Method and system for estimating parameters characterizing manual wheelchair propulsion

Abstract

Disclosed examples relate to a method and system for estimating parameters characterizing manual wheelchair propulsion. In at least one example, the method for estimating parameters characterizing manual wheelchair propulsion involves obtaining motion data over time, wherein the motion data is generated as a subject propels a wheel of the wheelchair; processing the motion data to determine timestamps associated with hand interaction events, wherein, the hand interaction events comprise one or more of a hand contact event and a hand release event, the hand contact event defining a first time instance when a subject's hand contacts the wheel of the wheelchair and initiates a propulsion stroke, and the hand release event defines a second time instance when the subject's hand releases from the wheel and ends the propulsion stroke; and generating an output based on the hand interaction events.

Description

METHOD AND SYSTEM FOR ESTIMATING PARAMETERS CHARACTERIZING MANUAL WHEELCHAIR PROPULSION CROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] The present application claims priority to, and the benefit of, United States Provisional Patent Application No. 63 / 607,940, titled “System and Method for Estimating Parameters Characterizing Manual Wheelchair Propulsion Using Inertial Measurement Units”, filed on December 8, 2023, the entire contents of which are incorporated herein by reference. FIELD

[0002] Disclosed examples generally relate to wheelchairs and manual use of wheelchairs, and more particularly, to a method and system for estimating parameters characterizing manual wheelchair propulsion. BACKGROUND

[0003] Existing methods for the analysis of wheelchair propulsion often involve specialized equipment, such as instrumented wheelchairs and motion capture systems, which are not portable or adaptable to real-world environments. Furthermore, current methods struggle with issues such as the lack of robustness, generalizability, and adaptability to dynamic propulsion speeds and patterns. Hence, there is a need for a portable and highly accurate system and method to estimate the parameters characterizing manual wheelchair propulsion and activities involved while individuals use manual wheelchairs. SUMMARY

[0004] Disclosed examples provide a continuous wavelet transform (CWT)-based method for body-mounted motion sensors to detect temporal events (e.g., hand contact, hand release events, active propulsion, among other examples disclosed herein) and estimating push phase and recovery phasedurations during wheelchair use. The provided examples illustrate that the CWT-based method for hand-mounted motion sensors achieved high accuracy, sensitivity, and precision in the detection of events and estimation of durations regardless of speed, propulsion pattern, and experience level.

[0005] It is believed that disclosed examples can be used as an alternative to the conventional SMARTWheelTMin estimating the temporal parameters which, in turn, (i) enables long-term monitoring of manual wheelchair propulsion and users function assessment, (ii) enables the provision of feedback on wheelchair propulsion to educate the users about cumulative exposure to repetitive tasks and avoid strain injuries, (iii) minimizes the accessibility limitations clinicians and user face to monitor manual wheelchair propulsion, hence providing an inclusive technology to all manual wheelchair users. Additionally, the high sensitivity and precision make it an ideal choice to be used for assistive technologies such as propulsion activated assistive wheelchairs.

[0006] In at least one broad aspect, there is provided a method for estimating parameters characterizing manual wheelchair propulsion, comprising: obtaining motion data over time, wherein the motion data is generated as a subject propels a wheel of the wheelchair; processing the motion data to determine timestamps associated with hand interaction events, wherein, the hand interaction events comprise one or more of a hand contact event and a hand release event, the hand contact event defining a first time instance when a subject’s hand contacts the wheel of the wheelchair and initiates a propulsion stroke, and the hand release event defines a second time instance when the subject’s hand releases from the wheel and ends the propulsion stroke; and generating an output based on the hand interaction events.

[0007] In some examples, the method further comprises, based on the timestamps for the hand interaction events, determining a time interval for a push phase and a recovery phase.

[0008] In some examples, the motion data comprises tri-axial acceleration data, and processing the acceleration data comprises, initially, determining a resultant acceleration data over time.

[0009] In some examples, the processing of the acceleration data is based on a continuous wavelet transform (CWT) technique.

[0010] In some examples, processing of the acceleration data further comprises: applying the CWT technique to the resultant acceleration data to identify one or more temporal regions corresponding to hand interaction events and cycle events; and in respect of each of the temporalregions associated with hand interaction events, analyzing the resultant acceleration data in these temporal regions to determine timestamps for the hand interaction events.

[0011] In some examples, applying the CWT to the resultant acceleration data comprises: segmenting the resultant acceleration data into one or more time window splits; determining CWT coefficients for each time window; for each time window, estimating a scale-independent energy; and based on the scale-independent energy, identifying the one or more temporal regions corresponding to the hand interaction events and the cycle events.

[0012] In some examples, further comprising applying a cadence adaptation mechanism after estimating the scale-independent energy, wherein the cadence adaption mechanism is applied to accommodate for changes in propulsion speed.

[0013] In some examples, analyzing the acceleration data in the temporal regions comprises applying a peak detection method to the resultant acceleration data.

[0014] In some examples, the one or more sensors part of a motion tracking unit.

[0015] In some examples, the motion tracking unit is coupled to one or more of the subject’s hand, wrist, forearm, upper arm and chest.

[0016] In another broad aspect, there is provided a system for estimating parameters characterizing manual wheelchair propulsion, comprising: one or more sensors for generating motion data, wherein the motion data is generated as a subject propels a wheel of a wheelchair; and at least one processor coupled to the one or more sensors, and configured for: processing the motion data to determine timestamps associated with hand interaction events, wherein, the hand interaction events comprise one or more of a hand contact event and a hand release event, the hand contact event defining a first time instance when a subject’s hand contacts the wheel of the wheel chair and initiates a propulsion stroke, and the hand release event defines a second time instance when the subject’s hand releases from the wheel and ends the propulsion stroke; and generating an output based on the hand interaction events.

[0017] In some examples, the at least one processor is further configured for: based on the timestamps for the hand interaction events, determining a time interval for a push phase and a recovery phase.

[0018] In some examples, the motion data comprises tri-axial acceleration data, and processing the acceleration data comprises, initially, determining a resultant acceleration data over time.

[0019] In some examples, the processing of the acceleration data is based on a continuous wavelet transform (CWT) technique.

[0020] In some examples, processing of the acceleration data further comprises the at least one processor being further configured for: applying the CWT technique to the resultant acceleration data to identify one or more temporal regions corresponding to hand interaction events and cycle events; and in respect of each of the temporal regions associated with hand interaction events, analyzing the resultant acceleration data in these temporal regions to determine timestamps for the hand interaction events.

[0021] In some examples, applying the CWT to the resultant acceleration data comprises the at least one processor being further configured for: segmenting the resultant acceleration data into one or more time window splits; determining CWT coefficients for each time window; for each time window, estimating a scale-independent energy; and based on the scale-independent energy, identifying the one or more temporal regions corresponding to the hand interaction events and the cycle events.

[0022] In some examples, the at least one processor being further configured for: applying a cadence adaptation mechanism after estimating the scale-independent energy, wherein the cadence adaption mechanism is applied to accommodate for changes in propulsion speed.

[0023] In some examples, analyzing the acceleration data in the temporal regions comprises applying a peak detection method to the resultant acceleration data.

[0024] In some examples, the one or more sensors form part of a motion tracking unit.

[0025] In some examples, the motion tracking unit is coupled to one or more of the subject’s hand, wrist, forearm, upper arm and chest.

