SYSTEM, METHOD AND COMPUTER PROGRAMS FOR ASSESSMENT OF BODY MOVEMENT'S CONDITIONS OR DISORDERS

Abstract

A system, method and computer program for assessment of body movement's conditions or disorders are proposed. The system comprises several monitoring sensors to be attached to different body areas of a person to obtain biomechanical and/or physiological variables thereof according to a specified sensor configuration; a memory or database, having stored therein at least one of: control biomechanical and/or physiological variables and pathology biomechanical and/or physiological variables obtained. The variables being stored classified in three different categories: cardiac, kinematics and plantar-pressure. A processing unit generating a normality model; getting biomechanical and/or physiological variables from a given user, classifying them into the three different categories, and selecting, for each category, those that are significant; computing, for the selected variables of the given user, a unified category score for each category; and computing a condition or disorder score as the deviation between the computed unified category scores with the generated normality model.

Claims

1. A system for assessment of body movement's conditions or disorders, comprising: several monitoring sensors, each one being configured to be attached to a different area of a body of a person to obtain one or more variables of the area according to a specified sensor configuration, the one or more variables including at least one of biomechanical and physiological variables, and the monitoring sensors comprising at least three sensors selected from the group consisting of: a heart rate sensor, an inertial sensor or a goniometer, and a plantar pressure sensor; a memory or database, having stored therein at least one of: control variables, obtained during different sessions of a same or different duration from healthy users while they performed a function test using the monitoring sensors, and pathology variables, obtained during different sessions of a same or different duration from unhealthy users while they performed the function test using the monitoring sensors, the unhealthy users suffering a given body movement condition or disorder; either or both of the control variables and the pathology variables for each session being stored classified in at least three different categories including cardiac, kinematics and plantar-pressure; and a processing unit operatively connected to the memory or database, the processing unit being configured to: generate a normality model for the given body movement condition or disorder by implementing a statistical-based feature selection process on either or both of the stored control variables and the pathology variables, the generated normality model configured to define a unified category score for each category of the at least three different categories, each unified category score outlining the variables that better characterize the given body movement condition or disorder; obtain variables from a given user while the given user performed the function test during a given session using the monitoring sensors, the given user suffering the given body movement condition or disorder; classify the obtained variables of the given user into the at least three different categories, and select, for each category, variables of the given user by considering the variables that better characterize the given body movement condition or disorder from the generated normality model; compute, for the selected variables of the given user, a unified category score for each category; and compute a condition or disorder score for the given user as the deviation between the computed unified category scores of the given user with the generated normality model.

2. The system of claim 1, wherein the at least three different categories comprise two additional categories, including spatial-temporal and electromyography, and wherein the monitoring sensors are selected from the group consisting of: a heart rate sensor, an electromyography sensor, an inertial sensor or a goniometer, and a plantar pressure sensor.

3. The system of claim 1, wherein the memory or database is configured to store both the control variables and the pathology variables, and wherein the statistical-based feature selection process comprises: comparing the control variables between two different sessions and considering the control variables that exhibited a same underlying distribution for both sessions as robust control variables; and comparing the pathology variables between two different sessions and considering the pathology variables that exhibited a same underlying distribution for both sessions as robust pathology variables.

4. The system of claim 3, wherein the statistical-based feature selection process further comprises comparing the robust control variables with the robust pathology variables, obtaining a compared set of robust variables as a result, and selecting from the compared set, the variables showing a difference lower than a given threshold as the variables that better characterize the given body movement condition or disorder.

5. The system of claim 1, wherein the specified sensor configuration comprises a time-synchronization with the other monitoring sensors.

6. The system of claim 1, wherein the function test comprises at least one of: moving the arms, getting up and sitting down from a seat, a 6-Minute Walking Test, a 10-Meter Walking Test, a Timed Up and Go Test, and a Stair Climb Test.

