 (FIG. 3e) using the following equations (see also box 211 in FIG. 2):

(28) $a_F\left(t\right)=\left[a_{V-\mathrm{inertial}}\left(t\right)+1\right]\sin \left(\left(t\right)\right)+a_{F-\mathrm{inertial}}\left(t\right)\cos \left(\left(t\right)\right);$$a_V\left(t\right)=\left[a_{V-\mathrm{inertial}}\left(t\right)+1\right]\cos \left(\left(t\right)\right)+a_{F-\mathrm{inertial}}\left(t\right)\sin \left(\left(t\right)\right);$$a_{V-\mathrm{filt}}\left(t\right)=\sqrt{{\left[a_F\left(t\right)\right]}^2+{\left[a_V\left(t\right)\right]}^2};$ (ii) low-pass filtering the resulting signal a.sub.V-filt(t) 313, leading to a.sub.V-filt2(t); and (iii) filtering this signal by a moving-average filter to obtain a.sub.V-filt3(t) (see also box 212 in FIG. 2); 4) as exemplified in FIGS. 3e-3f, determine the local peaks in a.sub.V-filt3(t) using a peak detection algorithm (box 213 in FIG. 2); the resulting positive and negative peaksP.sub.max 315 and P.sub.min 316, respectivelyexceeding a predefined threshold Th.sub.4, are identified (boxes 214 and 215 in FIG. 2); 5) classify the detected postural transition as sit-to-stand or stand-to-sit through the sequence by which P.sub.max and P.sub.min occur e.g., a Pa followed by a P.sub.min identifies the postural transition as a sit-to-stand pattern (box 316 in FIG. 2; see also FIGS. 3e-3f); 6) apply a post-processing algorithm to prevent misclassification of postures and postural transitions: for each postural transition, the classification as ST-SI or SI-ST will be corrected based on the preceding and subsequent sequences of postural transitions.
II. Analyzing Gait, and Identifying the Corresponding Walking Periods:

(29) FIG. 4 describes in flowchart form the software-based algorithm, developed as part of the invented system, to identify the subject's walking periods and measure his or her gait parameters. Using data recorded by the accelerometers, the algorithm can distinguish left and right gait steps, as well estimate the spatiotemporal gait parameters, e.g., swing, stance, double support, and gait speed.

(30) The algorithm consists of the following steps: 1) remove from consideration data during time periods associated with postural transitions and lying (boxes 401-402 in FIG. 4); 2) compute the time-varying norm (i.e., time-varying magnitude) of the vertical and horizontal accelerometer signals as:

(31) $a_F\left(t\right)=\left[a_{V-\mathrm{inertial}}\left(t\right)+1\right]\sin \left(\left(t\right)\right)+a_{F-\mathrm{inertial}}\left(t\right)\cos \left(\left(t\right)\right);$$a_V\left(t\right)=\left[a_{V-\mathrm{inertial}}\left(t\right)+1\right]\cos \left(\left(t\right)\right)+a_{F-\mathrm{inertial}}\left(t\right)\sin \left(\left(t\right)\right);$$a_{V-\mathrm{filt}}\left(t\right)=\sqrt{{\left[a_F\left(t\right)\right]}^2+{\left[a_V\left(t\right)\right]}^2};$
where (t) represents the time-varying trunk angle, and a.sub.V-interial(t) and a.sub.F-interial(t) represent the time-varying vertical and frontal acceleration components, respectively; FIG. 5b shows the resulting waveform, a.sub.V-filt3(t) 503see FIG. 1b for the free-body diagram leading to the above formulas; these formulas allow for suppression of the movement artifacts derived from the rotations of the subject's trunk; 3) remove the gravitational component from the vertical acceleration signal in two steps: first, use formula stated in step (2) to compute a.sub.V-filt3(t) 503; second, as shown by box 403 in FIG. 4, band-pass filter the result, leading to a.sub.V-filt4-(t) 504 (see FIG. 5c); 4) as represented by box 404 in FIG. 4, identify gait steps as the peaks 505 (see, FIG. 5c) in the a.sub.V-filt4(t) signal 504; 5) verify the sequence of the detected peaks according to pre-defined conditions for gait patterns (box 405 in FIG. 4); 6) distinguish left and right steps (box 407 in FIG. 4) using the signal a.sub.L(t) from the lateral accelerometerspecifically, (i) the subject's lateral velocity v.sub.L(t) is computed by integrating a.sub.L(t) during the recognized walking periods; (ii) the relationship between the locations of the positive and negative peaks in v.sub.L(t) with the identified peak in the filtered vertical acceleration signal, a.sub.V-filt4(t) 504, allows for left and right steps be distinguished.