[0026] Other features and advantages of the present application will become apparent from the following detailed description taken together with the accompanying drawings. It should be understood, however, that the detailed description and the specific examples, while indicating preferred embodiments of the application, are given by way of illustration only, since various changesand modifications within the spirit and scope of the application will become apparent to those skilled in the art from this detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] For a better understanding of the various embodiments described herein, and to show more clearly how these various embodiments may be carried into effect, reference will be made, by way of example, to the accompanying drawings which show at least one example embodiment, and which are now described. The drawings are not intended to limit the scope of the teachings described herein.

[0028] FIG. 1A shows an example system for estimating parameters characterizing manual wheelchair propulsion.

[0029] FIG.1B illustrates various placement locations for a motion tracking unit on a subject’s body.

[0030] FIG. 2A illustrates a push phase (a) and a recovery phase (b) during wheelchair propulsion.

[0031] FIG. 2B illustrates various sitting posture variations during manual wheelchair propulsion.

[0032] FIG. 2C illustrates various hand contact variations (i.e., from the relative point of contact to rim point of view), during manual wheelchair propulsion.

[0033] FIG.2D illustrates common wheelchair propulsion patterns.

[0034] FIG. 3A is an example method for estimating parameters characterizing manual wheelchair propulsion.

[0035] FIG.3B is an example method for hand interaction event detection using acceleration data.

[0036] FIG.3C is a visualized process flow for estimating parameters characterizing manual wheelchair propulsion.

[0037] FIG. 4A is a plot of continuous wavelet transform (CWT) coefficients for s ranging from 0 to 500, expanded on the time domain.

[0038] FIG.4B is a plot of CWT coefficients plotted against time.

[0039] FIG.4C is a plot of a CWT scalogram illustrating the CWT coefficients for different scales against time.

[0040] FIG. 4D is a plot of CWT coefficients plotted against time, and highlighting one or more regions of interest.

[0041] FIG.5A is an exemplary readout of an inertial measurement unit (IMU) from a hand- mounted IMU during wheelchair propulsion.

[0042] FIG.5B is a plot of filtered resultant acceleration.

[0043] FIG.5C is a plot of scale-independent energy across scales.

[0044] FIG.5D is the resultant acceleration recorded by the IMU of FIG.5A, during multiple consecutive propulsion cycles for two participants in subplots (a) and (b).

[0045] FIG. 6 is a boxplot illustrating the distribution of mean absolute temporal errors in estimating the push and recovery phases duration and detecting the tcontact and trelease across sex, level of experience, and wheelchair propulsion speed and duration. Each boxplot represents 110 samples that were randomly selected with respect to sex, level of experience, and wheelchair propulsion speed and duration.

[0046] FIG.7 is an example hardware configuration for an example motion tracking unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] Examples herein relate to a method and system for estimating parameters characterizing manual wheelchair propulsion.I. DEFINITIONS

[0048] Any term or expression not expressly defined herein shall have its commonly accepted definition understood by a person skilled in the art. As used herein, the following terms have the following meanings.

[0049] “Continuous wavelet transform (CWT)” refers to a mathematical technique for analyzing signals by decomposing them into scaled and translated wavelets to capture time and frequency information, and is based on using a mother wavelet.

[0050] “Hand contact event” refers to the moment (e.g., time instance) when a user's hand makes contact with the wheelchair push rim during propulsion.

[0051] “Hand release event” refers to the moment (e.g., time instance) when a user's hand disengages from the wheelchair push rim after propulsion.

[0052] “Memory” refers to a non-transitory tangible computer-readable medium for storing information in a format readable by a processor, and / or instructions readable by a processor to implement an algorithm. The term "memory" includes a plurality of physically discrete, operatively connected devices despite use of the term in the singular. Non-limiting types of memory include solid- state, optical, and magnetic computer readable media. Memory may be non-volatile or volatile. Instructions stored by a memory may be based on a plurality of programming languages known in the art, with non-limiting examples including the C, C++, Python ™, MATLAB ™, and Java ™ programming languages. To that end, it will be understood by those of skill in the art that references herein to a computing device (e.g., a motion tracking unit 102 or user device 118) as carrying out a function or acting in a particular way imply that processor is executing instructions (e.g., a software program) stored in memory and possibly transmitting or receiving inputs and outputs via one or more interfaces.

[0053] “Processor” refers to one or more electronic devices that is / are capable of reading and executing instructions stored on a memory to perform operations on data, which may be stored on a memory or provided in a data signal. The term "processor" includes a plurality of physically discrete, operatively connected devices despite use of the term in the singular. Non-limiting examples ofprocessors include devices referred to as microprocessors, microcontrollers, central processing units (CPU), and digital signal processors.

[0054] “Propulsion stroke” refers to the motion or action of applying force to the wheelchair push rim to move the wheelchair forward. The propulsion stroke is generally defined between the hand contact and hand release event, and involves the user’s hand being engaged with the push rim.

[0055] “Real time or near real time” means actions or processes performed either instantaneously after receiving specific inputs, or within a very short timeframe, typically measured in seconds (e.g., within 1-5 seconds). II. GENERAL OVERVIEW

[0056] The use of manual wheelchairs is a primary form of ambulation for individuals with mobility impairments. These individuals rely on manual wheelchairs for their daily activities and overall independence.

[0057] In particular, most wheelchair users utilize manual wheelchairs for mobility and activities of daily living, which means relying heavily on their upper extremities. Upper extremity pain, however, significantly restricts the functions and level of activity of manual wheelchair users and, therefore, overshadows their quality of life and independence. It has been previously shown that manual wheelchair propulsion on level ground and up a ramp, as well as body transfers, exert substantial mechanical stress on the shoulder, potentially leading to adverse upper limb kinematic patterns.

[0058] Despite the well-established link between activities involving a manual wheelchair and shoulder pain, the general understanding of the cumulative exposure of manual wheelchair users to repetitive tasks is constrained, and there are limited methods to qualitatively evaluate such exposure.

[0059] Previous methods and systems have focused on monitoring the physical activity of manual wheelchair users utilizing wheel-mounted or body-worn sensors. While these systems improve understanding of the overall physical activity in manual wheelchair users, the primary goal of these systems is to estimate high-level parameters describing physical activity and activity types. As a result,there remains a gap in our ability to accurately monitor and analyze specific parameters characterizing manual wheelchair propulsion that might lead to a reduction in the risk of the development of upper extremity pain in the users.

[0060] Additionally, in-lab motion capture cameras have been previously used as reference instruments to monitor the biomechanics of wheelchair propulsion, assess the users’ function and mobility and provide valuable feedback on manual wheelchair propulsion. Despite their potential benefits, motion capture cameras require specialized labs and expertise to operate effectively, making them costly and labor-intensive. As a result, these instruments face significant limitations in terms of accessibility for manual wheelchair users. These limitations have significant implications for the use of motion capture cameras in clinical and daily environment settings. Without access to these systems, clinicians and users may not be able to monitor manual wheelchair users’ overall propulsion patterns, which could lead to increased risk of upper extremity pain, injury, and other complications.

[0061] As a result, an accurate method to monitor manual wheelchair propulsion in a wide range of users is required to provide an in-depth understanding of their cumulative exposure to repetitive tasks. Also, it paves the way toward quantitative evaluation of manual wheelchair propulsion in daily activities in the real world. Such a method can thus contribute to reducing the risk of the development of upper extremity pain in manual wheelchair users.

