Step counting system and method for clinical research

Abstract

A system for enhanced analysis of activity signals from on-body devices for tracking activity of a plurality of clinical research subjects is provided. The system may comprise a plurality of on-body devices, each configured to be worn by a separate subject, and each device including an accelerometer and / or a gyroscope configured to generate an activity signal based on body movement of the subject wearing said device. A remote computing device may be configured to filter and decompose the received activity signal to identify an Intrinsic Mode Function ("IMF"), compute an instantaneous energy of the IMF, and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps. The steps derived from subjects that received a health intervention may then be compared against steps derived from subjects that did not receive the health intervention to assess an efficacy of the health intervention.

Description

STEP COUNTING SYSTEM AND METHOD FOR CLINICAL RESEARCHFIELD

[0001] The present disclosure pertains to step counting, and more particularly to systems and methods for conducting clinical research by counting steps of individuals using on-body devices in a free-living environment.BACKGROUND

[0002] Accurately counting steps of individuals provides information which is useful for clinical research as well as a variety of health-related applications. One common method of counting steps is to provide a user with an on-body device (e.g., an on-body device) including a sensor (e.g., an accelerometer and / or a gyroscope) that outputs a signal in response to movement of the user. The recorded signal could be regular, semi-regular or very irregular in different scenarios. In the recorded signal, the rhythm corresponding to steps taken by the user could be obscured by other motions of the body part to which the device is attached (e.g., the wrist) while the user is walking. Most existing step count algorithms fail to give an accurate estimation for steps from irregular or semi-regular signals. Accordingly, improvements to methods and systems for counting steps are desirable.SUMMARY

[0003] According to one embodiment of the present disclosure, a system is provided for enhanced analysis of activity signals from on-body devices for tracking activity of a plurality of clinical research subjects. The system may comprise a plurality of on-body devices each configured to be worn by a separate clinical research subject of the plurality of clinical research subjects. Each device may include at least one of an accelerometer and a gyroscope configured to generate an activity signal based on body movement of the subject wearing said device. The system may further comprise a remote computing device including at least oneprocessor and at least one memory device. The remote computing device may be configured to receive activity signals from each on-body device via a network and store the received signals in the at least one memory device. The at least one processor may be configured to execute computer-readable instructions stored in the at least one memory device to, for each respective received activity signal: compute an activity signal norm from the respective received activity signal; filter the activity signal norm to produce a filtered activity signal; decompose the filtered activity signal to identify an Intrinsic Mode Function (“IMF”); identify peaks of the IMF; compute an instantaneous energy of the IMF; and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps taken by the clinical research subject wearing the on-body device from which the respective received activity signal was received. In one aspect of the embodiment, the at least one processor may be further configured to compare the counted steps derived from activity signals for a first subset of the plurality of clinical research subjects to the counted steps derived from activity signals for a second subset of the plurality of clinical research subjects. For instance, the first subset of subjects may have been subjected to a health intervention, whereas the second subset of clinical research subjects were not subjected to the health intervention. By comparing the step counts derived for each subset of clinical research subjects, an efficacy of the health intervention at treating a health condition may be determined.

[0004] In another embodiment, a method is provided for conducting clinical research for evaluating efficacy of a health intervention in treating a health condition using enhanced analysis of activity signals from on-body devices. The method may comprise providing an on-body device to each subject of a plurality of subjects, each subject being a participant in clinical research for the health intervention, and each on-body device including at least one of an accelerometer and a gyroscope configured to derive an activity signal from body movement of the subject to which the device was provided; subjecting a first subset of the plurality of subjects to thehealth intervention, wherein a second subset of the plurality of subjects is not subjected to the health intervention; receiving at a computing device a plurality of activity signals communicated to the computing device from the plurality of on-body devices via a network, wherein a first subset of the plurality of activity signals is from the on-body device of subjects within the first subset of subjects and a second subset of the plurality of activity signals is from the on-body device of subjects within the second subset of subjects; for each respective activity signal in the plurality of received activity signals: computing an activity signal norm from the respective activity signal, filtering the activity signal norm to produce a filtered activity signal, decomposing the filtered activity signal to identify an Intrinsic Mode Function (“IMF”), identifying peaks of the IMF, computing an instantaneous energy of the IMF, and counting peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps; comparing (1 ) the counted steps derived from activity signals received from on-body devices provided to the first subset of subjects to (2) the counted steps derived from activity signals received from on-body devices provided to the second subset of subjects; and computing a measure indicative of the efficacy of the health intervention based on the comparison.

[0005] In another embodiment, a system for enhanced analysis of activity signals is provided, comprising: an on-body device configured to be worn by a subject, said device including at least one of an accelerometer and a gyroscope configured to generate an activity signal based on body movement of the subject; and a remote computing device including at least one processor and at least one memory device, wherein: the remote computing device is configured to receive the activity signals from said on-body device, and the at least one processor is configured to execute computer-readable instructions stored in the at least one memory device to: compute an activity signal norm from the received activity signal, filter the activity signal norm to produce a filtered activity signal, decompose the filtered activity signal to identify an Intrinsic Mode Function (“IMF”), identify peaks ofthe IMF; compute an instantaneous energy of the IMF, and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps taken by the subject.

[0006] In yet another embodiment, a computer-implemented method for counting steps is provided, the method comprising: receiving, by at least one processor, an activity signal from at least one of a three-axis accelerometer or a gyroscope of an on-body device; computing, by the at least one processor, an activity signal norm from the activity signal; filtering, by the at least one processor, the activity signal norm to produce a filtered activity signal; decomposing, by the at least one processor, the filtered activity signal to identify an Intrinsic Mode Function (“IMF”); computing, by the at least one processor, an instantaneous energy of the IMF; counting, by the at least one processor, peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps; and outputting the counted steps.BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The above-mentioned and other advantages and objects of this disclosure, and the manner of attaining them, will become more apparent, and the disclosure itself will be better understood, by reference to the following description of embodiments of the invention taken in conjunction with the accompanying drawings, wherein:

[0008] FIG. 1 A is a graph depicting a regular signal generated by an accelerometer of an on-body device;

[0009] FIG. 1 B is a graph depicting a semi-regular signal generated by an accelerometer of an on-body device;

[0010] FIG. 2 is a graph depicting an irregular signal generated by an accelerometer of an on-body device;

[0011] FIG. 3A is a flow chart of a method for conducting clinical research;

[0012] FIG. 3B is a flow chart of a step counting method according to one embodiment of the present disclosure;

[0013] FIG. 4 is a graph of an activity signal before and after Gaussian filtering;

[0014] FIG. 5 is a flow chart for extracting an Intrinsic Mode Function (“IMF”) from a filtered activity signal by using Empirical Mode Decomposition (“EMD”);

[0015] FIG. 6 is a graph of a filtered activity signal, a first spline fitted to the maxima of the filtered activity signal, a second spline fitted to the minima of the filtered activity signal, and a mean of the first and the second spline;

[0016] FIG. 7 is a flow chart of a method of decomposing a filtered activity signal using EMD to identify an IMF;

[0017] FIG. 8 is a graph of a filtered activity signal, a first spline fitted to the maxima of the filtered activity signal, a second spline fitted to the minima of the filtered activity signal, and a mean of the first and the second spline resulting from a first iteration of the method of FIG. 7;

[0018] FIGS. 9A-9E are graphs of the signals depicted in FIG. 8 after subsequent iterations of the method of FIG. 7;

[0019] FIG. 9F is a graph of an IMF identified as a result of iterations of the method of FIG. 7;

[0020] FIG. 10 provides graphs of an activity signal norm, a filtered activity signal, and an IMF obtained by using EMD;

[0021] FIG. 11 provides graphs of an IMF, the instantaneous energy of the IMF, and the IMF with identified steps;

[0022] FIG. 12 is a block diagram of obtaining an activity signal by using a Hilbert transform; and

[0023] FIG. 13 is a block diagram of a system according to an embodiment of the present disclosure.