(32) This algorithm, furthermore, enables both the recognition of undetected gait steps, and the removal of false detected steps.

(33) The system, through another algorithm, computes the times of heel-strike (initial contact) and toe-off (final contact) events using information extracted from the frontal and vertical acceleration signalsthis step corresponds to box 408 in FIG. 4. Specifically, the local minimum and maximum peaks in the frontal acceleration signal surrounding each identified vertical acceleration peak are used to identify heel-strike event and toe-off events. Following a heel-strike event, the subject's trunk continues to moves forward. As the toe-off event occurs, the trunk slows down, leading to a negative peak in the frontal accelerometer signal. Although a heel-strike event can be estimated using the vertical acceleration signal, when an impact is identified, the positive peak of the frontal acceleration pattern offers a significantly lesser noisy source for identification of the heel-strike event. Determination of these event times facilitates the measurement of the temporal parameters (e.g., stance, swing, double support, step time, gait cycle time, etc.) and other relevant information associated with the spatial parameters (i.e. stride velocity, step length and stride length).

(34) Gait speed (i.e., stride velocity) is computed (box 410 in FIG. 4) using information from the detected gait cycle and the amplitude of acceleration during the double support.

(35) III. Detecting and Classifying the Lying Posture.

(36) The system distinguishes lying from sitting and standing by comparing the angle of the vertical accelerometer signal a.sub.V(t) to that of the gravitational component. While the vertical accelerometer measures almost zero during lying periods, its value is significantly greater during sitting and upright posturesin some cases the value is close to the gravitational constant.

(37) The system identifies both the sit/stand-to-lying (SI/ST-L) and the mirror opposite (i.e., L-SI/ST) postural transitions using the following algorithm: 1) band-pass filter the vertical accelerometer signal (box 600 in FIG. 6); 2) calculate the gradient of the resulting the filtered signal a.sub.V-filt5(t) (box 601 in FIG. 6); 3) determine the maximum or minimum peak (P.sub.) of this gradient (box FIG. 6, box 602); 4) if the absolute value of the detected peak P.sub. exceeds a pre-defined threshold Th.sub.5 (box 603, FIG. 6), estimate the duration of lying postural transition using a local peak detection scheme to identify peaks preceding (L.sub.initial) and following (L.sub.terminal) P (box 604, FIG. 6); 5) identify a lying posture at the time of the detected peak when (i) the absolute value of the detected peak exceeds a threshold Th.sub.5 (box 603, FIG. 6); and (ii) the average value of a.sub.V-filt5(t) during the 3 seconds preceding the L.sub.initial is higher than a pre-defined threshold Th.sub.6 (boxes 605-606, FIG. 6); and (iii) the average value of a.sub.V-filt5(t) during the 3 seconds following the L.sub.terminal is lower than a threshold Th.sub.7 (boxes 605-606, FIG. 6); 6) detect/identify a lying-to-sit/stand (L-SI/ST) postural transition at the time of the detected peak (P.sub.) when (i) the absolute value of the detected peak exceeds a predefined threshold Th.sub.5 (box 603, FIG. 6); and (ii) the average value of a.sub.V-filt5(t) during the 3 seconds preceding the L.sub.initial is lower than Th.sub.8 (boxes 605-607, FIG. 6); and (iii) the average value of a.sub.V-filt5(t) during the 3 seconds following the L.sub.terminal is higher than a threshold Th.sub.9 (boxes 605-607, FIG. 6); 7) classify the lying posture further as lying on back, lying on the front, or on the sides (left or right) on the basis of the value of the frontal accelerometer signal (box 608, FIG. 6); 8) further classify lying on the side into lying on the right and lying on the left according to the value of the lateral accelerometer signal.
B. Computing the Risk of Falling and the Quality of the Subject's Physical Activity

(38) By monitoring the subject's physical activity, the invented system both evaluates the quality of the subject's physical activity, and computes the decline or progress in the subject's functional performance. FIG. 7 presents the flowchart of the corresponding software-based algorithm, developed as part of the invented system.