[0062] In view of the foregoing, examples herein provided for a method and system for estimating parameters characterizing manual wheelchair propulsion.

[0063] As explained herein, the disclosed examples rely on a wearable motion tracking unit. These units are portable and easy to use, and are inclusive to all manual wheelchair users. Further, the motion tracking unit provides accurate and effective motion assessment in applications such as clinical evaluation, athletic performance assessment and rehabilitation.

[0064] In at least one example, the motion tracking unit is used for evaluating various temporal parameters associated with hand interaction events in the context of a subject using a wheelchair.

[0065] Temporal parameters, including the start and end time of push and recovery phases, the push duration and recovery duration, as well as the propulsion speed and push duration to recovery duration ratio for each cycle, are primary measures to monitor manual wheelchair propulsion. Theseparameters provide valuable quantitative insights into the efficiency and effectiveness of manual wheelchair propulsion and can be instrumental for both clinicians and users in assessing and improving the function and mobility of manual wheelchair users.

[0066] More broadly, using the relationship between temporal parameters and injury development, it is possible to offer recommendations for manual wheelchair users. For example, it is strongly recommended to minimize the frequency of repetitive upper limb tasks to reduce the risk of overuse injuries. As well, it also suggested to reduce the frequency of wheelchair use and frequency of strokes during wheelchair propulsion because they are risk factors for strain injury at the shoulder and wrist. It is also recommended to adopt efficient propulsion techniques to decrease the strain on the shoulders, thereby lowering the risk of the development of upper extremity pain and injury. III. EXAMPLE SYSTEM

[0067] FIG. 1A shows an example system 100 for estimating parameters characterizing manual wheelchair propulsion.

[0068] As shown, the system 100 includes a motion tracking unit 102. Motion tracking unit 102 is used to monitor and generate acceleration values and / or other motion values (e.g., gyroscope values) as the subject 104 uses a wheelchair 106.

[0069] In some example, the motion tracking unit 102 comprises one or more accelerometers for generating tri-axial acceleration data (e.g., a tri-axial accelerometer). The tri-axial acceleration data can involve acceleration values over time in three Cartesian coordinates. In other examples, the motion tracking unit 102 comprises an inertial measurement unit (IMU). As known in the art, the IMU can incorporate a triaxial accelerometer, as well as a tri-axial gyroscope and magnetometer.

[0070] More generally, as explained herein with reference to FIG.7, the motion tracking unit 102 can include a processor 702 coupled to a memory 704 and one or more motion sensors 706. The motion sensors 706 can include a tri-axial accelerometer, a tri-axial gyroscope, an inertial measurement unit (IMU) or generally any other sensor configured for sensing motion. Processor 702 may also couple to a display interface 708 and / or a communication interface 710.

[0071] Continuing with reference to FIG.1A, the motion tracking unit 102 is couplable to any portion of a subject’s body, wherein the subject 104 is a user of a wheelchair 106. In at least one example, the motion tracking unit 102 is coupled to the subject’s hand 112. The advantage of this coupling location is that the tracking unit 102 is able to more closely monitor the user’s hand contact and release with the chair’s wheel 108. The significance of this monitoring is explained shortly, below.

[0072] In other cases, as shown in FIG.1B, the motion tracking unit 102 may be worn (e.g., couplable) to various other parts of the subjects body, such as their forearm and wrist, upper arm or chest. More generally, the motion tracking unit 102 can be a wearable device.

[0073] Although the motion tracking unit 102 is exemplified as being a single mechanical enclosure, it is possible that the motion tracking unit 102 in fact comprises multiple subunits. For example, each subunit can be a separate enclosure, and can house a different accelerometer. Each subunit can be coupled to (or disposed on) on the same or a different part of the subject’s body.

[0074] In at least one example, the motion tracking unit 102 is able to analyze acceleration data, generated from an accelerometer, to determine one or more hand interaction events.

[0075] A “hand interaction event” includes one or more of a timestamp for a hand contact event (tcontact) and a time stamp for a hand release event (trelease). For example, as shown in FIG.2A, a hand contact event (200b) refers to the time instance when the user’s hand contacts the wheel 108 to initiate a propulsion stroke of the wheel. A propulsion stroke is any force applied by the user to cause the wheel to rotate, e.g., forwardly or rearwardly. Further, a hand release event (200a) refers to the time instance when the user’s hand releases from the wheel after completing a propulsion stroke. In some cases, it is assumed the user’s hand is in contact with the wheel (e.g., push rim) during the interval of the propulsion stroke.

[0076] More broadly, wheelchair propulsion is broken down into the push phase and recovery phase. The hand contact event (tcontact) and release event (trelease) mark the beginning and end of the push phase and the end and beginning of the recovery phase, respectively, thus can be used to identify these phases. Push phase is defined as the duration in which user’s hand is physically in contact with the push-rim to push forward. Recovery phase is defined as the duration in which the hand moves back to the initial position to make contact with the push rim, respectively.

[0077] As shown in FIGs.2B – 2D, monitoring the push and recovery phases is important in a wide variety of contexts including in analyzing different sitting posture variations (FIG.2B), hand contact variations (FIG.2C) and different propulsion patterns (FIG.2D).

[0078] As explained herein, a method based on continuous wavelet transform (CWT) is used to detect the hand contact event (tcontact) and release event (trelease) from the accelerometer readouts obtained during different postures, propulsion techniques and speeds.

[0079] As provided below, the disclosed method detects hand contact and hand release instants (tcontact and trelease) to identify the push and recovery phases, which can then be used for estimating push duration, recovery duration, push duration to recovery duration ratio and cadence.

[0080] Continuing reference to FIG. 1A, the system 100 can also include a user device 118. User device 118 can include any computing device, including a personal computer, a smart phone, a tablet or the like.

[0081] In some examples, user device 118 includes a display interface 120 for displaying various graphical user interfaces (GUIs) and other output data. User device 118 can also include other output interfaces (e.g., audio speakers) for outputting other forms of data (e.g., audio data).

[0082] User device 118 can include a processor coupled to a memory and one or more of a communication interface, the display interface 120 as well as various other types of output interfaces.

[0083] In at least one example, the user device 118 is coupled to the motion tracking unit 102, e.g., via a communication network 150. Communication network 150 can be a wired or wireless network, as known in the art.

[0084] The communicative coupling between user device 118 and motion tracking unit 102 can allow user device 118 to receive various data (raw and / or processed data), from the motion tracking unit 102. Accordingly, in some examples, user device 118 can display various output data (as explained herein), including data about the subject’s use of the wheelchair.IV. EXAMPLE METHODS

[0085] The following is a description of various methods in accordance with the disclosed examples. The disclosed methods in FIGs.3A and 3B can be executed by a processor of one or more (or any combination) of the motion tracking unit 102, the user device 120 and / or any other computing device or server.

[0086] (a.) Method for Estimating Parameters Characterizing Manual Wheelchair Propulsion.

[0087] FIG. 3A shows an example method 300a for estimating parameters characterizing manual wheelchair propulsion.

[0088] At 302a, acceleration data is obtained from the motion tracking unit 102. For example, this can be tri-axial acceleration data, and which can be generated as the subject 104 is using the wheel chair 106. In some instances, the acceleration data is generated while the tracking unit 102 is coupled (e.g., worn on) the user’s hand, and the user is pushing the wheelchair wheel 108.