7. A computed-implemented method for assessment of body movement's conditions or disorders, wherein a memory or database comprises stored therein at least one of: control variables, obtained from healthy users while they performed a function test using several monitoring sensors during different sessions of a same or different duration, and pathology variables, obtained from unhealthy users while they performed the function test using the several monitoring sensors during different sessions of a same or different duration, the unhealthy users suffering a given body movement condition or disorder, each one of the different monitoring sensors, for each session and for each healthy and unhealthy user, being configured to be attached to a different area of a body to obtain one or more variables of the area according to a specified sensor configuration, the one or more variables including at least one of biomechanical and physiological variables; the monitoring sensors comprising at least three sensors that are selected from the group consisting of: a heart rate sensor, an inertial sensor or a goniometer, and a plantar pressure sensor; the control variables, the pathology variables, or both, for each session, being stored classified in at least three different categories, including cardiac, kinematics and plantar-pressure; the method comprising performing by one or more processors of a processing unit the following steps: generating a normality model for the given body movement condition or disorder by implementing a statistical-based feature selection process on the stored control or pathology variables, the generated normality model defining a unified category score for each category of at least three different categories, each unified category score outlining the variables that better characterize the given body movement condition or disorder; once stored in the memory or database, obtaining variables obtained from a given user while the given user performed the function test during a given session using the monitoring sensors, the given user suffering from the given body movement condition or disorder; classifying the gotten variables of the given user into the at least three different categories, and selecting, for each category, the variables of the given user by considering the variables that better characterize the given body movement condition or disorder from the generated normality model; computing, for the selected variables of the given user, a unified category score for each category; and computing a condition or disorder score for the given user as the deviation between the computed unified category scores of the given user with the generated normality model.

8. The method of claim 7, wherein the at least three different categories comprise two additional categories, including spatial-temporal and electromyography, and the monitoring sensors are selected from the group consisting of: a heart rate sensor, an electromyography sensor, an inertial sensor or goniometer, and a plantar pressure sensor.

9. The method of claim 7, wherein the memory or database stores both the control variables and the pathology variables, and the statistical-based feature selection process comprises: comparing the control variables between two different sessions and considering the control variables that exhibited a same underlying distribution for both sessions as robust control variables; and comparing the pathology variables between two different sessions and considering the pathology variables that exhibited a same underlying distribution for both sessions as robust pathology variables.

10. The method of claim 9, wherein the statistical-based feature selection process further comprises comparing the robust control variables with the robust pathology variables, obtaining a compared set of robust variables as a result, and selecting from the compared set, the variables showing a difference lower than a given threshold as the variables that better characterize the given body movement condition or disorder.

11. The method of claim 10, wherein the memory or database further has stored therein one or more control mobility metrics and one or more pathology mobility metrics obtained from the healthy and unhealthy users, respectively, during the different sessions, the method further comprising validating the computed condition or disorder score by comparing it with a normalized score obtained from both the one or more control mobility metrics and the one or more pathology mobility metrics.

12. The method of claim 7, wherein the specified sensor configuration comprises a time-synchronization with the other monitoring sensors.

13. The method of claim 7, wherein the cardiorespiratory function test comprises at least one of: moving the arms, getting up and sitting down from a seat, a 6-Minute Walking Test, a 10-Meter Walking Test, a Timed Up and Go Test, a Stair Climb Test.

14. A non-transitory computer-readable medium storing instructions configured to cause at least one processor of a computing system to perform the method according to claim 7.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The previous and other advantages and features will be more fully understood from the following detailed description of embodiments, with reference to the attached figures, which must be considered in an illustrative and non-limiting manner, in which:

[0034] FIG. 1 is a flow chart of a method for assessment of body movement's conditions or disorders, according to an embodiment of the present invention.

[0035] FIG. 2 is a flow chart illustrating how the normality model for a given condition or disorder is generated, according to an embodiment of the present invention.

[0036] FIG. 3 is a boxplot distribution for each one of the assessed categories. Control group achieves scores closer to one while DMD group achieves lower scores.

[0037] FIG. 4 are spider plots of the computed scores for a control subject and 5 DMD patients with different levels of affection. While NSAA scores decrease, so do the scores. However, these reductions are not uniform among categories, being highly dependent on the subjects.

[0038] FIG. 5 illustrates the accuracies of 6- and 2-minutes scores, 6MWT and 2MWT in comparison with NSAA score.

[0039] FIG. 6 illustrates the regression with the target variables of each score methodology with age.

DETAILED DESCRIPTION OF THE INVENTION AND OF PARTICULAR EMBODIMENTS

[0040] Present invention provides a new e-health platform/solution for assessment of body movement's conditions or disorders, particularly neuromuscular diseases or disorders. The present invention may be also applied for the assessment of other diseases/disorders or even neuropathies.

[0041] The invention integrates various (portable) monitoring sensors (e.g. heart rate, electromyography, inertial or goniometer, plantar pressure) to acquire different biomechanical and/or physiological parameters/variables during the performance of functional tests (e.g. 6-Minute Walking Test, 10-Meter Walking Test, Timed Up and Go Test, Stair Climb Test, moving the arms, getting up and sitting down from a seat, etc.). The monitoring sensors are attached to different areas of the body of a subject to obtain biomechanical and/or physiological variables of said area. Particularly, the different monitoring sensors are synchronized.