(39) The subject's risk of falling (RoF) during everyday life is computed by quantifying the quality of the subject's postural transitions and physical activities using the following algorithm: 1) estimate the lateral sway (.sub.sway) of the subject during PT by computing the standard deviation of the lateral accelerometer during PT (box 700, FIG. 7); 2) estimate the jerkiness in the subject's movement in all directions (.sub.V-jerk, .sub.F-jerk, and .sub.L-jerk)computed as the standard deviation of the band-pass filtered acceleration signals in the frontal, vertical and lateral directions (box 701, FIG. 7); 3) compute the mean (.sub.TD) and standard deviation (.sub.TD) of the durations of the subject's postural transitions (T.sub.2), over a day (box 702, FIG. 7); 4) compute the number of successive postural transitions (N.sub.Succ_PT) required for a subject to accomplish a single taskan example is multiple unsuccessful attempts by a subject to rise from a chair (box 703, FIG. 7); 5) evaluate the quality of physical activity by computing the fraction of the time that subject has active posture (including walking); the number of PTs per day; the number of walking episodes during a day; and the longest continuous walking period per day (boxes 704-706, FIG. 7); 6) evaluate the subject's risk of falling by inputting the above parameters to a statistical model (e.g., stepwise) that provides a linear combination of the calculated parameters to yield a single score representative the subject's RoF (box 707, FIG. 7). A subject is considered to be at a high-risk of falling if the linear combination passes beyond a threshold, which may be predefined, or may change adaptively.

(40) To identify a subject at a high risk of falling more accurately, the system continually adjusts the requisite threshold values based on the history of falls or other similar events detected by the algorithm (e.g., high-impact experienced shortly after a postural transition, very short ST-SI durations, etc.)

(41) I. Automatic Fall Detection.

(42) The present invention uses a novel algorithm, based solely on accelerometer signals, to automatically identify falls during the subject's everyday life with high sensitivity and specificity. The fall-detection algorithm described here uses information about the subject's physical activity, as well as posture. The flowchart in FIG. 8 describes in complete the algorithm developed to automatically detect the subject's falls. The following summarizes the algorithm: 1) compute the norm (magnitude) of acceleration in the transversal plane, a.sub.trans(t) from the frontal and lateral acceleration signalsa.sub.F(t) and a.sub.L(t), respectivelythrough:
a.sub.trans(t)={square root over ([a.sub.F(T)].sup.2+[a.sub.V(t)].sup.2)}(box 800); 2) apply a peak-detection algorithm (box 801) to a.sub.trans(t) to identify the presence of shocks a.sub.transP.sub.max; 3) confirm a fall event by considering the subject's PA and posture prior to impact times (marked by the identified shocks)this step is carried out using algorithms described above; 4) use different algorithms to identify a fall event, depending on the results of step (3) supra: (i) if impacts occur while subject is walking or turning, depending on whether the impacts occurred after right or left step, the algorithm chooses appropriate thresholds and coefficients required for subsequent steps (Th.sub.8: box 812; Th.sub.9: box 814; and coefficients of the multivariable model: box 816); (ii) if activity preceding the shock is not identified as walking, turning or any sequential locomotion (e.g., walking upstairs or downstairs,) the algorithm would identify as fall events only the shocks that occur after a postural transition to sitting or lying; (iii) Next, thresholds and coefficients required for subsequent steps are modified; 5) segment the shock-pattern following a postural transition into pre-shock, impact, and post-shock phases based on the location of local minimum peaks relative to the absolute maximum peak (p.sub.max) in the signal a.sub.trans(t) (box 810, FIG. 8); the set of thresholds chosen according to step (4) supra, and used by the algorithm depends on whether the post-shock posture is sitting or lying; 6) estimate the shock width (.sub.shock) using the local minimum peaks before and after each the peak p.sub.max (box 811, FIG. 8); consider the peak to be an artifact and subsequently ignored if its width does not exceed the threshold Th.sub.8 (box 812, FIG. 8); 7) if the peak is not an artifact, compute the subject's speed during the pre-shock phase by integrating the pattern of vertical accelerometerV.sub.V(t) (box 813, FIG. 8); for the peak to be recognized as a fall, the peak of the velocity profile must exceed the threshold Th.sub.9 (box 814, FIG. 8); 8) compute the following descriptors (box 815, FIG. 8): (i) sum of all accelerations at the time of impact t.sub.impact as:
a.sub.total(t.sub.impact)=a.sub.F(t.sub.impact)+a.sub.V(t.sub.impact+a.sub.V(t.sub.impact); (ii) the sum frontal and lateral accelerations at impact time:
a.sub.F+L(t.sub.impact)=a.sub.F(t.sub.impact)+a.sub.L(t.sub.impact); (iii) the difference of speed in each direction at the impact time (V.sub.F-impact, V.sub.V-impact, and V.sub.L-impact); and (iv) energy of the norm of vertical and frontal acceleration during the impact phase (.sub.shock):

