Machine learning classification system for patients with proprioceptive, mobility, and sensorimotor impairments

Abstract

An Individualized Signature Mapping (ISM) system objectively measures and assesses underlying musculoskeletal and / or neuromuscular parameters pertaining to impairments or conditions. The unique ISM signature obtained by the ISM system (100) provides an invaluable insight supporting critical decision making in the assessment, progression, and rehabilitation of the patient. The ISM system implements an artificial neural network to classify individuals as normal or impaired and generates recommendations to remediate the impairment that may be verified and further recommended by healthcare professionals.

Description

MACHINE LEARNING CLASSIFICATION SYSTEM FOR PATIENTS WITHPROPRIOCEPTIVE. MOBILITY. AND SENSORIMOTOR IMPAIRMENTS[1] CROSS-REFERENCE TO RELATED DISCLOSURES[2] This application incorporates by reference and claims priority to US Provisional Application No. 63 / 433,850 filed December 20, 2022.[3] FIELD OF THE DISCLOSURE[4] The disclosure relates to a system and method for classifying deviations and / or abnormalities of movements in the musculoskeletal system and associate the deviations / abnormalities with a particular condition, illness, or impairment.[5] BACKGROUND[6] Existing sensorimotor control tests, such as the Butterfly test for assessing musculoskeletal movement, have been shown to present with sufficient diagnostic accuracy for differentiating between patients with cervical spine impairment and healthy individuals. An exemplary method for assessment and graded training of sensorimotor functions is provided in international application No. PCT / IS2010 / 000010, filed on July 7, 2010, and published as WO 2011 / 004403 Al on January 13, 2011, which is incorporated herein by reference.[7] However, as impaired patients present a heterogenous group that frequently develop chronicity, there exists the elusive challenge of determining more targeted and individualized rehabilitation approaches for patients. Therefore, it is of importance to identify subgroups of impaired patients with specific kinaesthetic and / or mobility deficits.[8] The ability to differentiate between different subgroups of impaired patients represents an important challenge for the assessment tests. For example, the output of the Butterfly test provides different parameters that could enable more in-depth understanding of sensorimotor deficits of an individual neck pain patient. However, multiple parameters present a serious challenge for clinicians to interpret them appropriately. Therefore, there is a need for a system and method that can automatically identify clusters of parameters that present similar or different aspects of proprioceptive, mobility, and sensorimotor control and provide a more thorough analysis on the biometric characteristics and movement control of impaired patients. Such approach would ideally enable clinically relevant differentiation between subgroups of impaired patients and further provide group specific treatment guidelines. Additionally, suchapproach should ideally integrate the information provided by assessment test parameters, as well as impaired patient group biometric characteristics and enable easier clinical applicability.[9] Existing solutions for assessing patients with proprioceptive, mobility, and sensorimotor impairments do not carry out a fast, reliable, and cost-effective objective clinical assessments by objectively interpreting the physical quality and underlying abnormalities reflected in the movement deviations of the musculoskeletal system in patients.

[0010] There exists a need for proprioceptive, mobility, and sensorimotor impairment assessment that originates from creating or interpreting objective, individualized quantitative data that makes it possible to objectively identify deviations and abnormalities from musculoskeletal movements in patients. Such a solution would ideally assess the movement deviations and abnormalities to identify similarities between individuals having an impairment within a classified group that could be assessed or diagnosed with a strong statistical probability, thus linking their performance on the assessment test to a certain specific detrimental health condition or illness consistent with a predetermined classified group of individuals known to have similar performances and impairments.

[0011] The inventors of the disclosed system and method in the present application have found that by collecting and interpreting quantitative movement data derived from patients and healthy individuals, it is possible to classify deviations and / or abnormalities of movements or symptoms and associate the deviations / abnormalities with a particular condition or impairment in patients. Based on the classification(s), the system and method may further recommend appropriate rehabilitation or exercises related to addressing a condition or impairment. The system may also dynamically adjust the rehabilitation during therapy based on progress of the individual.

[0012] SUMMARY

[0013] In a first aspect of the present disclosure, an Individualized Signature Mapping (ISM) assessment system is provided to assess and classify performances of musculoskeletal impairment by analyzing and evaluating the biomechanics and physical qualities of the underlying musculoskeletal movement by including objective musculoskeletal performance data enabling for quantifying the statistical relevance / weight of the data in relation to a given or assumed detrimental health conditions or illness affecting individuals, based on the assessment test results; thus, enabling for objective assessment and classification of impairments affecting any of the musculoskeletal and neuromuscular system.