[0042] The different parameters/variables obtained are then stored and processed and the evolution of such conditions or disorders computed by assessing the functional capacity of the patient/user with long-lasting conditions such as, among others, muscular disturbances, respiratory diseases, etc.

[0043] When a test is finished, the different parameters obtained can be transmitted either to a software application (APP) run in a portable computing device such as a smartphone or a tablet, among others, or to a computer/server (e.g. a platform backend). In some embodiments, the different parameters can be stored in a database. If transmitted to the APP the latter can further transmit the session data to the platform backend for the session data be stored (in a memory of the platform), processed and analyzed. In some embodiments, the processing and analyzing of the session data is directly made by the APP.

[0044] The APP can be used by healthcare professionals to guide them through all the steps to facilitate the process of connecting the monitoring sensors and performing each function test. Either the platform or the APP can show the final assessment performed.

[0045] Referring to FIG. 1, the assessment of the body condition or disorder X comprises, at step 1001, the execution of the function tests, during different sessions of a the same or different extent, by the control group and the pathology group to obtain biomechanical and/or physiological variables thereof using the cited monitoring sensors. At step 1002, the control and the pathology biomechanical and/or physiological variables, for each session, are classified in different categories. In some embodiments, such classification is made by considering three categories, particularly cardiac, kinematics and plantar-pressure. In some embodiments, the classification is made by considering additional categories, for instance four, five, six, etc. In some particular embodiments, five categories are applied, including cardiac, spatial-temporal, electromyography, kinematics and plantar-pressure.

[0046] In yet other alternative or complementary embodiments, the classification is made by considering clinical parameters such as steppage/ataxia, proximal/distal weakness, equilibrium/stability; asymmetry, stiff joints, among others.

[0047] Once said categorization of the variables is made, at step 1003, a normality model of the body movement condition or disorder X is generated. This can be done by implementing a statistical-based feature selection process over the control and pathology biomechanical and/or physiological variables that involves the calculation of a unified category score for each category outlining the biomechanical and/or physiological variables that better characterize the body movement condition or disorder X. This possibility to divide the score into the different categories, obtaining multicategory score, eases treatment personalization.

[0048] In other embodiments of the invention, not illustrated, the assessment can be made by considering just one of the groups, i.e. the control or the pathology. In this alternative case, the normality model will be generated by executing the statistical-based feature selection process using the variables obtained from one of the groups only.

[0049] In some embodiments, machine learning algorithms are used in said statistical-based feature selection process.

[0050] Continuing with the explanation of FIG. 1, when the test results (i.e. biomechanical and/or physiological variables) of a new patient of condition or disorder X arrive or are obtained (step 1004), these are classified, step 1005, into the same categories applied for the control and pathology groups. Then, at step 1006, for each category applied, the significant biomechanical and/or physiological variables of the new patient are selected and for the latter a unified category score computed (step 1007) for each category. Finally, at step 1008, a score for condition or disorder X is computed for the new patient. This can be done by calculating the deviation of the computed unified score from the normality model.

[0051] The final score aims to be an indicator that resumes the overall performance of the function test. An indicator that is reliable and responsive to the function test and that can be used as an endpoint for the time evolution of the condition/disorder and for clinical trials.

[0052] Referring to FIG. 2, an embodiment of the normality model generation process is illustrated. Similar to the embodiment of FIG. 1, in this case both control and pathology groups are also considered. According to this particular embodiment, both the control and pathology biomechanical and/or physiological variables (2002C, 2002P) obtained during the different sessions (2001C, 2001P) can be first (optionally) cleansed (steps 2003C, 2003P).

[0053] Then, the statistical-based feature selection process is implemented. At step 2004C, a first feature selection process is executed to select the most robust, and repeated, variables. At step 2004P an equivalent or similar process for the pathology group is made. This first feature selection process can be done by comparing the (control/pathology) biomechanical and/or physiological variables between two different sessions. The (control/pathology) biomechanical and/or physiological variables exhibiting the same underlying distribution for both sessions are considered robust.

[0054] At step 2005, a second feature selection process is executed. This can be done by comparing the robust control biomechanical and/or physiological variables (obtained in 2004C) with the robust pathology biomechanical and/or physiological variables (obtained in 2004P), and by selecting from the compared set, the biomechanical and/or physiological variables showing a difference lower than a given threshold. Finally, at steps 2006 and 2007, a unified score for each category is computed and the normality model generated.

[0055] Following a particular example of the application of the present invention for the assessment of DMD is detailed.