(43) $E_{\mathrm{Impact}}={}_{\mathrm{Shock}}\sqrt{{a_F\left(t\right)}^2+{a_V\left(t\right)}^2\mathrm{dt}};$ 9) identify a fall event through a multivariable model (stepwise or linear combination) that uses the above descriptors as inputs and coefficients chosen in step (4) supra (box 816, FIG. 8); 10) identify a fall as serious if the post-fall activities represent an unusual activity pattern, such as a long-duration rest, or multiple unsuccessful postural transitions (boxes 818-819, FIG. 8); in one embodiment of the invention, an alarm will be set off following a serious fall.
II. Physical Activity Classification.

(44) The algorithms described above will classify the subject's physical activity and posture, determine his or her risk of falling and quality of movements. In addition, several rules will be applied to improve the classifications performed by the above algorithms. These rules include, but are not limited to, the following: 1) If two contradictory states are detected (e.g., lying with walking or sitting with walking) preference is first given to lying, then to walking, and finally to postural transitions. This rule is based on the rationale that the lying posture is classified with the least amount of error. It should be noted that since the algorithms for different postural detections operate independently, two contradictory sets of activities may be identified. 2) Two successive postural transitions classified as the same type (e.g., SI-ST followed by SI-ST) are not possiblethe classifications are modified according to the preceding and subsequent activities. 3) Elderly subjects cannot lean backwards after a SI-ST transition with a high likelihood. The algorithm estimates the trunk lean angle based on the trunk angle before (.sub.PT-pre) and/or following (.sub.PT-post) the postural transition. (i) Both .sub.PT-pre and .sub.PT-post are estimated based on the mean (E[.]) of the frontal acceleration during the rest period immediately before, or after a postural transition, according to the following formulas:
.sub.PT-pre=sin.sup.1(E[a.sub.F(t)|pre-PT-rest)
.sub.PT-post=sin.sup.1(E[a.sub.F(t)|post-PT-rest) where E[a.sub.F(t)pre-PT-rest] denotes the mean of the frontal acceleration signal during the rest period immediately before the postural transition; E[aF(t)post-PT-rest] denotes the corresponding mean after the postural transition. (ii) If the standard deviation of both frontal and vertical accelerations during a local interval before or after a postural transition were lower than a pre-defined threshold, the algorithm will classify that duration as a rest period. (iii) Sensor inclination (.sub.initial) is computed from the average of the frontal accelerometer signal during a recognized walking episode containing at least ten steps: .sub.initial=sin.sup.1([a.sub.F(t)|walking; 10 steps]. (iv) The backwards-leaning state is detected if, subtracting .sub.initial from .sub.PT-pre (or .sub.PT-post) yields a value lower than a pre-defined threshold. 4) The duration of the lying posture should be more than a specified length (e.g., 30 seconds). 5) For an episode to be classified as walking, it must include at least three successive steps within a predefined interval. 6) Since it is improbable for a person, especially an elderly subject, to stand for long periods without any movements, long standing periods without additional activity (e.g., more than three minutes) are interpreted as sitting. This rule applies if the standard deviations of both the vertical and frontal accelerations are below pre-defined thresholds.