[0024] Corresponding reference characters indicate corresponding parts throughout the several views. Although the drawings represent embodiments of the present disclosure, the drawings are not necessarily to scale, and certain features may be exaggerated or omitted in some of the drawings in order to better illustrate and explain the present disclosure.DETAILED DESCRIPTION

[0025] In many applications, inertial measurement units incorporated into on- body devices (such as devices configured to be worn around a user’s wrist, waist, chest, limb, or other body part, or devices incorporated into a user’s clothing, such as the user’s shoes or socks, etc.) are used to collect information about human body motion including the number of steps an individual takes in a particular period of time. The on-body devices are configured with accelerometers and / or gyroscopes that, along with a plurality of other electronic components in the on-body devices, permit collection, storage, analysis and transmission of movement data which includes data from which the number of steps can be determined. In one application, the step count of an individual is used as a proxy for how active the individual is during clinical research testing the efficacy of a health intervention. Examples of clinical research may include clinical studies or clinical trials.

[0026] For example, individuals in one arm of a clinical trial or clinical study may be subjected to a health intervention to treat a health condition while other individuals in another arm of the clinical trial or study may not be subject to the health intervention (e.g., be given a placebo). Non-limiting examples of health interventions that may be tested in a clinical trial or study include medications or drugs, medical devices, digital therapeutics, diet and / or exercise regimens, or any other therapy, regimen, or intervention that may be expected to affect the condition of subjects. The health condition being treated may, but need not be, a disease, such as Parkinsons, Alzheimer’s, or some other neurodegenerative disease. Alternatively or in addition, the health condition may be a physiological condition orsymptom, such as obesity, fatigue, chronic pain, and / or depression. The individuals in both arms of the clinical trial or clinical study may be provided with an on-body device to wear as they go about their daily lives over a monitoring period (e.g., a period of days, weeks or months) to provide motion data from which an indication of the efficacy of the health intervention may be obtained. If the subjects that were subjected to the health intervention were more active (e.g., exhibited higher step counts) compared to subjects that were not subjected to the health intervention, then this may indicate that the health intervention was effective in treating the health condition. The data collected by the on-body devices may be transmitted to a facility conducting the trial and analysed either while the trial I study is ongoing, or upon conclusion of the trial I study.

[0027] In the above-described example, after the health intervention is administered or provided to the designated subset of subjects, the subjects in the clinical research may go about their daily lives while providing the data to the facility. Thus, the individual subjects do not need to return periodically to the facility for testing and observation. Moreover, the subjects’ activity may be tracked in a natural, free-living environment such as the individuals’ home or workplace. This may provide more accurate data as compared to data collected in a clinical setting, such as the unfamiliar environment of a healthcare facility like a hospital, clinic, or doctor’s office.

[0028] The identification of actual steps from the motion data is made more difficult by the fact that during normal daily activity in a free-living environment (as opposed to, for example, performing a straight-line walking exercise in a controlled environment), an individual may move his or her body for a variety of reasons, may stop walking and start suddenly, may change the pace of his or her gait, etc. These behaviours introduce noise into the recorded signals which may obfuscate the rhythmic data representing actual steps. Generally speaking, such signals may be classified as regular, semi-regular or irregular, depending upon how rhythmic thesignal is. A regular signal in this context may be characterized as a quasiperiodic signal or function, meaning that it is similar to a periodic signal.

[0029] FIG. 1 A shows an example of a regular activity signal 10 generated by an accelerometer of an on-body device. It should be understood that while this description primarily refers to accelerometers, the signals representing movement of individuals may be provided by on-body devices that include other types of sensors such as gyroscopes, and the techniques applied to accelerometer data herein may also be applied to data derived from such other types of sensors. Also, the term “activity signal” as used herein is intended to mean any signal generated by an on- body device indicative of movement of a user’s body part (e.g., the user’s wrist). Referring back to FIG. 1 A, the time between peaks 12 and between valleys 14 is substantially constant, although not quite. In other words, the periodicity of the signal is substantially uniform across the sample time period. FIG. 1 B depicts a semi-regular activity signal 16 and FIG. 10 depicts an irregular activity signal 18. As shown in FIG. 1 C, the periodicity of the irregular activity signal 18 is entirely lost. In other words, the rhythmic data representing actual steps is not apparent from the raw activity signal. The semi-regular activity signal 16 of FIG. 1 B represents an example activity signal between the regular signal classification and the irregular signal classification. As shown, the semi-regular activity signal 16 includes portions (e.g., portion 20) where the activity signal 16 resembles a regular signal and portions (e.g., portion 22) where the activity signal 16 resembles an irregular signal. The principles of the present disclosure as described herein are most effective at identifying the rhythmic data representing actual steps in semi-regular and irregular activity signals.

[0030] Referring now to FIG. 3A, a simplified flow chart is shown representing the primary steps in a method for conducting clinical research, according to one embodiment of the present disclosure. The clinical research may be designed to evaluate the efficacy of a health intervention in treating a health condition usingenhanced analysis of activity signals from on-body devices. As shown, a method 23 according to the present disclosure begins at step 25 by providing an on-body device to each of a plurality of clinical research subjects. Each device may include at least one of an accelerometer and a gyroscope configured to derive an activity signal from body movement of the subject to which the device was provided. Each subject may be a participant in a clinical trial or clinical study designed to test the efficacy of the health intervention in treating the health condition. Devices may be provided to subjects upon an initial visit when they are enrolled in the clinical research, or devices may be mailed and / or delivered to subjects’ homes. Subjects may be instructed by researchers to wear the on-body devices as they go about their daily lives during the course of the clinical research. In some embodiments, a first subset of the plurality of subjects may be subjected to a health intervention, while a second subset of the plurality of subjects may not be subjected to the health intervention (e.g., given a placebo).

[0031] At step 27, activity signals from each on-body device provided to the subjects is received at a remote computing device. As used herein, the term “remote computing device” may comprise a single computing device, or multiple devices, such as a cluster or cloud of computing devices operating together to perform the functions described herein. In some embodiments, the activity signals may be received by the remote computing device via a network, such as the Internet. The activity signals may pass through one or more intermediate devices or networks before reaching the remote computing device. Depending on the embodiment, activity signals may be uploaded to the remote computing device at different frequencies. For instance, activity signals may be uploaded substantially in real time (e.g., within seconds or recording), or they may be uploaded once every few hours, or they may be uploaded once a day, once a week, once a month, or any other suitable frequency. In some embodiments, the frequency at which data is uploaded may be varied according to context or the needs of the clinical research. In someembodiments, activity signals may be uploaded in a single batch at the end of the clinical research. A first subset of the plurality of received activity signals may be from on-body devices of subjects within the first subset of subjects (who received the health intervention). A second subset of the plurality of received activity signals may be from on-body devices of subjects within the second subset of subjects (who did not receive the health intervention).

[0032] At step 29, the received activity signals may be analysed to derive step counts for each subject, or at least, for each activity signal that was received. These step counts are indicative of the number of steps that each corresponding subject took while wearing the on-body devices during the clinical research. Optionally, the derived step counts may be associated with a certain time period, such as a certain day or hour during the clinical research. Details regarding how the received activity signals are analysed to derive step counts are described below and herein.