[0089] In at least one example, to obtain effective acceleration data suitable for the analysis described herein – the X-axis, Y-axis, and Z-axis of each of the corresponding axial accelerometers (or tri-axial accelerometer) are oriented respectively dorsally perpendicular to the hand, medially toward the thumb, and proximally, respectively, to record the hand motion on the push-rim (FIG.1A).

[0090] FIG.5A shows a plot 500a of an exemplary IMU readout for acceleration and angular velocity for a hand-mounted IMU during manual wheelchair propulsion.

[0091] At 304a, the acceleration data is processed to determine timestamps associated with one or more hand interaction events. For example, this includes the timestamps for: (i) when the user’s hand contacts the wheelchair wheel at the start of a propulsion stroke (tcontact) (200b in FIG.2A); and (ii) when the user’s hand releases the wheelchair wheel at the end of a propulsion stroke (trelease) (200a in FIG.2A). In at least one example, as provided below, the hand interaction events are determined by initially determining a resultant acceleration from the tri-axial acceleration data.

[0092] At 306a, one or more motion parameters may be determined based on the timestamp data of the hand interaction events. The motion parameters include one or more of: (i) motion temporalparameters; (ii) identifying a motion type; and (ii) determining various kinetics and kinematics parameters characterizing the motion.

[0093] With respect to (i), as shown in FIG.2A, the motion temporal parameters can include the start and end time of push and recovery phases, the push duration and recovery duration, as well as the propulsion speed and push duration to recovery duration ratio for each cycle.

[0094] In particular, the push phase (200a) is the time duration in which user’s hand is physically in contact with the push-rim to push forward. This is determined based on the difference between (t(i)contact – t(i)release) in a single cycle (i), which also define the start and end times of the push phase.

[0095] The recovery phase (200b) is the time duration in which the hand moves back to the initial position to make contact with the push-rim, respectively. This is determined based on the difference of the release and contact between two consecutive cycles (i) and (i+t) (e.g., t(i)release - t(i+1)contact), which also define the start and end times of the recovery phase.

[0096] In some examples, the push and recovery phase time duration is further used to determine a push duration to recovery duration ratio and cadence.

[0097] With respect to the motion type (ii), this may be determined based on analyzing and extracting various features in the resulting acceleration data.

[0098] The types of features extracted include but are not limited to time-domain features, such as statistical metrics, parameters describing the acceleration peaks, zero-crossings, and temporal patterns, and frequency-domain features, spectral power, Fourier Transform coefficients, signal energy, and entropy. These features are used to classify motion by highlighting unique patterns specific to different motion types. Machine learning models or rule-based algorithms process these patterns to identify the motion of interest.

[0099] With respect to the kinetics and kinematic parameters (iii), this may be determined by extracting secondary temporal parameters and secondary spatial parameters. The secondary temporal parameters include but are not limited to event durations, event timings, and repetition rates about the motion of interest. The secondary spatial parameters include but are not limited to displacement,orientation metrics, and velocity or acceleration profiles. These parameters are aggregated using statistical or mathematical techniques, as known in the art, such as averaging or regression to represent overall patterns for repeated cycles. The aggregated parameters are then translated to explain kinetics and kinematics. This process helps feed biomechanical models to assess physical performance, efficiency, or fatigue during motion activities like manual wheelchair propulsion.

[0100] More generally, the disclosed methods are able to estimate the temporal parameters of wheelchair propulsion using acceleration data. This method is focused on robustness, generalizability, and adaptability with dynamic speeds, different postures, and propelling techniques.

[0101] At 308a, an output is generated, based on the data processing. The disclosure herein is not limited to the form of output generated. For example, the output can be a visual or graphic (or audio) output on a display interface of the motion tracking device 102 and / or the user device 118.

[0102] For instance, the output can be any one of, or any combination of, the time instances for the hand interaction events, including one or more of the hand contact event (tcontact) and hand release event (trelease). It may also include various determined motion parameters, including the push and recovery phase time duration, and the push duration to recovery duration ratio and cadence. In other examples, the output can include one or more other motion parameters, including an identification of the motion type and / or various kinetics and kinematics parameters.

[0103] More generally, disclosed examples enable for estimating the temporal parameters which (i) enables long-term monitoring of manual wheelchair propulsion and users function assessment, (ii) enables the provision of feedback on wheelchair propulsion to educate the users about cumulative exposure to repetitive tasks and avoid strain injuries, (iii) minimizes the accessibility limitations clinicians and user face to monitor manual wheelchair propulsion, hence providing an inclusive technology to all manual wheelchair users. Additionally, the high sensitivity and precision make it an ideal choice to be used for assistive technologies such as propulsion activated assistive wheelchairs.

[0104] (b.) Method for Determining Timestamps for Hand Interaction Events.

[0105] FIG. 3B shows an example method 300b for determining timestamps for hand interaction events, using acceleration data.

[0106] At 302b, the acceleration data can be pre-processed. The pre-processing can involve determining the resultant acceleration from the tri-axial acceleration data. The pre-processing can also involve applying a low pass filter to the resultant acceleration data.

[0107] The resulting acceleration can be determined based on tri-axial acceleration data, using any method known in the art. In at least one example, the resulting acceleration is determinable by calculating the Euclidean norm of the three tri-axial acceleration components, measured by the accelerometer. The resultant acceleration may be calculated at each (or any desired interval of) time sampling instance, such as to produce resultant acceleration data over time.

[0108] In some cases, a low pass filter may also be applied to the resultant acceleration data. In some examples, the low pass filter comprises a 6th-order low-pass Butterworth filter. The cut-off frequency of the filter can be 10 Hz.

[0109] To this end, FIG.5B shows an example plot 500b of filtered resultant acceleration.

[0110] At 304b, the resultant acceleration data is segmented into one or more time windows. For example, the resultant acceleration data is segmented (or split) into “M” overlapping sliding time windows. In some examples, the window size is two (2) seconds. In at least one example, each window is longer than the expected duration for a single propulsion stroke.

[0111] At 306b, for each time window, a CWT decomposition is applied to the resultant acceleration data in that window. In turn, this allows determining corresponding CWT coefficient for each time window split.

[0112] In at least one example, the CWT decomposition is performed using a Morlet mother wavelet (which is a modified version of Daubechies wavelet with increased symmetry). The CWT coefficients are estimated for the scales ranging from 1 to a maximum scaling factor (Smax).

[0113] More broadly, as known in the art, CWT produces a time-frequency decomposition where both short-duration high-frequency and long-duration low-frequency information can becaptured simultaneously using a mother wavelet. In this example, a Morlet wavelet is used as the mother wavelet. The Morlet wavelet can be a complex symmetric sinusoid (i.e., to avoid spectral domain skewness) modulated by a Gaussian.

[0114] The CWT coefficients of a discrete time signal, x(t), are calculated based on Equation (1): ^^ 1 ା^ ^^ − ^^ (1)^^^^^^ ^^, ^^ = ^ ^^(^^)ψ൬^^ ^ ^^^^√^^ ି^where ψ is the and location (i.e., time

[0115] The relationship between the dilation (i.e., scale) and spectral components is inverse. This means that higher (lower) values for dilation that are equivalent to lower (higher) frequencies present more global (local) information about the signal and the patterns.