[0014] The ISM system comprises a tracking device having a sensor unit to obtain characteristic data pertaining to physiological and / or behavioral biometric information of an individual. In an embodiment, the sensor unit is attachable to the individual and includes at least one inertial measurement unit (IMU) for obtaining musculoskeletal movement data from an individual and at least one computing device or computer having a processor for processing the musculoskeletal movement data from the individual. Using one or more hardware storage devices that store instructions, encoded in an assessment application, that are executable on the computer, the system generates metrics based on the characteristic data collected from the sensor unit, and the assessment application compares the characteristic data and derived metrics from the individual to a baseline of characteristic data and metrics and classifies the characteristic data and metrics as healthy or unhealthy. In an embodiment, the tracking device is directly connected to the computer and configured to capture physiological or behavioral information of an individual without being directly attached to the user (e.g., motion capture camera).

[0015] The assessment application generates an objective diagnostic assessment based on the characteristic data and, using data analytics and machine learning algorithms, evaluates a range of motion, proprioception, and neuromuscular control related to the characteristic data. Results from the assessment tests can further be combined with analysis of symptoms and / or subjective assessments (e.g., questionnaires) to classify the individual into a group or subgroup classification having shared biometric characteristics or results with other individuals known to have a diagnosed impairment. The assessment application is a computer program configured to apply a statistical analysis to identify different groups of parameters having a specific functional or clinical meaning using the database of patient classification data. The different groups of parameters that can be identified include various combinations corresponding to other individual or patient populations. Based on performance of the assessment tests, the ISM data is subsequently classified.

[0016] In an embodiment, the computer comprises a monitor for displaying the application of the ISM system. The application generates a specialized movement procedure test for the individual to perform using the tracking device and projects angular displacement derived by the tracking device. In an exemplary embodiment, the tracking device is wearable by the individual and comprises a processor configured to calculate angular displacement based on measurement of acceleration, rotational velocity, etc. obtained by an IMU housed within the sensor unit. The tracking device may further comprise a storage device (e.g., storage 116)containing executable instructions for the processor to perform the calculations. The calculation of a correlation and / or deviation between the trajectory path of a tracing cursor displayed by a graphical user interface to the individual and the displayed computer-generated path may be implemented through the computer program (e.g., assessment application) and analyzed using various accuracy parameters.

[0017] The present disclosure also relates to a method for operating a ISM system for identifying deviations and abnormalities from musculoskeletal movements in individuals, comprising the steps of: (1) obtaining characteristic data from an individual using sensor, (2) processing the characteristic data from the individual using a first computer comprising a processor, (3) producing 3D data and derived metrics of the musculoskeletal movements; and (4) comparing the characteristic data and metrics from the individual to a baseline of characteristic data and metrics and further classifying the characteristic data as healthy or unhealthy, wherein the data classified as unhealthy is further classified into one or more subgroups consistent with an impairment or characteristic. Based on the subgroup classification, the system may recommend one or more rehabilitative programs or exercises for the individual to complete using the assessment application. The assessment application is further programed to dynamically adjust and update the recommended rehabilitative programs or exercises based on the performance of the individual during and / or after completion of the one or more rehabilitative programs or exercises.

[0018] The ISM optimizes the accuracy and quality of objective musculoskeletal assessment / diagnosis by utilizing individualized movement data for innovative assessment / diagnostic input in order to improve the quality of the interpretation of musculoskeletal movements / assessments of musculoskeletal movements in individuals and their reported diagnostic outcomes. By using the innovative objective data, the ISM accelerates the diagnostics and the rehabilitation process of subjects while reducing the cost of care and enhancing the quality of the healthcare delivery.

[0019] In an embodiment, the assessment application comprises a deep artificial neural network (ANN) to automatically evaluate impairment level and progress of performance, and to automatically recommend a rehabilitation or exercise strategy using data, e.g., physiological and behavioral biometric information, from the tracking device. The rehabilitation or exercise strategy may be recommended based on both biometric information of the tracking device and patient reported data from at least one questionnaire, the patient reported data being manually input into the processor.

[0020] The present disclosure also relates to a computer-implemented method for classifying sensorimotor impairments and generating rehabilitation strategies for a human subject, comprising the steps of (1) obtaining characteristic data from an individual using a tracking device and assessment application, (2) producing the characteristic data corresponding to the tracking device, (3) comparing the characteristic data from the individual to a baseline of characteristic data, and (4) classifying the characteristic data as normal (e.g., healthy) or impaired, wherein an impaired classification is defined as relating to one or more impairments identified from previously obtained characteristic data from a group of individuals.

[0021] These and other features, aspects, and advantages of the present disclosure will help better understand the following description, appended claims, and accompanying drawings.

[0022] BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to describe the manner in which the advantages and features of the systems and methods described herein can be obtained, a more particular description of the embodiments briefly described above will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. Understanding that these drawings depict only typical embodiments of the systems and methods described herein, and are not therefore to be considered to be limiting of their scope, certain systems and methods will be described and explained with additional specificity and detail through the use of the accompanying drawings.