[0033] At step 31 , the derived step counts are compared to compute a measure indicative of the efficacy of the health intervention at treating the health condition. For instance, an average step count for the first subset of subjects (who received the health intervention) may be compared to an average step count for the second subset of subjects (who did not receive the health intervention). If the difference between these two averages is statistically significant, this may indicate that the health intervention was effective at treating the health condition. To determine whether the difference between the two averages is statistically significant, the method may further include (at step 31 ) an analysis of the noisiness or variability in the derived step counts for both the first and the second subset of subjects. If the difference between the two averages is larger than can be accounted for by the observed noisiness or variability for each subset of subjects, then the difference may be statistically significant.

[0034] Referring now to FIG. 3B, a simplified flow chart is shown representing the primary steps for analyzing activity signals to derive step counts, according toone embodiment of the present disclosure. Each of the steps depicted in FIG. 3B will be described in greater detail below. As shown, a method 24 according to the present disclosure begins at step 26 by determining the norm of the X, Y and Z components of an activity signal from an accelerometer of an on-body device. Signals derived from accelerometers of existing on-body devices may sometimes be recorded at a frequency of 60-100 Hz. In some embodiments, the signal from the on- body device may be aggregated and / or downsampled to a lower frequency in order to reduce the computational burden of subsequent steps of method 24. For example, the signal may be downsampled to 50, 25, 20, 15, 10, 5 Hz, or any other appropriate frequency. At step 28, the activity signal resulting from step 26 is filtered using a Gaussian filtering technique as described below. The filtered activity signal from step 28 is then decomposed using Empirical Mode Decomposition (“EMD”) at step 30 to permit the identification and extraction of an Intrinsic Mode Function (“IMF”) as is further described below. At step 32, peaks of the IMF are identified. At step 33, the Instantaneous Energy (“IE”) of the IMF is then calculated, and those identified peaks having lEs that fall below a predetermined IE threshold are discarded or removed at step 34. Finally, at step 36 actual steps are counted using the peaks that exceed the predetermined IE threshold.

[0035] Although the on-body device may also use gyroscopes, the example provided in this description assumes that the sensor of the on-body device is a three-axis accelerometer consisting of three accelerometers mounted in a block formation, each providing acceleration measurements along one of the three axes. The norm of the X, Y and Z components of the overall activity signal is computed at step 26 by squaring the value of each axis component for each sample, adding the squared values together, then optionally computing the square root of the result. In other words, the norm of the components is derived by computing the sum of the squares of the component signals, and then optionally computing the square root of the sum. The output signal of step 26 is therefore a one-dimensional time seriessignal with varying amplitude and frequency over time. This signal is used as the basis for the remaining steps of the method 24 described herein and is referred to as “the activity signal norm.”

[0036] At step 28, the activity signal norm from step 26 is filtered using a Gaussian filtering technique employing the Weierstrass transform, which provides a smoothed version of the activity signal norm by averaging the values of f, weighted with a Gaussian centered at x. One exemplary function F used in this filtering technique is defined as follows:............ v 4wherein the convolution of f is with the Gaussian function 1 / (V4 nt) eA(— (xA2) / 4). Examples of an activity signal norm and a filtered activity signal are provided in FIG. 4. The activity signal norm 38 is a jagged signal as shown. The filtered activity signal 40 is a smoothed function, each point of which is a weighted average of all of the points in the original function. The weight used is determined by a Gaussian distribution.

[0037] After filtering in the manner described above, the filtered activity signal 40 is subject to decomposition at step 30 using EMD. One method for using EMD to identify an IMF is generally depicted in simplified form in FIG. 5. The initial signal, x(t), is the filtered activity signal 40 as shown in FIG. 6. At step 42, the filtered activity signal 40 is analysed to extract the local extrema of the filtered activity signal 40. These extrema are shown in FIG. 6 as maxima 44 (only three labelled) and minima 46 (only three labelled). At step 48, a spline is fitted to the local maxima according to conventional curve fitting techniques to get the upper envelope signal 50 as depicted in FIG. 6. At step 52, a spline is fitted to the local minima according to conventional curve fitting techniques to get the lower envelop signal 54 as depicted in FIG. 6. At step 56, a mean signal is computed using the values of the upper envelope signal 50 and the lower envelope signal 54 over time. Thisaveraging function (labelled 58 in FIG. 6) is subtracted from the original filtered activity signal 40 (i.e. , x(t)) at the difference junction 60. The output or difference signal, Xi(t), is evaluated to determine whether it meets the criteria for an IMF, at which point the signal is subject to further processing as is described below. If the difference signal is not an IMF, then it is used as the input signal for step 42 in an iterative sifting process as indicated by step 62 and further described below.

[0038] A more detailed flow chart depicting a method 64 for implementing signal decomposition using EMD step 30 is depicted in FIG. 7. The method 64 begins at step 66 with the filtered activity signal from step 28 of FIG. 3 (after Gaussian filtering) as the input signal, x(t). At step 68, an index variable k is initialized and set to 0. Also at step 68, a placeholder function hk-n(t) is set equal to the input signal x(t). Since at step 68, the index variable k is initialized to 0, the placeholder function hk+i(t) is hi(t). At step 70, k is set equal to k + 1 (i.e., at this first iteration, k is set equal to 7). At step 72, the minima points 46 (FIG. 6) and the maxima points 44 (FIG. 6) of hi (t) (i.e., the local extrema) are extracted. Local extrema (i.e., local minima or local maxima) of hi(t) may be identified using derivatives, wavelet analysis, convolutions, and / or any other conventional technique known to the person of ordinary skill in the art. At step 74, a maxima spline 50 (FIG. 6) is curve fitted to the maxima points 44 and a minima spline 54 (FIG. 6) is curve fitted to the minima points 46. At step 76, the mean m(t) (i.e., line 58 in FIG. 6) of the set of data values corresponding to the maxima spline and the minima spline is determined. The mean m(t) (line 58) may be derived by averaging, for every time point, the values of the maxima spline and the minima spline. Given that during this iteration k = 1, at step 78 hk(t) is hi(t) and hk+i(t) is h2(t). Thus, at step 78 h2(t) is set equal to hi(t) - m(t). In other words, hz(t) is the difference between the mean extrema 58 and the original filtered activity signal 40. At step 80, it is determined whether h2(t) is an IMF as is further described below.

[0039] FIG. 8 depicts the signals described above during this first iteration of the method 64. As the signals in FIG. 8 simply provide an example of decomposition of a different filtered activity signal than that shown in FIG. 6, the same reference designations will be used but with an apostrophe. Thus, FIG. 8 shows the original filtered activity signal 40’, the maxima points 44’, the minima points 46’, the maxima spline 50’, the minima spline 54’ and the mean 58’. As indicated above, at step 78 of FIG. 7, the activity signal h?(t) is the difference between the mean extrema 58’ and the original filtered activity signal 40’. FIG. 9A shows h2(t) as signal 402’ which may be used for the next iteration of the method 64 if h2(t) is not determined to be an IMF at step 80.