[0116] FIGs. 4A – 4C illustrate various exemplary plots of CWT coefficients based on an exemplary application, including: (i) CWT coefficients plotted for s ranging from 0 to 500, expanded on the time domain (FIG.4A); (ii) plot of CWT coefficients plotted against time (FIG.4B); and (iii) a plot of CWT scalogram illustrating the CWT coefficients for different scales against time (FIG.4C).

[0117] At 308b – 314b, the time instances corresponding to hand interaction events are determined. The hand interaction events include: (i) hand contact events (i.e., subject’s hand contacts wheelchair wheel at the start of a propulsion stroke), and (ii) hand release event (i.e., subject’s hand releases wheelchair wheel at the end of the propulsion stroke). This is performed using a generalized two-step process:

[0118] At step one (acts 308b – 312b), a general estimation of the temporal region corresponding to the hand interaction events is identified. This general estimation is based on the determined CWT coefficients.

[0119] Subsequently, at step two (act 314b), based on the general estimation, the specific time instance for each hand interaction events is determined. This more specific determination is based onapplying peak detection to the resultant acceleration within the general temporal estimate, determined at step one.

[0120] To this end, it is appreciated that a CWT operation is particularly advantageous in this application because it captures both time and frequency information, and is more robust to amplitude and time changes. This accommodates the cyclical motion in wheelchair propulsion which expresses amplitude peaks in each cycle, but where sometimes these specific peaks are not obvious or distinguishable in the complex resultant acceleration signal (e.g., as shown in FIG. 5D). More specifically, in time instances having slower and less impulsive propulsion, the CWT is able to effectively identify the region of interest. Further, as demonstrated further herein, the exemplified method based on CWT is accurate, and is more robust to speed changes and propulsion changes.

[0121] Accordingly, for this reason, CWT is initially used to determine the general temporal region where the relevant peaks occur (i.e., as amongst many peaks in the resultant acceleration signal), as related to hand interaction events. Once the general temporal region is identified, peak detection is then used to further quantify the exact timestamp for the relevant peak, which expresses a hand interaction event. In this manner, the unique combination of a CWT method followed by acceleration peak detection, enables a more accurate and robust technique to quickly identify hand interaction events in otherwise complex acceleration data. In some examples, this allows applying the method with faster computational processing, enabling real-time or near real-time application.

[0122] More specifically, with continued reference to FIG.3B, at 308b, the CWT coefficients are used to estimate scale-independent energy (^^^) for each time window, in accordance with Equation(2):^^^ = ∑ெି^^ୀ^ |^^^^^^(^^,^^)|ଶ , for s ranging from 1 to Smax (2)wherein ^^^^^^(^^,^^) is a 2D wavelet energy density function that measures the distribution of signal energy. ^^^the peaks accounting for most of the signal energy in the spectral domain (FIG.4C).

[0123] Therefore, the ^^^of the CWT coefficients can be approximated as a combination of two 1D Gaussian distributions, with each Gaussian distribution representing the spectral signal energy ofthe event and cycle, respectively. This is exemplified in plot 500c of FIG. 5C, which shows scale- independent energy determined from various example sample data, each represented by its own line.

[0124] Due to periodic nature of manual wheelchair propulsion, ^^^should peak at two particular scales (^^^and ^^^) (FIG.5C). One represents the cycles (strokes, propulsions), and the other represents the hand interaction events (e.g., contact and hand release events), as shown in Equation(3):^^^ = ∑ெି^^ୀ^ |^^^^^^(^^,^^)|ଶ , for s ranging from 1 to Smax (3)

[0125] At 310b, in some examples, a cadence adaptation mechanism is applied to accommodateThe method of applying the cadence adaptation mechanism is explained in greater detail below.

[0126] At 312b, the temporal signals estimating the regions in which the cycle and hand interaction events occurred are defined as CWT coefficients at these particular scales where the ^^^inthat scale peaks, in accordance with Equations (4) and (5) (see e.g., FIG. 4D).^^^ = [^^^^^^(^^^ , 1),^^^^^^(^^^ , 2), … .. ,^^^^^^(^^^ , ^^), … . ,^^^^^^(^^^,^^)] (4)^^^ = [^^^^^^(^^^, 1), ^^^^^^(^^^, 2), … .. , ^^^^^^(^^^, ^^), … . , ^^^^^^(^^^, ^^)] (5)wherein ^^^^^^(^^, ^^) represents the CWT coefficients at scale ^^ and for ^^ th splits in the signal, ^^^ is thetemporal signal regions – in each time window – where ^^^peaks within the scale ^^^and represents the time period for the cycle event in that window, and ^^^is the temporal signal regions – in each time window – where ^^^peaks within the scale ^^^and represents the time period for a hand interaction event in that window.

[0127] Accordingly, the ^^^peaks within the scale ^^^identifies the CWT coefficients corresponding to the hand contact and release events. Thee ^^^peaks within the scale ^^^can help to indicate the cycle event, which can help also differentiate between the hand contact and release events between cycles.

[0128] In some examples, the ^^^peaks within the scale ^^^(504c) is typically represented as being larger than the ^^^peaks within the scale ^^^(502). Accordingly, the size or width of the peak isdetermined to map it to the correct scale. In some examples, the system compares the two peaks and determines the relative size (e.g., thresholding or area under the curve) to classify the peaks.

[0129] At 314b, the distinct temporal signal representations match the frequency of the event and cycle and are then used to estimate the regions and identify the time instants in which the hand contact and hand release occurred.

[0130] More generally, the distinct temporal signal representation of the hand interaction events (^^^) (Equation (4)) are cross-referenced back to a plot of the timewise acceleration data. Peak detection is then applied in these temporal regions to identify instants with the highest unfiltered resultant acceleration values in the suggested region as tcontact and trelease instants. This is shown, by way of example, in the indicators in plot 500f in FIG.5D.

[0131] To this end, any method of peak detection can be used as known in the art (e.g., thresholding).

[0132] Accordingly, based on this, the time stamps for the hand interaction events are identified in the resulting acceleration data, including the timestamps for tcontactand trelease.

[0133] (iii.) Method for Cadence Adaptation.

[0134] As explained in act 310b (method 300b in FIG. 3B), in some examples, the method 300b may include applying a cadence adaptation mechanism.

[0135] In particular, to account for the alterations in propulsion speed, a cadence adaptation mechanism is implemented that employes the frequency relationship between the event and cycle. More generally, the purpose of the cadence adaptation mechanism is dealing with propulsion speed changes, fine-tuned to decrease false negatives / positives.

[0136] When applying the cadence adaptation mechanism, the scale-independent energy is determined for each window, as explained at 308b (FIG.3B).

[0137] The scale-independent energy ^^^(^^) for a given time window (^^) is then cross-correlatedwith an a priori estimate of the scale-independent energy (^^^ି ) that is equal to ^^^(^^ − 1), i.e., for theprevious time window.