[0056] A pilot study was conducted on 28 male children (age 6.3-15.4 years) with DMD and 2 male children with (Becker Muscular Dystrophy) BMD (age 17.0-17.9 years) who attended their control visits to the Rehabilitation Service of the Sant Joan de Du Hospital (Barcelona, Spain). Most of the patients were treated with corticoids. The control group consisted of 27 healthy male children (age 9.0-13.9 years) that attended the Maristes la Inmaculada School (Barcelona). As DMD affects mainly males, all tests were performed with males.

Performed Tests

[0057] The 6MWT was carried out on a standardized, level, undisturbed, and well-lit hallway that measured 25 meters. Two cones were used to define the start and the end of the walkway. Prior to each test, surface EMGs were calibrated by having the participant perform a maximal voluntary contraction of the hamstrings, quadriceps, and gluteus muscles. Furthermore, the insoles were also calibrated. All calibration was done according to a standard procedure set on the proposed system. The participants were instructed to walk as fast as possible, without running, in a straight line back and forth between two cones for 6 minutes. They were permitted to rest if needed, by leaning against the wall. Stopwatch did not stop during rest period.

[0058] The completed distance was measured at two minutes (2MWT) and at 6 minutes.

[0059] The 6MWT test was conducted on three sessions: [0060] Session 1 (t0). The first session, tagged as t0, was aimed at collecting baseline data. During this session, the participant performed the 6MWT once. [0061] Session 2 (tor). The goal of the second session was to determine the test-retest reliability and was tagged as t0r. A subgroup of participants performed the test with a time gap of approximately 3-4 weeks. [0062] Session 3 (t1). The third session was tagged as t1 and was planned to validate the responsiveness. The test was performed with a time gap of approximately 6 months.

[0063] During the sessions, the monitoring sensors were attached to different areas of the body to measure the biomechanics/physics of the 6MWT. The type of attached sensors and their information is summarized in Table 1. The list of all acquired variables was grouped into five categories: cardiac, spatial-temporal, electromyography, kinematics and plantar-pressure.

TABLE-US-00001 TABLE 1 Sensor type Description Category Heart rate A strap applied below the chest, to measure Cardiac cardiorespiratory parameters. Surface EMGs EMGs were integrated in custom made, size- EMG appropriate shorts that enabled recording of muscle activity. Inertial sensors Sensors were placed at the thigh, shank, and spatial-temporal, in the plantar insoles to measure joint kinematics kinematics. Plantar pressure Pressure sensors were integrated in insoles Plantar pressure, which were size appropriate and used to obtain spatial-temporal, information on weight distribution during gait. kinematics

[0064] Furthermore, in each of the sessions, clinicians computed the North Star Ambulatory Assessment (NSAA), a scale used to measure functional motor abilities, for DMD patients. Control subjects received the maximum NSAA score. Subjects with medium NSAA values (from 15 to 20) at to suffered a steep decline between t0 and t1. The distance for 6MWT (6MWTD) was also computed. These two metrics (NSAA and 6MWTD) were used for posterior comparison with the computed score. These measurements were evaluated on a time window of six minutes, and on a time window of two minutes. The data analysis pipeline was executed for both time-windows.

Data Analysis

[0065] To obtain a multidimensional score that takes into consideration the information classified into the five listed categories the method was divided in two stages. The purpose of the first stage is the selection of the most robust variables which in turn present significantly different values between control and pathology subjects. These variables will be finally united into one representative variable for each of the categories. The objective of the second stage is to generate a normality model (or physiological walking model) composed of the five categories. The final score for a specific pathological subject is subsequently determined by analysing the distance of its variables to the normality model.

First Stage:

[0066] The purpose of this stage was to discard the variables that were not useful for discriminating pathologic and normal behaviours. It involved the following sub-stages: [0067] Data cleansing. Due to the nature of the DMD disorder, it is considered that the affectation is almost symmetrical between the right and left limbs. Thus, the variables corresponding to the sensors placed in both right and left parts of the body were merged into a single measure representing its mean value. [0068] Robust variable selection. The most robust and repeatable variables were selected considering the t0 and tor sessions for control and DMD subjects. The most robust variables were expected to have similar values between these two sessions as they were recorded in a short time-window. As the sets of data to compare arise from the same individuals at different sessions, t0 and tor, the data was paired. To determine which statistical test was the most suitable for each variable, a Shapiro-Wilk test was applied. If the variable followed a normal distribution for both the t0 sessions set and the tor sessions set, a Paired Sample T-Test was applied. If this condition was not met, the chosen test was Wilcoxon Signed-Rank Test. The variables that exhibited the same underlying distribution for both sessions (p>0.05) were considered robust and used for subsequent analysis. [0069] Selection of the variables that best characterize the condition/disorder. To identify the variables that characterize the pathology, this step compares how the evolution of the data differs between the control group and the pathology group. To do so, a Kruskal-Wallis test for unpaired data was used to compare the values of each one of the variables between control and pathology group. The values that shown a significant difference between groups (threshold p<0.01) were selected for further analysis. [0070] Generation of a unified score for each one of the categories. The selected variables, which belonged to one of the five categories, were firstly normalized (being 0 the minimum and 1 the maximum value). If the control median value was lower than the pathology median value, the variable was inverted to have a lower DMD score than a control score. Finally, the unified variable for each of the categories was computed as the mean of the normalized values of that category and normalized between 0 and 1.