[0040] In one embodiment of the present disclosure, an IMF is defined according to the criteria 84 depicted in FIG. 7, i.e. , an IMF is a signal that (1 ) has a number of extrema (e.g., maxima points 44’ plus minima points 46’) that is either equal to the number of zero crossings of the signal or differs from the number of zero crossings by at most one, and (2) a mean signal calculated by averaging a spline fitted to local maxima of the signal and a spline fitted to local minima of the signal is less than a threshold value at all points. Intuitively, condition (2) indicates that a signal that is an IMF has maxima and minima that define an envelope that is essentially symmetric about the horizontal X-axis. In certain exemplary embodiments, the threshold value may be user-defined and very small, such as but not limited to 0.0001 or 0.00001 . In some embodiments, the threshold value may be set and / or adjusted by a user depending on the user’s preferred balance of accuracy vs. computational efficiency. A lower threshold would result in a signal that is closer to a true IMF, while a higher threshold would result in an algorithm that requires fewer iterations and is therefore more computationally efficient. In this example, the signal 4O2’ of FIG. 9A has 34 extrema and 34 zero crossings and so satisfies condition (1 ). However, the mean value (682’) of the maxima spline 5O2’ and the minima spline 542’ is not less than the threshold value at all points (i.e., the envelopedefined by the maxima and minima spline is not essentially symmetric about the X- axis). Accordingly, the method 64 returns from step 80 back to step 70 of FIG. 7 to begin another iteration.

[0041] At step 70 of this next (second) iteration, k is incremented by 1 , so k now equals 2. Thus, at step 72, the reference to hk(t) now refers to h2(t), which is the signal 402’ of FIG. 9A resulting from the first iteration. At step 72, the minima points 462’ (FIG. 9A) and the maxima points 442’ (FIG. 9A) of h2(t) (i.e. , the local extrema) are extracted as described above. At step 74, a maxima spline 5O2’ (FIG. 9A) is curve fitted to the maxima points 442’ and a minima spline 542’ (FIG. 9A) is curve fitted to the minima points 462’. At step 76, the mean signal, m, 582’ (FIG. 9A) of the set of data values corresponding to the maxima spline 5O2’ and the minima spline 542’, m, is determined. In some embodiments, since the local extrema (462’, 442’), maxima spline 5O2’, minima spline 542’, and mean signal 582’ were all previously determined as part of step 80 in the previous (first) iteration, steps 72, 74, and 76 may be accomplished by merely referencing or retrieving some or all of these previously-determined points and / or signals from memory. In some embodiments, for all iterations of method 64 after the first iteration (i.e., for the second and higher iterations), steps 72 and 74 may be omitted entirely, and step 76 may be modified to merely “retrieve mean m(t) of upper and lower envelopes determined from previous iteration.” Given that during this iteration k - 2, at step 78 hk(t) is h2(t) and hi+i(t) is ti3(t). Thus, at step 78, hs(t) is set equal to hs(t) - m. In other words, hs(t) is the difference between the mean extrema 582’ and the signal 4O2’ resulting from the first iteration. This hs(t) is shown as signal 40s’ in FIG. 9B. At step 80 of FIG. 7 it is determined whether hs(t) (40a’ of FIG. 9B) is an IMF.

[0042] In this example, the filtered activity signal 40a’ of FIG. 9B has 34 extrema and 34 zero crossings, but the mean value of the envelope defined by the maxima spline 503 and the minima spline 54s’ is not less than the threshold value atall time points (i.e. , the envelope is not essentially symmetric about the X-axis). Accordingly, the method 64 returns from step 80 back to step 70 of FIG. 7.

[0043] At step 70 of this next (third) iteration, k is incremented by 1 , so k now equals 3. Thus, at step 72 if this next iteration, the reference to hk(t) now refers to tuft), which is the signal 40a’ of FIG. 9B resulting from the last (second) iteration. At step 72, the minima points 46a’ (FIG. 9B) and the maxima points 44a’ (FIG. 9B) of tuft) (i.e., the local extrema) are extracted by identifying the inflection points as described above. At step 74, a maxima spline 50a’ (FIG. 9B) is curve fitted to the maxima points 44a’ and a minima spline 54a’ (FIG. 9B) is curve fitted to the minima points 46a’. At step 76, the mean m, 58a’ (FIG. 9B) of the set of data values corresponding to the maxima spline 50a’ and the minima spline 54a’, m, is determined. Again, since the local extrema (46a’, 44a’), maxima spline 50s’, minima spline 54a’, and mean signal 58a’ were all previously determined as part of step 80 in the previous (second) iteration, steps 72, 74, and 76 in the current (third) iteration may be accomplished in some embodiments by merely referencing or retrieving these previously-determined points and / or signals from memory. Again, for some embodiments, for all iterations of method 64 after the first iteration (i.e., for the second and higher iterations), steps 72 and 74 may be omitted entirely, and step 76 may be modified to merely “retrieve mean mft) of upper and lower envelopes determined from previous iteration.” Given that during this iteration k - 3, at step 78 hk(t) is tuft) and hk+i(t) is tuft). Thus, at step 78, tuft) is set equal to tuft) - m. In other words, tuft) is the difference between the mean extrema 583’ and the signal 4O3’ resulting from the last iteration. FIG. 9C shows tuft) as the signal 404 for the next iteration of the method 64. At step 80 of FIG. 7 it is determined whether tuft) (404 of FIG. 9C) is an IMF.

[0044] In this example, the filtered activity signal 404’ of FIG. 9C has 34 extrema and 34 zero crossings, but the mean value of the envelope defined by the maxima spline 504 and the minima spline 544’ is not less than the threshold value atall points (i.e. , the envelope is not essentially symmetric about the X-axis). Accordingly, the method 64 returns from step 80 back to step 70 of FIG. 7.

[0045] The fourth iteration of the above-described signal decomposition using EMD method 64 is depicted in FIG. 9D. As the envelope defined by the maxima spline 50s’ and the minima spline 54s’ of the signal 40s’ does not have a mean value that is less than the threshold value at all points, the method 64 again returns from step 80 back to step 70 of FIG. 7. The process repeats as described previously for multiple iterations until an IMF is found. For example, FIG. 9E depicts the results of the ninth iteration of the method 64. After this ninth iteration, the filtered activity signal 40 ’ satisfies both criteria for an IMF - the number of extrema and the number of zero crossings are equal and the mean value of the envelop defined by the maxima spline 5Oio’ and the minima spline 54io’ is less than the threshold value at all points. Accordingly, the filtered activity signal 4Oio’ shown in FIG. 9F is stored at step 82 as an IMF of the method 64 (FIG. 7).

[0046] It should be appreciated that the IMF identified and used to count steps in the methods disclosed herein is a first-order IMF, also known as a “first” IMF to persons of ordinary skill in the art. EMD may be used to also derive higher-order IMFs, e.g., a “second”, “third”, “fourth” IMF, etc. Higher-order IMFs generally are lower frequency signals (i.e., have lower energy at higher frequencies) compared to lower-order IMFs, and as a result appear smoother and less jagged when graphed over time compared to lower-order IMFs. In other words, the “second” IMF is generally lower in frequency compared to the “first” IMF, and the “third” IMF is lower in frequency compared to the second IMF, etc. The sum of all IMFs, including the “first” IMF as well as all higher-order IMFs, would approximate the signal from which all said IMFs were derived using EMD. However, such higher-order IMFs are not used in the methods and techniques for counting steps described herein. Since steps generally appear as sharp and jagged peaks in the activity signal, such steps are easiest to detect in the first-order IMF, which has the highest frequency contentcompared to higher-order IMFs (which tend to smooth over and reduce the prominence of such sharp signal peaks). Consequently, in some embodiments, the presently disclosed techniques for counting steps need not compute or use higher- order IMFs.

[0047] Returning to application of the principles of the present disclosure to step counting, after performing the signal decomposition process described above to identify an IMF, the instantaneous energy (“IE”) of the IMF is determined as described below. An example of the results of the process up until this point is shown in FIG. 10. From left to right, FIG. 10 depicts the norm 108 (i.e., the square root of the sum of the squares) of the X, Y and Z components of the output of a three-axis accelerometer, the filtered activity signal norm 110 after applying Gaussian filtering, and the IMF 112 extracted using the EMD method 64 described above.