[0024] Fig. 1 illustrates a schematic diagram depicting an exemplary signature mapping assessment system in accordance with an embodiment of the present disclosure.

[0025] Fig. 2 illustrates the signature mapping assessment system of Fig. 1 in connection with a service.

[0026] Fig. 3 illustrates a flowchart of the method incorporating an artificial neural network implemented by the system.

[0027] Fig. 4 illustrates the graphical user interface of the disclosed signature mapping assessment system with an assessment session display.

[0028] Fig. 5 illustrates pathways followed by a moving dot for the easy, medium, and difficult test patterns.

[0029] Fig. 6 illustrates visualizations of movement assessment test performance for an asymptomatic subject and a whiplash subject.

[0030] Fig. 7 illustrates a basic flowchart of the classification process implemented by thesystem.

[0031] The drawing figures are not necessarily drawn to scale. Instead, they are drawn to provide a better understanding of the components and are not intended to be limiting in scope but providing exemplary illustrations.

[0032] DEFINITIONS

[0033] For ease of understanding the disclosed embodiments of the present disclosure and associated method and system elements, a description of a few terms is necessary.

[0034] The term ‘assessment test’ generally refers to an interactive test or exercise conducted using the assessment application (i.e., Butterfly test, range of motion test, relocation test, rehabilitation exercise) performed by an individual, intended to evaluate a certain physical function, wherein the performance of the assessment test may be used in classifying the individual as having an impairment or condition. An assessment test may be related to motion, strength, balance, etc. and may also be used as a rehabilitative exercise for treating an impairment.

[0035] The term ‘characteristic data’ refers to a collection of discrete or continuous values that convey information, describing the quantity, quality, fact, statistics, other basic units of meaning, or simply sequences of symbols that may be further interpreted by the system. Exemplary characteristic data may include musculoskeletal, neuromuscular information, and / or biometric information, such as physiological, and / or behavioral biometrics. Characteristic data also refers to recorded information (e.g., in a network database) concerning proprioceptive, mobility, and sensorimotor impairments.

[0036] The term ‘computer’ or ‘computing device’ may include any device that comprises at least one processor and that electronically executes one or more programs, such as a user interface program and / or software program, and may include personal computers, laptop computers, servers, portable media players, hand-held devices, cellular phones, microprocessor-based programmable consumer electronic and / or appliances, and other similar electronic devices that include circuitry for wirelessly sending and / or receiving information.

[0037] The term ‘impairment’ refers to a condition, disability, injury, or symptomatic state affecting the muscular, skeletal, and / or nervous systems of an individual.

[0038] The term ‘individual’ refers to a user of the system or, more specifically, a patient or person using the tracking device and conducting the assessment test.

[0039] The term ‘individualized signature mapping’ or ‘ISM’ refers to a unique performance (i.e., record of characteristic data points) obtained by an individual using the tracking device and during an assessment test that may be used in classifying the performance of the individual as being impaired or not. The signature map may correspond to an image, video, motion path, infrared radiation reading, audio recording, frequency reading, and / or other types of unique identifiable data that define one or more biometric characteristics of interest. These biometric characteristics are defined as being ‘preselected’ and corresponding to a specific assessment test to evaluate physiological and / or behavioral biometrics of interest.

[0040] Unless otherwise specified, the term ‘network’ refers to one or more data links, e.g., comprising a database, that enable the wired or wireless transport of electronic data between computer systems and / or modules and / or other electronic devices.

[0041] The term ‘processor’ or ‘processing unit’ refers to one or more devices, circuits, and / or processing cores configured to process data, such as computer program instructions, and includes personal computers, desktop computers, laptop computers, message processors, handheld devices, multi-processor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile telephones, PDAs, tablets, pagers, routers, switches, and the like. Unless otherwise stated, references to a first processor may also apply to a second processor and vice versa.

[0042] The term ‘service’ refers to an automated program that is tasked with performing different actions based on input. As used herein, the terms ‘executable module,’ ‘executable component,’ ‘component,’ ‘module,’ ‘service,’ or ‘engine’ can refer to hardware processing units or to software objects, routines, or methods that may be executed on or with the system.

[0043] The term ‘software’ generally refers to computer-executable instructions, code, data, applications, programs, program modules, or the like maintained in or on any form or type of computer-readable media that is configured for storing computer-executable instructions or the like in a manner that is accessible to a computing device.

[0044] As used herein, reference to any type of machine learning or artificial intelligence may include any type of machine learning algorithm or device, convolutional neural network(s), multilayer neural network(s), recursive neural network(s), deep neural network(s), decision tree model(s) (e.g., decision trees, random forests, and gradient boosted trees) linear regression model(s), logistic regression model(s), support vector machine(s) (SVM), artificial intelligence device(s), or any other type of intelligent computing system. Any amount of training data maybe used (and perhaps later refined) to train the machine learning algorithm to dynamically perform the disclosed operations.