[0138] As a result, the scale delay which reflects the changes in the propulsion speeds (fromone split to the next split) is obtained in accordance with Equation (6):^^ = ^^^^^^^^^^^^ ^^^ ∗ ^^^ି |^^ ∈ [1, ^^^^௫]| (6)

[0139] In at least one example, ^^^and ^^^ are estimated knowing the scale delay and ^^^ି and^^^ି . As referenced herein, ^^^and ^^^ estimated from ^^^ି and ^^^ି are called ^̂^^ and ^̂^^, as shown inEquations (7) and (8):^̂^^ = ^^^^^^^^^^^^ {^^ ∈ [^^௧ , ^^^^௫]|^^ᇱ^(^^)(7) =0 ^^^^^^ ^^"^(^^) < 0}^^^(^^)^̂^ = ^^^^^^^^^^^^ {^^ ∈

[0001] ᇱ^^ , ^^௧ |^^ (^^)= 0 ^^^^^^ ^^"^(^^) < 0}^^^(^^)^^௧ = ^^^^^^^^^^^^{ ^^ ∈ [^^^ି + ^^, ^^^ି + ^^]}^^^(^^) (8)

[0140] This process extracts the event and cycle spectral information from ^^^using the Gaussian approximation.

[0141] The parameters in ^^^are updated based on the extracted information to form an a posteriori estimate of the scale-independent energy (^^^^) which serves as the prior for the next window. This, in turn, is how the CWT method adapt with the alternation in propulsion speed.

[0142] It is expected that very fast transitions in propulsion speed cause large shifts in the scale-independent energy. These transitions are identified when the following constraints are notsatisfied:^^^ି + ^^ > 1(9) ^^^ି + ^^ < ^^^^௫^̂^ฬ ^^̂^ − 2ฬ < 0.2^

[0143] In this case, the ^^^ିmay be very different from that of ^^^. Hence, ^^^and ^^^estimated fromare not accurate resulting in incorrect ^^^^due to pooralignment of the two signals. For this case, in some examples, a 1D two-term Gaussian mix is fitted over ^^^ିand ^^^to estimate ^^^^for this part of the signal where fast transition in propulsion speed occurs.

[0144] Subsequently, to locate the temporal event regions representing the cycles and events (act 314b in method 300b), the parameters obtained from the cadence adaptation mechanism are used. These parameters contain ^̂^^and ^̂^^stored in every time window split that hold information about the local frequency of the event and cycle for the time duration of that window.

[0145] Using ^̂^^and ^̂^^obtained for all the splits (from all the overlapping windows) and calculating the CWT coefficients at those particular scales, two temporal signal representation areidentified (^^^^ and ^^^^):^^^^ = [^^^^^^(^̂^^ , 1),^^^^^^(^̂^^ , 2), … .. ,^^^^^^(^̂^^ , ^^), … . ,^^^^^^(^̂^^,^^)] (10)^^^^ = [^^^^^^(^̂^^, 1), ^^^^^^(^̂^^, 2), … .. , ^^^^^^(^̂^^, ^^), … . , ^^^^^^(^̂^^, ^^)] (11)

[0146] The system may then apply act 316b to determine the exact time stamps for the hand interaction events, as explained previously.

[0147] (iv.) Overall Method.

[0148] Method 300c (FIG.3C) exemplifies a generalized description of the overall method, as described above, that may include the cadence adaption mechanism. Method portion 302 refers to the experimental verification, as described below. V. EXPERIMENTAL TESTING AND RESULTS

[0149] The disclosed methods were validated against an instrumented wheelchair, able to determine the forces and moments applied to the push-rim during the wheelchair propulsion and thus detect tcontactand treleaseinstants.

[0150] (i.) Experimental Design and Setup.

[0151] An instrumented wheelchair equipped with two force- and moment-sensing push-rim wheels (SMARTWheel,Three Rivers Holdings LLC, AZ, USA) was used as a reference criterion for the detection of hand contact with push-rim.

[0152] The instrumented wheelchair was placed on a roller platform setup 114 (FIG.1A) in which the user could freely propel the wheelchair in a controlled environment.

[0153] In addition, an IMU (MTws, XSENS Technologies, NL) was placed on the dorsal aspect of the hand, over the third metacarpal, between the wrist and fingers, using double-sided medical tape. Its X-axis, Y-axis, and Z-axis were oriented dorsally perpendicular to the hand, medially toward the thumb, and proximally, respectively, to record the hand motion on the push-rim (FIG.1A).

[0154] SMARTWheel recorded the force and moment applied to the push-rim with precision of 2N, resolution of 0.2 N, and sampling frequency of 240 Hz. IMUrecorded the triaxial acceleration with a full-scale range of ±160 m / sec2 with a sampling frequency of 100 Hz (FIG.5A).

[0155] Twenty (23) participants (14 males and 9 females, including 9 experienced manual wheelchair users) consented to participate. Once the instrumented wheelchair and the IMU were set up, the participants were asked to take time and adjust to the instrumented wheelchair and roller platform. Additionally, participants were advised to freely propel the instrumented wheelchair to become comfortable and familiar with the experimental setup.

[0156] Then, the participants were asked to propel at three different self-selected speeds: slow (0.4 ± 0.2 m / s), medium (0.6 ± 0.2 m / s), and fast (0.8 ± 0.2 m / s), for 1 minute at each speed level.

[0157] Additionally, to account for any possible variations in wheelchair propulsion over an extended period of time, participants were asked to propel the wheelchair for more than five (5) minutes with a self-selected speed. The participants were encouraged to experiment with different body postures, including bending forward and backward and different propulsion patterns during the trials (e.g., FIGs.2B – 2D).

[0158] These instructions were part of the approach to simulate a variety of real-world conditions. Also, participants were instructed to tap the push-rim three times at both the start and endof each trial. These actions produced six distinct spikes in the data recorded by the IMU and the SMARTWheel, enabling the synchronization of these two instruments.

[0159] (ii.) Experimental Results.

[0160] FIG. 5D shows a visual representative of the tcontact and trelease measured by the SMARTWheel and those detected by the IMU.

[0161] The disclosed system and method illustrated average accuracy of 96% and 96%, average F1-score of 98% and 98%, average precision of 96% and 96%, and average sensitivity of 99% and 99% in detecting tcontact and trelease, respectively (Table I). Cumulative (%) tcontact (%) trelease (%) Accuracy 96(5) 96(7) 96(7) F1-Score 98(4) 98(4) 98(4) Precision 96(3) 96(6) 96(6) Sensitivity 99(3) 99(3) 99(3) Table I – The Average (SD) of accuracy, F1-score, precision, and sensitivity of disclosed example method in detecting tcontactandtrelease

[0162] The comparison of the IMU measurements against SMARTWheel revealed a temporal error in detecting tcontactand treleasethat were comparable to one sampling interval of the IMU (10 milliseconds), indicating that the mean temporal event detection error lies within a range of 3 recording samples for tcontact and trelease. The absolute temporal errors were 10± 20 milliseconds (mean ± SD) for tcontactand 20 ± 30 milliseconds for trelease. Correspondingly, the SEM for these temporal errors were both less than 10 milliseconds. The relative errors in detecting tcontact and trelease, expressed as error divided by the wheelchair propulsion cycle duration (tcontact to trelease), were 0.4±1.1% and 0.7±1.0%, respectively. Correspondingly, the SEM for these detections was both less than 0.1% (FIG.6).

[0163] Additionally, the temporal error in push duration and recovery duration detection were 10 ± 60 milliseconds (mean ± SD) and −20 ± 80 milliseconds, respectively. Correspondingly, SEM for these mean temporal errors were less than 1 millisecond.