[0048] Referring now to FIG. 11 , the IMF 112 of FIG. 12 is reproduced in the upper left portion of the figure. This IMF 112 is then analysed to detect local maxima in the peak detection step 32 of FIG. 3. In certain embodiments of the present disclosure, the peaks 114 are identified by first constructing an analytic signal of the IMF 112. The analytic signal may be derived from the IMF 112 by using a Hilbert transform 118 as represented in FIG. 12. The Hilbert transform is a specific singular integral that takes a function, such as the IMF 112, which is a real variable, and produces another function of the real variable. In FIG. 12, the IMF 112 is represented as the input function u(t). The Hilbert transform is a convolution of the function u(t) with the function h(t) = 1 / (nt) and is defined using the Cauchy principal value, PV. Specifically, the Hilbert transform of the function u(t) is as set forth below:The analytic signal is the sum of the real components, u(t) and the imaginary components iv(t), denoted X(t) in FIG. 12.

[0050] The instantaneous energy, IE, of the analytic signal is computed by calculating the norm of X(t) according to the equation IE[u(t)] = IX(t)l2. In other words, the IE may be derived from the sum of the squares of the real component u(t) and the imaginary component iv(t). Next, at step 34 of the method 24 of FIG. 3, the IE values of the peaks 114 are compared to a predetermined IE threshold such as the threshold 120 shown in the lower left portion of FIG. 11. In the example shown, the threshold 120 is 0.005. Also at step 34, the peaks 114 having IE values that are below the peak IE threshold 120 are discarded. The remaining peaks 114 are mapped back onto the IMF 112 and are identified as valid peaks 122 as shown in the right-hand portion of FIG. 11 . These remaining peaks 114 are counted as valid steps.

[0051] Table 1 below shows experimental results of the accuracy of the step count achieved using the principles of the present disclosure relative to the Verisense Step Count algorithm for the three different types of signals obtained from the benchmark Clemson Dataset. The Clemson Dataset is a large dataset of raw accelerometer data which is marked with the time occurrences of all steps to permit pedometer algorithms to use a common dataset for evaluation against a gold standard. The Verisense Step Count algorithm is an open-source step counting algorithm created by Shimmer Research. The accuracies listed below are computed as 1 - (ABS (algorithm counted steps - actual steps) ) / actual steps.Table 1

[0052] As shown in Table 1 , the method of step counting according to the present disclosure provided substantial accuracy improvements relative to the Verisense algorithm for all three types of input motion signals, with the most pronounced improvement for the irregular signal type (i.e., almost 21 %).

[0053] Referring now to FIG. 13, a system 200 according to the present disclosure is shown in block diagram form. The system 200 generally includes an on-body device 202, at least remote computing device 204 and a network 206 connecting the on-body device 202 and the remote computing device 204. The on- body device 202 includes an accelerometer 208, at least one processor 210 in electrical communication with the accelerometer 208, at least one memory device 212 in electrical communication with the processor 210, and a transceiver 216 in electrical communication with the processor 210. Although in this figure only one on- body device 202 is shown, in some embodiments, system 200 may include a plurality of on-body devices, wherein each device is provided to and configured to be worn by a separate clinical research subject.

[0054] The remote computing device 204 includes at least one processor 218, at least one memory device 220 in electrical communication with the processor 218, a transceiver 222 in electrical communication with the processor 218, and a display 224 in electrical communication with the processor 218. The remote computing device 204 may be a desk top computer, a laptop computer, a tablet, a smartphone, or any other suitable computing device. Although the remote computing device 204 is shown as a single device in FIG. 13, in some cases the remote computing device 204 may comprise multiple computing devices working together to implement the functions described herein. This may be the case, for instance, if the remote computing device is a computing cluster or cloud of networked processing devices.

[0055] The network 206 may be any suitable network for communicating data from the on-body device 202 to the remote computing device 204. In certain embodiments, the network 206 may include one or more networks, including any of a Local Area Network, a Metropolitan Area Network, a Wide Area Network, a wireless network and an Inter Network such as the internet. The network 206 may include a cloud environment comprising virtual servers, storage and / or processors hosted and / or maintained by a service provider. The network 206 may include one or more intermediate devices and / or communication links owned, maintained, and / or hosted by entities separate from the clinical research subjects or researchers conducting the clinical research. In other embodiments, the network 206 is omitted and the on-body device 202 communicates directly with the remote computing device 204 through a direct wireless link using a wireless network protocol (Wi-Fi) or other type of wireless technology such as, but not limited to, Bluetooth®, Zigbee, Z- Wave, etc.

[0056] As indicated above, the accelerometer 208 of the on-body device 202 may be a three-axis accelerometer consisting of three accelerometers mounted in a block formation, each providing acceleration measurements along one of the three axes (i.e. , configured to output an accelerometer signal including an X, Y and Z component). Alternatively, in some embodiments, a gyroscope may be used in lieu of or in addition to the accelerometer 208. The processor 210 of the on-body device 202 may include one or more processors, and one or more of the functions described above may be distributed among the one or more processors. The memory device 212 of the on-body device 202 may include volatile or non-volatile memory, random access memory, read only memory, etc. The transceiver 216 of the on-body device 202 may include a transmitter and a receiver to provide two-way communication between the on-body device 202 and the remote computing device 204 via the network 206. In other embodiments, the transceiver 216 is simply atransmitter configured to provide data to the remote computing device 204 via the network 206.

[0057] Similarly, the processor 218 of the remote computing device 204 may include one or more processors, and one or more of the functions described above may be distributed among the one or more processors. The memory device 220 of the remote computing device 204 may include volatile or non-volatile memory, random access memory, read only memory, etc. The display 224 may be any suitable display for providing a visual representation of information to the user of the remote computing device 204, including input / output displays such as touch screen displays which function as a graphical user interface. The transceiver 222 of the remote computing device 204 may include a transmitter and a receiver to provide two-way communication between the remote computing device 204 and the on-body device 202 via the network 206. In other embodiments, the transceiver 222 is simply a receiver configured to receive data from the on-body device 202 via the network 206.

[0058] In operation of one embodiment of the present disclosure, the processor 210 of the on-body device 202 performs the various functions described above for counting steps. The processor 210 may execute non-transitory computer- readable instructions stored in the memory device 212, which cause the processor 210 to perform the various functions described above. The processor 210 may also store the results of one or more of the computation functions described in the memory device 212. The processor 210 may cause the transceiver 216 to transmit the results of the method 24 (FIG. 3) as the results are computed.

[0059] In certain embodiments, the processor 218 of the remote computing device 204 may analyze the counted steps to determine the efficacy of a drug provided to the user of the on-body device 202. In such an embodiment, the processor 218 may compare the number of steps counted by the on-body device202 of one or more users who received a drug to a number of steps counted by the on-body device of one or more other users who received a placebo.

[0060] In another embodiment, the processor 210 of the on-body device 202 receives the activity signal from the accelerometer 208 and causes the transceiver 216 to transmit data representing the activity signal to the remote computing device 204 via the network 206. Such data may be transmitted by the on-body device 202 to the network 206 at various frequencies. In some embodiments, data may be transmitted in substantially real time, e.g., every few milliseconds or seconds. In some embodiments, data may be transmitted at regular intervals, e.g., once an hour, once every three hours, or once a day. In some embodiments, data may be held by the on-body device 202 until a certain quantity of data has been collected, at which point the collected data may be sent on to the network 206 in a single batch - since the rate at which data may be collected may vary depending on how a user uses the on-body device (or how often the user wears the device), the rate at which collected data is sent on to the network 206 may also vary accordingly. In some cases, the activity signal from the accelerometer 208 may be aggregated and / or downsampled to compress the activity signal before it is sent to the remote computing device 204. For example, the activity signal may be downsampled from a frequency of 100 Hz or more 50, 25, 20, 15, or 10 Hz. In such embodiments, the processor 218 of the remote computing device 204 performs the various functions described above for counting steps based on the activity signal received from the on-body device 202 via the network 206.