[0045] DETAILED DESCRIPTION OF VARIOUS EMBODIMENTS

[0046] A better understanding of different embodiments of the disclosure may be had from the following description read with the accompanying drawings in which reference characters refer to like elements. While the disclosure is susceptible to various modifications and alternative constructions, certain illustrative embodiments are in the drawings and are described below. It should be understood, however, that there is no intention to limit the disclosure to the embodiments disclosed; on the contrary, the intention covers all modifications, alternative constructions, combinations, and equivalents falling within the spirit and scope of the disclosure.

[0047] With respect to the use of plural and / or singular terms herein, those skilled in the art may translate the terms from the plural to the singular and / or from the singular to the plural as is appropriate to the context and / or application. The various singular / plural permutations may be expressly set forth herein for sake of clarity.

[0048] It will be understood that unless a term is defined to possess a described meaning, there is no intent to limit the meaning of such term, either expressly or indirectly, beyond its plain or ordinary meaning.

[0049] The disclosed assessment and diagnostic system is based on the functional and statistical interpretation of objective innovative data derived from measurements of musculoskeletal movements of individuals, facilitating for linking specific assessment / diagnostic outcomes with a specific illness / condition in individuals. The serology and hematology of an individual are sources of objective data that can aid in disease assessment and diagnostics.The disclosed solution objectively measures, derives, and assesses underlying physical parameters that play important part in the function and wellbeing of individuals. By assembling, comparing, and processing the movement data, the quality of movements can be assessed and used to classify impairments and assist healthcare professionals in diagnosing an underlying disease or condition affecting individuals. This assessment / diagnostic process is referred to as the Individualized Signature Mapping (ISM).

[0050] The ISM method enables for enhancing the statistical likelihood of a better and more accurate diagnosis in healthcare of individuals and serves as a tool to indicate a particular healthissue or diseases. ISM is defined as an innovative musculoskeletal and neuromuscular assessment and diagnostic method for both chronic and non-chronic conditions in individuals.

[0051] Measuring, storing, comparing, and classifying diverse musculoskeletal movement variables facilitates for the creation of a unique identifiable signature that enables the identification of abnormal physical condition or illness, enabling for objective diagnostics of a particular disorder / ailment. The unique ISM signature provides an invaluable insight supporting critical decision making in the assessment, progression and rehabilitation of the patient. In addition, it facilitates increased quality and possibly reduction of time and cost in the healthcare delivery.

[0052] By applying data feature ranking capabilities to identify the most important data (weight / rank) to decide on the most accurate disease diagnostics, augments the ISM process and facilitates for a greater accuracy and facilitates for a reduced cost, time and quality in the healthcare delivery.

[0053] Operating the ISM preferably involves the following: (1) licensed or trained operator to run the ISM using dedicated secure web service; (2) medical professional trained to operate the ISM using dedicated secure web service; (3) self-service interactive turnstiles or kiosks that provide the user with a guiding semi or fully automotive assessment / diagnostic tests that can be performed either indoor, outdoor, in the car, at the clinic, via mobile phone, at home or one's workplace using secure web service; and (4) user access to the preventive and predictive assessment / diagnostic interface of the ISM in real-time via ISM in TeleMedicine / TeleHealth mode.

[0054] Exemplary Embodiment to Objectively Assess and Diagnose Cervical Impairment

[0055] Affecting nearly two-thirds of the general population at least once in their life, neck pain is a growing healthcare concern. Common causes for this condition can include whiplash, a blow to the head, strenuous working conditions, or sustained poor posture. For example, professionals who spend many hours hunched over their workspace — such as surgeons and dentists — frequently develop neck pain. Those who wear heavy protective helmets, including athletes, jet and helicopter pilots, warriors and firefighters, are also at risk.

[0056] The techniques clinicians currently use to assess neck impairment have significant drawbacks because majority of them rely on cumbersome tests, such as manually operated range-of-motion (ROM) tests, that make it difficult to gauge the extent of an injury or track progress during therapy. These manually operated ROM tests are also prone to operator errorand may lack consistency in implementation and practice. Some techniques also require labor- intensive manual procedures involving a laser pointer attached to the patient’s head.

[0057] Fig. 1 illustrates a schematic diagram of the machine learning classification and assessment system 100 for evaluating patients with proprioceptive, mobility, and sensorimotor impairments. The system 100 comprises a computing device 106 having the disclosed ISM assessment application 112, a tracking device 102, and secure cloud-based network 124. The assessment application 112 of the system 100 uses objective data input received from the tracking device 102 capable of generating reliable objective movement data. Various movement data is received from the tracking device 102 and subsequently stored and processed with the computing device 106 during and after the data capture process. Advantageously, clinicians using the system 100 may remotely track an individual’s progress, obviating the reliance on self-reporting to confirm patient compliance.