Second Stage

[0071] While the determination of a category-by-category score may be useful to assess the subject performance in different aspects of the gait, there is interest in the definition of a score that characterizes the decline of the subject in the moment of the evaluation. The purpose of this stage was to determine the normality model and subsequent condition or disorder score (or pathology score) for one or more DMD subjects. [0072] Generation of the normality model. The unified variable for each of the categories and for the control group was further used to define a 5-dimensional gaussian model, where each dimension corresponded to one of the categories. [0073] Obtention of the pathology score. Each DMD and control subject obtains a score. The score was computed as the weighted log probability of belonging to the normality model. If the probability is the highest (which is expected for the control subjects) the score is 1. The lower the probability, the further the subject is from normality, and therefore has a greater impairment.

Score Validation

[0074] The computed pathology score (6M+ and 2M+ for 6 and 2 minutes, respectively) was validated by comparing it with the normalized NSAA score (from 0 to 1). For further comparisons, the 6MWTD and 2MWTD were also normalized and compared with NSAA. For the four measurements, accuracy was computed as:

100*(1|score-pscore-v|)

[0075] Where score-p and score-v represented the predicted score and the NSAA normalized score, respectively.

[0076] Finally, as DMD produces a progressive muscle degeneration, it is expected to obtain worse performance while age increase. To evaluate this effect, the linear regression between age and NSAA, the predicted scores (2M+ and 6M+), and the 6MWT and 2MWT distances was computed.

Results

Generation of a Unified Score for Each Category

[0077] The measurements of the 5 categories were assessed for 2- and 6-minutes intervals, obtaining a unified score for each category and for each period. FIG. 3 show as an example, the distribution of each one of the unified variables for both control an DMD groups, for 6 minutes.

[0078] The generation of these categorical scores allow to represent the deviations of the pathological subjects in diverse fields related to normal gait. As an example, FIG. 4 shows how different DMD patients presented different performances in the evaluated categories.

The Pathology 2M+ and 6M+ Score and its Comparison with NSAA and Distances.

[0079] To validate the computed score, the NSAA scale was taken into consideration as the gold-standard. FIG. 5 shows the global accuracy of the pathology scores (2M+ and 6M+) and 6MWTD and 2MWTD in comparison with NSAA. 6-minutes and 2-minutes test presented the highest accuracies for controls. For DMD, 6M+ and 6MWTD performances were similar (86.00 and 86.30 respectively). While the performance of 2M+ in comparison with 2MWTD was slightly higher (85.03 and 79.68 respectively). The highest decreases were observed in BMD, with a drop in performance of 19 points for 6M+vs. 6MWTD and a drop of 15 points for 2M+/2MWTD.

[0080] Finally, the regression with the ages of the patients is shown in FIG. 6. As a progressive disorder, it is expected to have a negative correlation with age. 6M+ and 2M+ scores were the most negatively correlated with age (R.sup.2 of 0.41 and 0.49 respectively), with higher values than NSAA score.

[0081] The present invention has been described in particular detail with respect to specific possible embodiments. Those of skill in the art will appreciate that the invention may be practiced in other embodiments. For example, the nomenclature used for components, capitalization of component designations and terms, the attributes, data structures, or any other programming or structural aspect is not significant, mandatory, or limiting, and the mechanisms that implement the invention or its features can have various different names, formats, and/or protocols. Further, the system and/or functionality of the invention may be implemented via various combinations of software and hardware, as described, or entirely in software elements. Also, particular divisions of functionality between the various components described herein are merely exemplary, and not mandatory or significant. Consequently, functions performed by a single component may, in other embodiments, be performed by multiple components, and functions performed by multiple components may, in other embodiments, be performed by a single component.

[0082] Certain aspects of the present invention include process steps or operations and instructions described herein in an algorithmic and/or algorithmic-like form. It should be noted that the process steps and/or operations and instructions of the present invention can be embodied in software, firmware, and/or hardware, and when embodied in software, can be downloaded to reside on and be operated from different platforms used by real-time network operating systems.

[0083] The scope of the present invention is defined in the following set of claims.