[0061] In the context of a clinical trial or study, all data collected from on-body devices 202 may be collected and held within the memory device 220 of the remote computing device 204 until the conclusion of the clinical research. In some embodiments, only after the clinical trial / study has concluded may the collected data be analysed to count steps. Since clinical trials / studies may involve data collected from dozens or hundreds of subjects over a period of multiple weeks, theamount of data to be analysed at the end of the trial may be voluminous. To analyse the data efficiently, the collected data may be divided into multiple subsets, e.g., by subject, by day, or by subject-day. Each subset of data may then be analysed in parallel according to the various functions described above, e.g., in a high- performance computing cluster or multiple processing devices operating in parallel. Alternatively, the data may be analysed while the trial I study is ongoing at regular intervals. For instance, the collected data may be analysed to derive step counts once a day or once a week.

[0062] Any directional references used with respect to any of the figures, such as right or left, up or down, or top or bottom, are intended for convenience of description, and do not limit the present disclosure or any of its components to any particular positional or spatial orientation. Additionally, any reference to rotation in a clockwise direction or a counter-clockwise direction is simply illustrative. Any such rotation may be implemented in the reverse direction as that described herein.

[0063] Although the foregoing text sets forth a detailed description of embodiments of the disclosure, it should be understood that the legal scope of the invention is defined by the words of the claims set forth at the end of this patent and equivalents. The detailed description is to be construed as exemplary only and does not describe every possible embodiment. Numerous alternative embodiments may be implemented, using either current technology or technology developed after the filing date of this patent, which would still fall within the scope of the claims.

[0064] The following additional considerations apply to the foregoing description. Throughout this specification, plural instances may implement components, operations, or structures described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations may be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components inexample configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.

[0065] In various embodiments, a hardware module may be implemented mechanically or electronically. For example, a hardware module may comprise dedicated circuitry or logic that is permanently configured (e.g., as a special-purpose processor, such as a field programmable gate array (FPGA) or an applicationspecific integrated circuit (ASIC)) to perform certain operations. A hardware module may also comprise programmable logic or circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations. It will be appreciated that the decision to implement a hardware module mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0066] Accordingly, the term “hardware module” should be understood to encompass a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering embodiments in which hardware modules are temporarily configured (e.g., programmed), each of the hardware modules need not be configured or instantiated at any one instance in time. For example, where the hardware modules comprise a general-purpose processor configured using software, the general-purpose processor may be configured as respective different hardware modules at different times. Software may accordingly configure a processor, for example, to constitute a particular hardware module at one instance of time and to constitute a different hardware module at a different instance of time.

[0067] Hardware modules may provide information to, and receive information from, other hardware modules. Accordingly, the described hardware modules may be regarded as being communicatively coupled. Where multiple of such hardware modules exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) that connect the hardware modules. In embodiments in which multiple hardware modules are configured or instantiated at various times, communications between such hardware modules may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware modules have access. For example, one hardware module may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware module may then, at a later time, access the memory device to retrieve and process the stored output. Hardware modules may also initiate communications with input or output devices, and may operate on a resource (e.g., a collection of information).

[0068] The various operations of example methods described herein may be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may, in some example embodiments, comprise processor-implemented modules.

[0069] Similarly, the methods or routines described herein may be at least partially processor-implemented. For example, at least some of the operations of a method may be performed by one or more processors or processor-implemented hardware modules. The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the processor or processors may be located in a single location (e.g., within a homeenvironment, an office environment or as a server farm), while in other embodiments the processors may be distributed across a number of locations.

[0070] The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the one or more processors or processor-implemented modules may be located in a single device or geographic location (e.g., within a home environment, an office environment, or a server farm). In other example embodiments, the one or more processors or processor-implemented modules may be distributed across a number of devices or geographic locations.

[0071] In addition, the aspects and functionalities described herein may operate over distributed systems (e.g., cloud-based computing systems and / or network-based computing systems), where application functionality, memory, data storage and retrieval and various processing functions may be operated remotely from each other over a distributed computing network, such as the Internet or an intranet. User interfaces and information of various types may be displayed via onboard computing device displays or via remote display units associated with one or more computing devices. For example, user interfaces and information of various types may be displayed and interacted with on a wall surface onto which user interfaces and information of various types are projected. Interaction with the multitude of computing systems with which aspects of the invention may be practiced include, keystroke entry, touch screen entry, voice or other audio entry, gesture entry where an associated computing device is equipped with detection (e.g., camera) functionality for capturing and interpreting user gestures for controlling the functionality of the computing device, and the like.

[0072] Unless specifically stated otherwise, use herein of words such as “processing,” “computing,” “calculating,” “determining,” “presenting,” “displaying,” or the like may refer to actions or processes of a machine (e.g., a computer) thatmanipulates or transforms data represented as physical (e.g., electronic, magnetic, or optical) quantities within one or more memories (e.g., volatile memory, nonvolatile memory, or a combination thereof), registers, or other machine components that receive, store, transmit, or display information.

[0073] As used herein any reference to “one embodiment” or “an embodiment” means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

[0074] Some embodiments may be described using the expression “coupled” and “connected” along with their derivatives. For example, some embodiments may be described using the term “coupled” to indicate that two or more elements are in direct physical or electrical contact. The term “coupled,” however, may also mean that two or more elements are not in direct contact with each other, but yet still cooperate or interact with each other. The embodiments are not limited in this context.

[0075] Additionally, some embodiments may be described using the expression “communicatively coupled," which may mean (a) integrated into a single housing, (b) coupled using wires, or (c) coupled wirelessly (i.e. , passing data I commands back and forth wirelessly) in various embodiments.

[0076] As used herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having” or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0077] In addition, use of the “a” or “an” are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give a general sense of the description. This description, and the claims thatfollow, should be read to include one or at least one and the singular also includes the plural unless it is obvious that it is meant otherwise.

[0078] The patent claims at the end of this patent application are not intended to be construed under 35 U.S.C. § 112(f) unless traditional means-plus-function language is expressly recited, such as “means for” or “step for” language being explicitly recited in the claim(s).

[0079] Multiple aspects are disclosed, which include, but are not limited to, the following aspects:

[0080] 1 . A system for enhanced analysis of activity signals from on-body devices for tracking activity of a plurality of clinical research subjects, comprising: a plurality of on-body devices each configured to be worn by a separate clinical research subject of the plurality of clinical research subjects, each device including at least one of an accelerometer and a gyroscope configured to generate an activity signal based on body movement of the subject wearing said device; and a remote computing device including at least one processor and at least one memory device, wherein: the remote computing device is configured to receive the activity signals from each on-body device via a network and store the received signals in the at least one memory device, and the at least one processor is configured to execute computer-readable instructions stored in the at least one memory device to, for each respective received activity signal: compute an activity signal norm from the respective received activity signal; filter the activity signal norm to produce a filtered activity signal; decompose the filtered activity signal to identify an Intrinsic Mode Function (“IMF”); identify peaks of the IMF; compute an instantaneous energy of the IMF; and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps taken by the clinical research subject wearing the on-body device from which the respective received activity signal was received.

[0081] 2. The system of aspect 1 , wherein the at least one processor is further configured to execute the computer-readable instructions to compare the countedsteps derived from activity signals for a first subset of the plurality of clinical research subjects to the counted steps derived from activity signals for a second subset of the plurality of clinical research subjects.