[0058] In an embodiment, the tracking device 102 is a distinct hardware component of the system 100. For example, the tracking device 102 comprises a sensor unit or sensor 104 having one or more inertial measurement units (IMUs) (or similar) attached to an individual. In an embodiment, the tracking device 102 an adjustable headgear that measures head movement via a Bluetooth-enabled, custom-built sensor 104 that syncs with a computing device 106 or application 112. The tracking device 102 advantageously features a minimalist design that allows a patient to easily progress through physical therapy without excess weight or resistance aggravating their impairment. In an embodiment, the tracking device 102 weighs less than 55g, preferably less than 100g. In a preferred embodiment, the tracking device 102 is directly and operatively connected to the computing device 106 and arranged to monitor the individual without being directly attached to the individual. For example, the tracking device 102 may be a motion capturing device or camera.

[0059] In an embodiment, each sensor 104 contains one or more IMUs that report changes in head angular position by using the IMUs built-in accelerometers and gyroscopes and algorithm. In an embodiment, the tracking device 102 comprises a first sensor 104 attachable to a head, neck, or limb of the human subject and a second sensor (i.e., similar to sensor 104) attachable to a torso or a trunk of the human subject. IMUs of the sensor 104 detect movement in each of the three cardinal axes and can precisely measure three-dimensional rotational velocity, linear acceleration, and magnetic field. The assessment application 112 utilizes a sensor fusion algorithm to derive the resultant rotational position (orientation) and reduce drift. In an exemplary embodiment, the sensor fusion algorithm is implemented via processor (e.g.,processor 114a) embedded in the sensor 104 of the tracking device 102 before communicating the derived data to the assessment application 112 for further processing by the processor 114b on the computing device 106. Using Bluetooth, the measurements are wirelessly transmitted to a computing device 106 and, via the assessment application 112, displayed on a monitor or display 122 the head position and movement in real time. The wireless connection and transmission capability make the system 100 an ideal solution to support patients when practicing their physical therapy exercises in the clinic or at home. In an embodiment, an initial calibration for the tracking device 102 is carried out while the tracking device 102 is stationary (i.e., not moving) and synching to the computing device 106, thus resetting acceleration and angular velocity values. A calibration of the sensor 104, which is specific to the individual, may be carried out by following a predefined set of instructions via the assessment application 112 on the display 122.

[0060] The ISM algorithm processes the objective measurements from the tracking device 102 and produces quantitative ID, 2D, and / or 3D metrics on head-neck movements. By comparing the data received from the tracking device 102, it is possible to compare and identify healthy and unhealthy musculoskeletal movements and generate an objective diagnostic assessment based on the outcome of the assessment. In an embodiment, the diagnostic assessment is presented as a probability or likelihood that the performance of an individual on the assessment test corresponds to a certain impairment or condition. Applying data analysis and machine learning algorithms with the assessment application 112 further increases the quality of the outcome by providing a classification scheme e.g., in the form of digital and printable reports.

[0061] The computing device 106 includes a memory 108, operating system 110, assessment application 112, one or more processor(s) 114, storage 116, input-output interface 118, graphical user interface 120, and display 122. The computing device 106 may represent any suitable type of computer, computing system, server, disk array, or programmable devices such as a handheld device, a networked device, or an embedded device, etc. The computing device 106 may be in communication with one or more networked computers via one or more networks 124, such as a cluster or other distributed computing system, through the I / O interface 118. The I / O interface 118 is configured to transmitted data between the computing device 106 and the tracking device 102 e.g., via wired or wireless connection.

[0062] The processor 114 may include one or more devices selected from processors, microcontrollers, digital signal processors, microcomputers, central processing units, field programmable gate arrays, programmable logic devices (i.e., digital logic circuitry), statemachines, logic circuits, analog circuits, digital circuits, or any other devices that manipulate signals (analog or digital) based on operational instructions that are stored in the memory 108. Memory 108 may be a single memory device or a plurality of memory devices including but not limited to read-only memory (ROM), random access memory (RAM), volatile memory, non-volatile memory, static random access memory (SRAM), dynamic random access memory (DRAM), flash memory, or cache memory. Memory 108 may also include a mass storage device such as a hard drive, optical drive, tape drive, non-volatile solid state device, or any other device capable of storing digital information.

[0063] The processor 114 may operate under the control of an operating system 110 that resides in memory 108. The operating system 110 may manage computer resources so that computer program code embodied as one or more computer programs, such as the assessment application 112 communicatively connected to memory 108 may have instructions executed by the processor 114. In an alternative embodiment, the processor 114 may execute the assessment application 112 directly, in which case the operating system 110 may be omitted.