REFERENCES

(45) [1] B. Najafi and K. Aminian, Body movement monitoring system for elderly people, determines time and duration of postural transition (2000, European and US Patent), EP1195139-A1 EP810920 5 Oct. 2000; US2004015103-A1 US398462 4 Apr. 2003, 2000. [2] B. Najafi, K. Aminian, F. Loew, Y. Blanc, and P. A. Robert, Measurement of stand-sit and sit-stand transitions using a miniature gyroscope and its application in fall risk evaluation in the elderly, Ieee Transactions on Biomedical Engineering, vol. 49, pp. 843-851, 2002. [3] B. Najafi, K. Aminian, A. Paraschiv-Ionescu, F. Loew, C. J. Bula, and P. Robert, Ambulatory system for human motion analysis using a kinematic sensor: Monitoring of daily physical activity in the elderly, Ieee Transactions on Biomedical Engineering, vol. 50, pp. 711-723, 2003. [4] R. W. Bohannon, A. W. Andrews, and M. W. Thomas, Walking speed: reference values and correlates for older adults, J Orthop Sports Phys Ther, vol. 24, pp. 86-90, 1996. [5] K. Aminian, B. Najafi, C. Bula, P. F. Leyvraz, and P. Robert, Spatio-temporal parameters of gait measured by an ambulatory system using miniature gyroscopes, Journal of Biomechanics, vol. 35, pp. 689-699, 2002. [6] K. Aminian, B. Najafi, J. Gramiger, P. Morel, and N. Bijan, Autonomous measuring unit for human movement has sensors, conditioning circuit, display, and circuit for recording kinematic parameters of body segment, ECOLE POLYTECHNIQUE FEDERALE LAUSANNE (ECOL-Non-standard) AMINIAN K (AMIN-Individual) BUAN N (BUA-Individual) GRAMIGER J (GRAM-Individual) MOREL P (MORE-Individual). [7] K. Aminian, K. Rezakhanlou, E. De Andres, C. Fritsch, P. F. Leyvraz, and P. Robert, Temporal feature estimation during walking using miniature accelerometers: an analysis of gait improvement after hip arthroplasty, Medical & Biological Engineering & Computing, vol. 37, pp. 686-691, 1999. [8] S. R. Cummings, M. C. Nevitt, and S. Kidd, Forgetting falls. The limited accuracy of recall of falls in the elderly, J Am Geriatr Soc, vol. 36, pp. 613-6, 1988. [9] D. Oliver, M. Britton, P. Seed, F. C. Martin, and A. H. Hopper, Development and evaluation of evidence based risk assessment tool (STRATIFY) to predict which elderly inpatients will fall: case-control and cohort studies, Bmj, vol. 315, pp. 104953, 1997. [10] M. E. Tinetti, T. F. Williams, and R. Mayewski, Fall risk index for elderly patients based on number of chronic disabilities, Am J Med, vol. 80, pp. 429-34, 1986. [11] K. Doughty, R. Lewis, and A. McIntosh, The design of a practical and reliable fall detector for community and institutional telecare, J Telemed Telecare, vol. 6 Suppl 1, pp. S150-4, 2000. [12] U. Lindemann, A. Hock, M. Stuber, W. Keck, and C. Becker, Evaluation of a fall detector based on accelerometers: a pilot study, Med Biol Eng Comput, vol. 43, pp. 548-51, 2005. [13] Y. Depeursinge, J. Krauss, and M. El-Khoury, Device for monitoring the activity of a person and/or detecting a fall, U.S. Pat. No. 6,201,476, 2001. [14] N. Noury, G. Barralon, G. Virone, P. Boissy, M. Hamel, and P. Rumeau, A smart sensor based on rules and its evaluation in daily routines, presented at 25th Annual International Conference of the IEEE Eng. Med. Biol. Society, 2003.