[0082] 3. The system of aspect 2, wherein the first subset of clinical research subjects were subjected to a health intervention, whereas the second subset of clinical research subjects were not subjected to the health intervention.

[0083] 4. The system of aspect 3, wherein the at least one processor is configured to execute the computer-readable instructions to evaluate an efficacy of the health intervention at treating a health condition based on the comparison of counted steps.

[0084] 5. The system of any one of aspects 1-3, wherein the identified IMF is a first order IMF.

[0085] 6. The system of any one of aspects 1-5, wherein the at least one processor is further configured to filter the activity signal norm by applying a Gaussian filter to the activity signal norm.

[0086] 7. The system of any one of aspects 1-6, wherein each activity signal is an acceleration signal including an X component, a Y component and a Z component, and computing the activity signal norm includes computing a sum of the X component squared, the Y component squared and the Z component squared of the respective received activity signal.

[0087] 8. The system of any one of aspects 1-7, wherein the at least one processor is further configured to downsample each respective received activity signal before computing the activity signal norm.

[0088] 9. The system of any one of aspects 1-8, wherein decomposing the filtered activity signal includes: analyzing the filtered activity signal to identify local maxima and local minima; fitting a first spline to the local maxima and a second spline to the local minima; calculating a mean signal from values of the first splineand the second spline; computing a difference signal between the mean signal and the filtered activity signal; and determining whether the difference signal is an IMF.

[0089] 10. The system of aspect 9, wherein determining whether the difference signal is an IMF includes: determining whether a number of extrema of the difference signal is equal to a number of zero crossings of the difference signal or differs from the number of zero crossings by at most one; determining whether at all points, a second mean signal calculated by averaging a third spline fitted to local maxima of the difference signal and a fourth spline fitted to local minima of the difference signal is less than a threshold value.

[0090] 11 . The system of any one of aspects 9-10, wherein the at least one processor is further configured to execute the computer-readable instructions to: set the filtered activity signal equal to the difference signal upon determining that the difference signal is not an IMF; and repeat the analyzing, fitting, calculating, computing and determining steps of aspect 9 until the difference signal is determined to be an IMF.

[0091] 12. The system of any one of aspects 1-11 , wherein the at least one processor is further configured to execute the computer-readable instructions to: construct an analytic signal of the IMF, the analytic signal being a sum of real components and imaginary components of a Hilbert transform of the IMF; and compute the instantaneous energy of the IMF by computing a norm of the analytic signal.

[0092] 13. A method for conducting clinical research for evaluating efficacy of a health intervention in treating a health condition using enhanced analysis of activity signals from on-body devices, the method comprising: providing an on-body device to each subject of a plurality of subjects, each subject being a participant in clinical research for the health intervention, and each on-body device including at least one of an accelerometer and a gyroscope configured to derive an activity signal from body movement of the subject to which the device was provided; subjecting a firstsubset of the plurality of subjects to the health intervention, wherein a second subset of the plurality of subjects is not subjected to the health intervention; receiving at a computing device a plurality of activity signals communicated to the computing device from the plurality of on-body devices via a network, wherein a first subset of the plurality of activity signals is from on-body devices of subjects within the first subset of subjects and a second subset of the plurality of activity signals is from on- body devices of subjects within the second subset of subjects; for each respective activity signal in the plurality of received activity signals: computing an activity signal norm from the respective activity signal, filtering the activity signal norm to produce a filtered activity signal, decomposing the filtered activity signal to identify an Intrinsic Mode Function (“IMF”), identifying peaks of the IMF, computing an instantaneous energy of the IMF, and counting peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps; comparing (1 ) the counted steps derived from activity signals received from on-body devices provided to the first subset of subjects to (2) the counted steps derived from activity signals received from on-body devices provided to the second subset of subjects; and computing a measure indicative of the efficacy of the health intervention based on the comparison.

[0093] 14. The method of aspect 13, wherein the identified IMF is a first orderIMF.

[0094] 15. The method of any one of aspects 13-14, wherein the health intervention is a medication and the health condition is a disease.

[0095] 16. The method of any one of aspects 13-15, wherein the comparing step comprises: calculating a first average number of counted steps derived from on- body devices provided to the first subset of subjects; calculating a second average number of counted steps derived from on-body devices provided to the second subset of subjects; and calculating a difference between the first average and the second average.

[0096] 17. The method of any one of aspects 13-16, wherein, for each respective activity signal in the plurality of received activity signals, filtering the activity signal norm comprises applying a Gaussian filter to the activity signal norm.

[0097] 18. The method of any one of aspects 13-17, wherein each activity signal comprises an acceleration signal including an X component, a Y component, and a Z component, and computing the activity signal norm for each respective activity signal includes computing a sum of the X component squared, the Y component squared and the Z component squared.

[0098] 19. The method of any one of aspects 13-18, further comprising, for each respective activity signal in the plurality of received activity signals, downsampling the respective activity signal before computing the activity signal norm for the respective activity signal.

[0099] 20. The method of any one of aspects 13-19, wherein, for each respective activity signal in the plurality of received activity signals, decomposing the filtered activity signal includes: analyzing the filtered activity signal to identify local maxima and local minima; fitting a first spline to the local maxima and a second spline to the local minima; calculating a mean signal from values of the first spline and the second spline; computing a difference signal between the mean signal and the filtered activity signal; and determining whether the difference signal is an IMF.

[0100] 21 . The method of aspect 20, wherein, for each respective activity signal in the plurality of received activity signals, determining whether the difference signal is an IMF includes: determining whether a number of extrema of the difference signal is equal to a number of zero crossings of the difference signal or differs from the number of zero crossings by at most one; and determining whether at all points, a second mean signal calculated by averaging a third spline fitted to local maxima of the difference signal and a fourth spline fitted to local minima of the difference signal is less than a threshold value.

[0101] 22. The method of any one of aspects 20-21 , further comprising, for each respective activity signal in the plurality of received activity signals: setting the filtered activity signal equal to the difference signal upon determining that the difference signal is not an IMF; and repeating the analyzing, fitting, calculating, computing and determining steps of aspect 20 until the difference signal is determined to be an IMF.

[0102] 23. The method of any one of aspects 13-22, further comprising, for each respective activity signal in the plurality of received activity signals, constructing an analytic signal of the IMF, the analytic signal being a sum of real components and imaginary components of a Hilbert transform of the IMF, and wherein computing the instantaneous energy of the IMF includes computing a norm of the analytic signal.

[0103] 24. A system for enhanced analysis of activity signals, comprising: an on-body device configured to be worn by a subject, said device including at least one of an accelerometer and a gyroscope configured to generate an activity signal based on body movement of the subject; and a remote computing device including at least one processor and at least one memory device, wherein: the remote computing device is configured to receive the activity signals from said on-body device, and the at least one processor is configured to execute computer-readable instructions stored in the at least one memory device to: compute an activity signal norm from the received activity signal, filter the activity signal norm to produce a filtered activity signal, decompose the filtered activity signal to identify an Intrinsic Mode Function (“IMF”), identify peaks of the IMF; compute an instantaneous energy of the IMF, and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps taken by the subject.

[0104] 25. A computer-implemented method for counting steps comprising: receiving, by at least one processor, an activity signal from at least one of a three- axis accelerometer or a gyroscope of an on-body device; computing, by the at leastone processor, an activity signal norm from the activity signal; filtering, by the at least one processor, the activity signal norm to produce a filtered activity signal; decomposing, by the at least one processor, the filtered activity signal to identify an Intrinsic Mode Function (“IMF”); computing, by the at least one processor, an instantaneous energy of the IMF; counting, by the at least one processor, peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps; and outputting the counted steps.