[0164] The relative temporal errors, expressed as error divided by the wheelchair propulsion cycle duration (tcontact to trelease), were 0.4 ± 2.0% and −0.6 ± 3.0%, respectively (FIG.6).

[0165] A three-way mixed ANOVA was run on bootstrapped data to understand the effect of wheelchair propulsion speed, sex, and experience level on absolute temporal error. Absolute temporal errors were not normally distributed, as assessed by Kolmogorov-Smirnov and Shapiro-Wilk’s test (p < 0.05), and there were outliers in the data, as assessed by inspection of a boxplot.

[0166] There was homogeneity of variances for absolute temporal error during the medium speed propulsion (p = 0.143), but not for absolute temporal error during the slow and medium speed propulsions (p < 0.05), as assessed by Levene’s test for equality of variances.

[0167] It was assumed that three-way mixed ANOVA is robust to non-normality and heterogeneity of variance since the group sample sizes exceeded 30 samples for each level and were approximately equal. Also, there are only three levels of the within-participants factor, thus the assumption of sphericity is automatically met.

[0168] The three-way interaction between wheelchair propulsion speed, sex and experience level was not statistically significant (p = 0.152). As a result, the two-way interactions were analyzed.

[0169] There was a statistically significant two-way interaction between wheelchair propulsion speed and sex and wheelchair propulsion speed and experience level (p < 0.05). Statistical significance of a simple main effect was accepted at a Bonferroni-adjusted alpha level of 0.017 (0.05 / 3). Looking at two-way interaction between propulsion speed and sex, there was a statistically significant simple main effect of sex at the medium speed propulsion (p < .017), but not at the slow and fast speed propulsions (p = 0.835 and p = 0.091).

[0170] In addition, there was a statistically significant simple main effect of experience level at the medium and fast speed propulsion (p < .017) but not at the slow speed propulsions (p = 0.333). All pairwise comparisons were performed for statistically significant simple main effects. Bonferroni corrections were made with comparisons within each simple main effect considered a family of comparisons. It was observed that the absolute temporal error was higher in experienced manual wheelchair users than inexperienced ones during medium speed propulsion with a mean difference of22 milliseconds (95% CI, 17 to 27 milliseconds, p < 0.001) and fast speed propulsions with a mean difference of 26 milliseconds (95% CI, 18 to 34 milliseconds, p < 0.001).

[0171] Also, the absolute temporal error was higher in female manual wheelchair users during medium speed propulsion with a mean difference of 12 milliseconds (95% CI, 7 to 17 milliseconds, p < 0.001). In summary, our proposed method demonstrates great performance in detecting hand contact and hand release events and identifying push phase and recovery phase with high accuracy, F1-score, precision, and sensitivity.

[0172] Moreover, the proposed method illustrated high temporal accuracy in detecting events and estimating the temporal durations, as evidenced by the low mean absolute temporal error, SD and SEM, values, which are below 20, 90, and 1 milliseconds, respectively. VI. EXAMPLE HARDWARE CONFIGURATION FOR MONITORING DEVICE

[0173] FIG. 7 shows an example electrical hardware configuration for an example motion tracking unit 102.

[0174] As shown, the motion tracking unit 102 includes a processor 702 coupled to a memory 704, and one or more of motion sensor(s) 706, a display interface 708, and a communication interface 710.

[0175] In some examples, memory 702 stores any of methods 300a – 300b.

[0176] Motion sensor(s) 806 can include any sensor for detecting and tracking motion. In some examples, this includes accelerometers such as tri-axial accelerometer. In other examples, it may include gyroscopes, such as tri-axial gyroscopes.

[0177] Display interface 808 can be an output interface for displaying data (e.g., an LCD screen).

[0178] Communication interface 810 may comprise a cellular modem and antenna for wireless transmission of data to the communications network, such as network 150.VII. ALTERNATE OR SPECIFIC EXAMPLES

[0179] While disclosed examples use tri-axial accelerometer data, the same methods and concepts are applicable to tri-axial gyroscope data or various other types of motion data. Accordingly, in some examples, act 302a involves obtaining relevant “motion data” broadly, and act 304a involves processing the obtained motion data. Further, method 300b is applied in respect of such motion data.

[0180] Additionally, in some examples, the methods of FIGs.3A and 3B are applicable in real time or near real time. For examples, the methods are applied in real time or near real time as the motion tracking unit is obtaining real time or near real time acceleration data. In other examples, the methods are applied after the fact. For example, after motion data is captured, it is analyzed in accordance with methods 300a and 300b. VIII. INTERPRETATION

[0181] Various systems or methods have been described to provide an example of an embodiment of the claimed subject matter. No embodiment described limits any claimed subject matter and any claimed subject matter may cover methods or systems that differ from those described below. The claimed subject matter is not limited to systems or methods having all of the features of any one system or method described below or to features common to multiple or all of the apparatuses or methods described below. It is possible that a system or method described is not an embodiment that is recited in any claimed subject matter. Any subject matter disclosed in a system or method described that is not claimed in this document may be the subject matter of another protective instrument, for example, a continuing patent application, and the applicants, inventors or owners do not intend to abandon, disclaim or dedicate to the public any such subject matter by its disclosure in this document.

[0182] Furthermore, it will be appreciated that for simplicity and clarity of illustration, where considered appropriate, reference numerals may be repeated among the figures to indicate corresponding or analogous elements. In addition, numerous specific details are set forth in order to provide a thorough understanding of the embodiments described herein. However, it will be understood by those of ordinary skill in the art that the embodiments described herein may be practicedwithout these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the embodiments described herein. Also, the description is not to be considered as limiting the scope of the embodiments described herein.

[0183] It should also be noted that the terms “coupled” or “coupling” as used herein can have several different meanings depending in the context in which these terms are used. For example, the terms coupled or coupling may be used to indicate that an element or device can electrically, optically, or wirelessly send data to another element or device as well as receive data from another element or device. As used herein, two or more components are said to be “coupled”, or “connected” where the parts are joined or operate together either directly or indirectly (i.e., through one or more intermediate components), so long as a link occurs. As used herein and in the claims, two or more parts are said to be “directly coupled”, or “directly connected”, where the parts are joined or operate together without intervening intermediate components.

[0184] It should be noted that terms of degree such as "substantially", "about" and "approximately" as used herein mean a reasonable amount of deviation of the modified term such that the end result is not significantly changed. These terms of degree may also be construed as including a deviation of the modified term if this deviation would not negate the meaning of the term it modifies.

[0185] Furthermore, any recitation of numerical ranges by endpoints herein includes all numbers and fractions subsumed within that range (e.g.1 to 5 includes 1, 1.5, 2, 2.75, 3, 3.90, 4, and 5). It is also to be understood that all numbers and fractions thereof are presumed to be modified by the term "about" which means a variation of up to a certain amount of the number to which reference is being made if the end result is not significantly changed.

[0186] The example embodiments of the systems and methods described herein may be implemented as a combination of hardware or software. In some cases, the example embodiments described herein may be implemented, at least in part, by using one or more computer programs, executing on one or more programmable devices comprising at least one processing element, and a data storage element (including volatile memory, non-volatile memory, storage elements, or any combination thereof). These devices may also have at least one input device (e.g. a pushbuttonkeyboard, mouse, a touchscreen, and the like), and at least one output device (e.g. a display screen, a printer, a wireless radio, and the like) depending on the nature of the device.