[0064] The computer storage 116 typically includes at least one hard disk drive and may be located externally to the computing device 106, such as in a separate enclosure or in one or more networked computers, one or more networked storage devices (including, for example, a tape or optical drive), and / or one or more other networked devices (including, for example, a server). The storage 116 may also host one or more databases communicatively connected to data collected and received in association with the assessment application 112, wherein the storage 116 includes a cloud-based secured HIPAA-compliant database. The storage 116 may be included in the computing device 106 or in a cloud-based network 124 (or service 126).

[0065] The graphical user interface 120 may be operatively coupled to the processor 114 of computing device 106 in a known manner to allow a system operator to interact directly with the computing device 106. The graphical user interface 120 may include or be communicatively connected to output devices (e.g., display 122) such as video and / or alphanumeric displays, a touch screen, a speaker, and any other suitable audio and visual indicators capable of providing information to the system operator. The graphical user interface 120 may also include input devices (e.g., tracking device 102) and controls such as an alphanumeric keyboard, a pointing device, keypads, pushbuttons, control knobs, microphones, etc., capable of accepting commands or input from the operator and transmitting the entered input to the processor 114.

[0066] Those skilled in the art will recognize that the computing environment illustrated in Fig. 1 is not intended to limit the solution(s) in the present disclosure. In addition, various program code described herein may be identified based upon the application or software component within which it is implemented in a specific embodiment of the invention. However, it should be appreciated that any particular program or hardware nomenclature that follows is used merely for convenience, and thus the disclosed system and method should not be limited to use solely in any specific application identified and / or implied by such nomenclature. It should be further appreciated that the various features, applications, and devices disclosed herein may also be used alone or in any combination. Moreover, given the typically endless number of ways in which computer programs may be organized into routines, procedures, methods, modules, objects, and the like, as well as the various ways in which program functionality may be allocated among various software layers that are resident within a typical computing system (e.g., operating systems, libraries, APIs, applications, applets, etc.), and / or across one or more hardware platforms, it should be appreciated that the invention is not limited to the specific organization and allocation of program or hardware functionality described herein.

[0067] Fig. 2 illustrates the system 100 further including a service 126 being communicatively coupled to both the computing device 106 and network 124. In some cases, the service 126 can be a deterministic service that operates fully given a set of inputs and without a randomization factor. In other cases, service 126 can be or can include a machine learning (ML) or artificial intelligence engine to enable the service 126 to operate even when faced with a randomization factor. In some implementations, the service 126 is a cloud service operating in a cloud environment. In some implementations, the service 126 is a local service operating on a local device. In some implementations, the service 126 is a hybrid service that includes a cloud component operating in the cloud and a local component operating on a local device. These two components can communicate with one another. The service 126 is generally tasked with operatively connecting the computing device 106 with the network 124 or another computer. In an embodiment, the service 126 functions as a proxy server or intermediary between the computing device 106 and the network 124. In an alternative embodiment, the service 126 functions as a virtual machine. The service 126 may receive incoming data, i.e., a software update, from a network 124.

[0068] Fig. 2 further illustrates the general architecture and testing setup for the system 100. The tracking device 102, i.e., being placed on the head, neck, or limb of a user, is configuredto detect movement in each of the three cardinal axes (x, y, z) and can precisely measure movement in three-dimensional space, e.g., the IMU measures linear acceleration, rotational velocity, magnetic field, and / or other parameters and uses these measurements to calculate rotational movement. The computing device 106 collects movement data from the tracking device 102, and displays the position (corresponding to the head, neck, or limb of the user) and movement as a cursor 130 in real time. The graphical user interface 120 connected to the assessment application 112 displays a test widget 128 for a certain sensorimotor control test, which prompts the user to conduct the sensorimotor control test. The test widget 128 may be a moving target, computer-generated path, or another interface element to map the cursor 130 and corresponding movement of the tracking device 102. The sensorimotor control test instructs the subject to follow a computer-generated path (e.g., test widget 128) with the tracing cursor (e.g., cursor 130) by moving the head, neck, or limb of the subject, wherein the computer-generated path is presented on the display 122 (e.g., monitor) either as a visible path or a target that moves along an invisible path. In an embodiment, the assessment application 112 records and displays a pattern comprising a trajectory path of the tracing cursor, and a correlation and / or deviation between the trajectory path of the tracing cursor 130 and the computer-generated path is subsequently recorded. The correlation and / or deviation can further be analyzed using movement parameters.