Claims

WHAT IS CLAIMED IS:1 . A system for enhanced analysis of activity signals from on-body devices for tracking activity of a plurality of clinical research subjects, comprising: a plurality of on-body devices each configured to be worn by a separate clinical research subject of the plurality of clinical research subjects, each device including at least one of an accelerometer and a gyroscope configured to generate an activity signal based on body movement of the subject wearing said device; and a remote computing device including at least one processor and at least one memory device, wherein: the remote computing device is configured to receive the activity signals from each on-body device via a network and store the received signals in the at least one memory device, and the at least one processor is configured to execute computer-readable instructions stored in the at least one memory device to, for each respective received activity signal: compute an activity signal norm from the respective received activity signal; filter the activity signal norm to produce a filtered activity signal; decompose the filtered activity signal to identify an Intrinsic Mode Function (“IMF”); identify peaks of the IMF;compute an instantaneous energy of the IMF; and count peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps taken by the clinical research subject wearing the on-body device from which the respective received activity signal was received.

2. The system of claim 1 , wherein the at least one processor is further configured to execute the computer-readable instructions to compare the counted steps derived from activity signals for a first subset of the plurality of clinical research subjects to the counted steps derived from activity signals for a second subset of the plurality of clinical research subjects.

3. The system of claim 2, wherein the first subset of clinical research subjects were subjected to a health intervention, whereas the second subset of clinical research subjects were not subjected to the health intervention.

4. The system of claim 3, wherein the at least one processor is configured to execute the computer-readable instructions to evaluate an efficacy of the health intervention at treating a health condition based on the comparison of counted steps.

5. The system of any one of claims 1 -3, wherein the identified IMF is a first order IMF.

6. The system of any one of claims 1 -5, wherein the at least one processor is further configured to filter the activity signal norm by applying a Gaussian filter to the activity signal norm.

7. The system of any one of claims 1 -6, wherein each activity signal is an acceleration signal including an X component, a Y component and a Z component, and computing the activity signal norm includes computing a sum of the X component squared, the Y component squared and the Z component squared of the respective received activity signal.

8. The system of any one of claims 1 -7, wherein the at least one processor is further configured to downsample each respective received activity signal before computing the activity signal norm.

9. The system of any one of claims 1 -8, wherein decomposing the filtered activity signal includes: analyzing the filtered activity signal to identify local maxima and local minima; fitting a first spline to the local maxima and a second spline to the local minima; calculating a mean signal from values of the first spline and the second spline; computing a difference signal between the mean signal and the filtered activity signal; and determining whether the difference signal is an IMF.

10. The system of claim 9, wherein determining whether the difference signal is an IMF includes: determining whether a number of extrema of the difference signal is equal to a number of zero crossings of the difference signal or differs from the number of zero crossings by at most one;determining whether at all points, a second mean signal calculated by averaging a third spline fitted to local maxima of the difference signal and a fourth spline fitted to local minima of the difference signal is less than a threshold value.11 . The system of any one of claims 9-10, wherein the at least one processor is further configured to execute the computer-readable instructions to: set the filtered activity signal equal to the difference signal upon determining that the difference signal is not an IMF; and repeat the analyzing, fitting, calculating, computing and determining steps of claim 9 until the difference signal is determined to be an IMF.

12. The system of any one of claims 1 -11 , wherein the at least one processor is further configured to execute the computer-readable instructions to: construct an analytic signal of the IMF, the analytic signal being a sum of real components and imaginary components of a Hilbert transform of the IMF; and compute the instantaneous energy of the IMF by computing a norm of the analytic signal.

13. A method for conducting clinical research for evaluating efficacy of a health intervention in treating a health condition using enhanced analysis of activity signals from on-body devices, the method comprising: providing an on-body device to each subject of a plurality of subjects, each subject being a participant in clinical research for the health intervention, and each on-body device including at least one of an accelerometer and a gyroscope configured toderive an activity signal from body movement of the subject to which the device was provided; subjecting a first subset of the plurality of subjects to the health intervention, wherein a second subset of the plurality of subjects is not subjected to the health intervention; receiving at a computing device a plurality of activity signals communicated to the computing device from the plurality of on-body devices via a network, wherein a first subset of the plurality of activity signals is from on-body devices of subjects within the first subset of subjects and a second subset of the plurality of activity signals is from on-body devices of subjects within the second subset of subjects; for each respective activity signal in the plurality of received activity signals: computing an activity signal norm from the respective activity signal, filtering the activity signal norm to produce a filtered activity signal, decomposing the filtered activity signal to identify an Intrinsic Mode Function (“IMF”), identifying peaks of the IMF, computing an instantaneous energy of the IMF, and counting peaks of the IMF having an instantaneous energy that is above a predetermined threshold as steps; comparing (1 ) the counted steps derived from activity signals received from on-body devices provided to the first subset of subjects to (2) the counted steps derived from activity signals received from on-body devices provided to the second subset of subjects; andcomputing a measure indicative of the efficacy of the health intervention based on the comparison.

14. The method of claim 13, wherein the identified IMF is a first order IMF.

15. The method of any one of claims 13-1 , wherein the health intervention is a medication and the health condition is a disease.

16. The method of any one of claims 13-15, wherein the comparing step comprises: calculating a first average number of counted steps derived from on-body devices provided to the first subset of subjects; calculating a second average number of counted steps derived from on-body devices provided to the second subset of subjects; and calculating a difference between the first average and the second average.

17. The method of any one of claims 13-16, wherein, for each respective activity signal in the plurality of received activity signals, filtering the activity signal norm comprises applying a Gaussian filter to the activity signal norm.

18. The method of any one of claims 13-17, wherein each activity signal comprises an acceleration signal including an X component, a Y component, and a Z component, and computing the activity signal norm for each respective activity signal includes computing a sum of the X component squared, the Y component squared and the Z component squared.

19. The method of any one of claims 13-18, further comprising, for each respective activity signal in the plurality of received activity signals, downsampling the respective activity signal before computing the activity signal norm for the respective activity signal.

20. The method of any one of claims 13-19, wherein, for each respective activity signal in the plurality of received activity signals, decomposing the filtered activity signal includes: analyzing the filtered activity signal to identify local maxima and local minima; fitting a first spline to the local maxima and a second spline to the local minima; calculating a mean signal from values of the first spline and the second spline; computing a difference signal between the mean signal and the filtered activity signal; and determining whether the difference signal is an IMF.21 . The method of claim 20, wherein, for each respective activity signal in the plurality of received activity signals, determining whether the difference signal is an IMF includes: determining whether a number of extrema of the difference signal is equal to a number of zero crossings of the difference signal or differs from the number of zero crossings by at most one; and determining whether at all points, a second mean signal calculated by averaging a third spline fitted to local maxima of the difference signal and a fourth spline fitted to local minima of the difference signal is less than a threshold value.

22. The method of any one of claims 20-21 , further comprising, for each respective activity signal in the plurality of received activity signals: setting the filtered activity signal equal to the difference signal upon determining that the difference signal is not an IMF; and repeating the analyzing, fitting, calculating, computing and determining steps of claim 20 until the difference signal is determined to be an IMF.

23. The method of any one of claims 13-22, further comprising, for each respective activity signal in the plurality of received activity signals, constructing an analytic signal of the IMF, the analytic signal being a sum of real components and imaginary components of a Hilbert transform of the IMF, and wherein computing the instantaneous energy of the IMF includes computing a norm of the analytic signal.

Images