[0187] It should also be noted that there may be some elements that are used to implement at least part of one of the embodiments described herein that may be implemented via software that is written in a high-level computer programming language such as object oriented programming or script-based programming. Accordingly, the program code may be written in Java, Swift / Objective- C, C, C++, Javascript, Python, SQL or any other suitable programming language and may comprise modules or classes, as is known to those skilled in object oriented programming. Alternatively, or in addition thereto, some of these elements implemented via software may be written in assembly language, machine language or firmware as needed. In either case, the language may be a compiled or interpreted language.

[0188] At least some of these software programs may be stored on a storage media (e.g. a computer readable medium such as, but not limited to, ROM, magnetic disk, optical disc) or a device that is readable by a general or special purpose programmable device. The software program code, when read by the programmable device, configures the programmable device to operate in a new, specific and predefined manner in order to perform at least one of the methods described herein.

[0189] Furthermore, at least some of the programs associated with the systems and methods of the embodiments described herein may be capable of being distributed in a computer program product comprising a computer readable medium that bears computer usable instructions for one or more processors. The medium may be provided in various forms, including non-transitory forms such as, but not limited to, one or more diskettes, compact disks, tapes, chips, and magnetic and electronic storage. The computer program product may also be distributed in an over-the-air or wireless manner, using a wireless data connection.

[0190] The term “software application” or “application” refers to computer-executable instructions, particularly computer-executable instructions stored in a non-transitory medium, such as a non-volatile memory, and executed by a computer processor. The computer processor, when executing the instructions, may receive inputs and transmit outputs to any of a variety of input oroutput devices to which it is coupled. Software applications may include mobile applications or “apps” for use on mobile devices such as smartphones and tablets or other “smart” devices.

[0191] A software application can be, for example, a monolithic software application, built in- house by the organization and possibly running on custom hardware; a set of interconnected modular subsystems running on similar or diverse hardware; a software-as-a-service application operated remotely by a third party; third party software running on outsourced infrastructure, etc. In some cases, a software application also may be less formal, or constructed in ad hoc fashion, such as a programmable spreadsheet document that has been modified to perform computations for the organization’s needs.

[0192] Software applications may be deployed to and installed on a computing device on which it is to operate. Depending on the nature of the operating system and / or platform of the computing device, an application may be deployed directly to the computing device, and / or the application may be downloaded from an application marketplace.

[0193] The present invention has been described here by way of example only, while numerous specific details are set forth herein in order to provide a thorough understanding of the exemplary embodiments described herein. However, it will be understood by those of ordinary skill in the art that these embodiments may, in some cases, be practiced without these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the description of the embodiments. Various modification and variations may be made to these exemplary embodiments without departing from the spirit and scope of the invention, which is limited only by the appended claims.

Claims

CLAIMS:

1. A method for estimating parameters characterizing manual wheelchair propulsion, comprising: - obtaining motion data over time, wherein the motion data is generated as a subject propels a wheel of the wheelchair; - processing the motion data to determine timestamps associated with hand interaction events, wherein, the hand interaction events comprise one or more of a hand contact event and a hand release event, the hand contact event defining a first time instance when a subject’s hand contacts the wheel of the wheelchair and initiates a propulsion stroke, and the hand release event defines a second time instance when the subject’s hand releases from the wheel and ends the propulsion stroke; and - generating an output based on the hand interaction events.

2. The method of claim 1, wherein the method further comprises, based on the timestamps for the hand interaction events, determining a time interval for a push phase and a recovery phase.

3. The method of any one of claims 1 or 2, wherein the motion data comprises tri-axial acceleration data, and processing the acceleration data comprises, initially, determining a resultant acceleration data over time.

4. The method of claim 3, wherein the processing of the acceleration data is based on a continuous wavelet transform (CWT) technique.

5. The method of claim 4, wherein processing of the acceleration data further comprises:- applying the CWT technique to the resultant acceleration data to identify one or more temporal regions corresponding to hand interaction events and cycle events; and - in respect of each of the temporal regions associated with hand interaction events, analyzing the resultant acceleration data in these temporal regions to determine timestamps for the hand interaction events.

6. The method of claim 5, wherein applying the CWT to the resultant acceleration data comprises: - segmenting the resultant acceleration data into one or more time window splits; - determining CWT coefficients for each time window; - for each time window, estimating a scale-independent energy; and - based on the scale-independent energy, identifying the one or more temporal regions corresponding to the hand interaction events and the cycle events.

7. The method of claim 6, further comprising applying a cadence adaptation mechanism after estimating the scale-independent energy, wherein the cadence adaption mechanism is applied to accommodate for changes in propulsion speed.

8. The method of any one of claims 5 to 8, wherein analyzing the acceleration data in the temporal regions comprises applying a peak detection method to the resultant acceleration data.

9. The method of any one of claims 1 to 8, wherein the one or more sensors part of a motion tracking unit.

10. The method of claim 9, wherein the motion tracking unit is coupled to one or more of the subject’s hand, wrist, forearm, upper arm and chest.

11. A system for estimating parameters characterizing manual wheelchair propulsion, comprising:- one or more sensors for generating motion data, wherein the motion data is generated as a subject propels a wheel of a wheelchair; and - at least one processor coupled to the one or more sensors, and configured for: - processing the motion data to determine timestamps associated with hand interaction events, wherein, the hand interaction events comprise one or more of a hand contact event and a hand release event, the hand contact event defining a first time instance when a subject’s hand contacts the wheel of the wheel chair and initiates a propulsion stroke, and the hand release event defines a second time instance when the subject’s hand releases from the wheel and ends the propulsion stroke; and - generating an output based on the hand interaction events.

12. The system of claim 11, wherein the at least one processor is further configured for: based on the timestamps for the hand interaction events, determining a time interval for a push phase and a recovery phase.

13. The system of any one of claims 11 or 12, wherein the motion data comprises tri-axial acceleration data, and processing the acceleration data comprises, initially, determining a resultant acceleration data over time.

14. The system of claim 13, wherein the processing of the acceleration data is based on a continuous wavelet transform (CWT) technique.

15. The system of claim 14, wherein processing of the acceleration data further comprises the at least one processor being further configured for:- applying the CWT technique to the resultant acceleration data to identify one or more temporal regions corresponding to hand interaction events and cycle events; and - in respect of each of the temporal regions associated with hand interaction events, analyzing the resultant acceleration data in these temporal regions to determine timestamps for the hand interaction events.

16. The system of claim 15, wherein applying the CWT to the resultant acceleration data comprises the at least one processor being further configured for: - segmenting the resultant acceleration data into one or more time window splits; - determining CWT coefficients for each time window; - for each time window, estimating a scale-independent energy; and - based on the scale-independent energy, identifying the one or more temporal regions corresponding to the hand interaction events and the cycle events.

17. The system of claim 16, wherein the at least one processor being further configured for: applying a cadence adaptation mechanism after estimating the scale-independent energy, wherein the cadence adaption mechanism is applied to accommodate for changes in propulsion speed.

18. The system of any one of claims 15 to 18, wherein analyzing the acceleration data in the temporal regions comprises applying a peak detection method to the resultant acceleration data.

19. The system of any one of claims 11 to 18, wherein the one or more sensors form part of a motion tracking unit.

20. The system of claim 19, wherein the motion tracking unit is coupled to one or more of the subject’s hand, wrist, forearm, upper arm and chest.

Images