[0069] The assessment application 112 compares characteristic data from the individual to a baseline of characteristic data and classifies the characteristic data as healthy or unhealthy. In an exemplary embodiment, reference values are based on measurements that have been taken on a healthy and a whiplash-associated-disorder cohort, age and gender matched. The assessment application 112 may also take into account a Boolean variable specific to a musculoskeletal classification (e.g., neck pain = yes / no) without going further into details regarding type and / or etiology. The movement data variables, e.g., kinematic outcome, may subsequently be connected to answers from clinically validated questionnaires. In an embodiment, the characteristic data is musculoskeletal movement data of an individual.

[0070] The sensor 104, via processor, employs a sensor fusion algorithm to infer, in real time, a three-dimensional head / neck orientation from acceleration, angular velocity, and magnetic field data communicated to the assessment application 112 from tracking device 102. The assessment application 112 preferably comprises a deep artificial neural network (ANN) to automatically evaluate impairment level and progress of performance, and to automatically recommend a rehabilitation or exercise strategy using movement data of the tracking device102. The rehabilitation or exercise strategy is recommended at least based on movement data of the tracking device 102 and may also be based on patient reported data from at least one questionnaire, wherein the patient reported data is manually input into the processor 114. In other words, the system 100 generates a deep-learning-based rehabilitation strategy recommendation engine for the suggestion of individual rehabilitation strategies based on results gathered from the assessment tests. The system may recommend one or more rehabilitative programs or exercises for the individual to complete using the assessment application. The assessment application is further programed to dynamically adjust and update the recommended rehabilitative programs or exercises based on the performance of the individual during and / or after completion of the one or more rehabilitative programs or exercises. In an embodiment, the recommended rehabilitation strategy (or exercises) may be modified based on progress during rehabilitation as intermittently evaluated with the assessment tests.

[0071] As depicted in Fig. 3, the system 100 implements computer-implemented method 103 utilizing a neural network for impairment classification. The ANN is created to automatically evaluate an impairment level based on recorded or derived characteristic data from an individual’s performance on assessment test implemented using the assessment application 112. Following the creation of the ANN, the ANN is trained using characteristic data collected from assessment tests completed by the group of individuals. The training of the neural network may be performed using data collected from individuals that have completed one or more assessment tests. The neural network learns to extract features (i.e., identified as corresponding to normal or impaired conditions) from a newly provided input (i.e., results of a new assessment test), determine whether the features match those of individuals having one or more known impairments, and, via the assessment application, indicate matching results at an output of the neural network. The matching results indicating one or more impairments may be further classified into subclassifications that may be used to recommend a rehabilitative or exercise program for the individual (e.g., after receiving confirmed authorization and recommendation from healthcare professional). Finally, the ANN is validated using known data that has been correctly labeled or classified.

[0072] In an embodiment, the ANN is trained using one or more of the following algorithms: gradient descent, Newton method, conjugate gradient, quasi-Newton method, Levenberg- Marquardt algorithm. One skilled in the art will recognize that other methods to train and implement the ANN may be utilized to train and validate the neural network to improve theaccuracy of classifying impairments and conditions.

[0073] Because the system 100 provides quantitative data on impairment, the ANN assists in determining the degree of impairment based on data collected with the tracking device 102 during execution of assessment tests. This approach of using the assessment application 112 with the ANN avoids a significant level of manual intervention required to interpret the collected data.

[0074] The data gathered by the assessment application 112 can be used to develop a machine learning based approach using deep ANNs to automatically evaluate impairment level and progress and recommend a rehabilitation strategy using the movement data obtained by the tracking device 102 and, optionally, patient reported data from questionnaires. In an embodiment, the ANN approach is trained using the ground truth assessment test patterns collected from multiple individuals labeled with 1) a condition or diagnosis for each individual (e.g., normal, or having a first impairment, or having a second impairment, or have first and second impairments, and so on) and 2) the type of established assessment test(s) or exercise(s) completed by the individual corresponding to the labeled condition or diagnosis. The ANN may also be trained with or without unlabeled patterns.

[0075] The training of the ANN is such that the cross-entropy based loss function allows for both labelled data and unlabeled data and uses consistency training or a semi-supervised training strategy in order to allow for the use of the more readily available unlabeled measurements. The input into the deep ANN will include the raw trajectories measured by the tracking device 102 and the output is a classification layer determining the individual’s rehabilitation needs (i.e., proprioceptive deficiencies, or range of motion) and suggesting certain rehabilitative exercises, executable by the assessment application 112 of the system 100, specifying the level of difficulty for these exercises and numbers of repetitions. Advanced data augmentation methods designed for the assessment tests are employed to facilitate a semisupervised training of the ANN wherein simulations of additional assessment tests may be used to enrich the collected data. Parts of the labelled dataset may also be used as a blinded test set to validate the sensitivity and the specificity of the recommendation engine. Thus, the assessment application 112 comprises an Al-based individual therapy protocol engine configured to generate on-the-fly adjustments of a proposed therapy plan according to an individual’s performance and progress corresponding to the assessment test(s) and predefined or previously classified impairments.

[0076] For individuals performing the assessment tests and receiving regular assessments by a physical therapist, rehabilitation progress of the individual can be monitored both by the physical therapist and assessment application 112. The rehabilitation strategy may be automatically adjusted according to the individual’s performance improvement during each subsequent assessment test (or rehabilitative exercise).

[0077] The assessment application 112 provides a novel method for the evaluation and recommendation of rehab approaches based on the patient’s movement data. During an individual’s rehabilitation process there may either be a need to immobilize or to mobilize an area of interest of the individual (e.g., neck) at a certain time. In the same way, individuals may either need to improve their movement control, proprioception, or range of motion. The system 100 offers exercises via the assessment application 112 that train these parameters e.g., of the neck. The proposed assessment application 112 helps in determining the right type of exercises, difficulty- and intensity levels for each individual, taking into account some or all of the following: (1) the clinical assessment by a physical therapist, (2) the assessment tests using the system 100, and (3) the individual’s reported data, such as responses to the validated Neck disability index questionnaire and pain rating. By providing a first classification of a patient’s rehabilitation needs and an individual therapy plan that will complement other therapy prescribed by an attending physical therapist, the proposed assessment application will include an artificial-intelligence-based recommendation engine for the (semi)-automatic, individual and adaptive rehab plan creation based on patient data and changes in the data that can be observed during the course of the patient’s rehabilitation.

[0078] By comparing data from healthy subjects with data from patients, the ISM assessment application 112 can accurately classify asymptomatic cases and accurately identify those suffering from impairment. Using the tracking device 102, along with data analytics and machine learning algorithms of the assessment application 112, a greater variety and accurate assessments / diagnostics of abnormal musculoskeletal or neuromuscular conditions can be classified and / or grouped by similar impairments.

[0079] In some impairment cases insights into the state of the individual’s ability to control movements can reveal certain aspects of the type of impairment and which rehabilitation path is most appropriate. To evaluate sensorimotor or neuromuscular control, a specialized movement procedure test is conducted. Range of tests can be applied in this setting and the following is an example of one such test and described as follows:

[0080] A human subject is seated in front of a computer monitor wearing the tracking device 102. The subject is instructed to visually track the movement of the test widget 128 on the graphical user interface 120 as it moves on the display 122, following different movement trajectories, visible or invisible, ranging from easy, medium, and difficult.

[0081] Fig. 4 shows an example of predefined movement patterns appearing in random order at three difficulty levels: easy pathway 132, medium pathway 134, and difficult pathway 136. Different shades indicate variations in the test widget 128 velocity, the easy pathway 132 having slower instances of velocity variation and the difficult pathway 136 having faster instances of velocity variation. As depicted, the ‘easy’ pathway 132 has less variation in velocity and the ‘difficult’ pathway 136 has greater variation in velocity.

[0082] Referring to Fig. 5, during the sensorimotor control test, the tracking device 102 continuously measures changes to the head’ s orientation as the subject follows the test widget 128. Among other things, the assessment application 112 records rotational data at a predefined frequency (Hz), derived from angular velocity and acceleration etc. of the tracking device 102. This recorded movement data is stored and processed using data analysis and machine learning algorithms by means of statistical and machine learning techniques (e.g., via deep artificial neural network) that uses movement data and derived parameters to enhance the accuracy and quality of the assessment / diagnostics.

[0083] The assessment application 112 objectively evaluates a subject’s ability to control their head and neck as they follow the test widget 128 in the defined sensorimotor control test. In this particular test, the assessment application 112 projects the rotational displacement captured by the tracking device 102 onto the 2D plane coinciding with the screen surface of the display 122 (e.g., monitor). Using this projection, the assessment application 112 can then compare the path of the test widget 128 with the path traced by the cursor 130 that corresponds to the test subject’s movement. As observed in Fig. 6, by plotting overlays of these paths, it is easy to see differences in the performance of an asymptomatic test subject and a test subject with a neck injury. The level of deviation between the healthy subject for easy, medium, and difficult paths 132a, 134a, 136a, of testing and the impaired subject for easy, medium, and difficult paths 132b, 134b, 136b, has then been objectively quantified. The graphical user interface 120 may present a miniplayer view 138 of the user and / or tracking device 102 on the display 122 during the sensorimotor control test.

[0084] Referring again to Fig. 6, visualizations of the quality of the performed movementtrajectories (e.g., paths 132a, 134a, 136a, 132b, 134b, 136b) are generated. The assessment application 112 computes multiple statistical metrics to accurately quantify differences between asymptomatic and symptomatic subjects by taking into account key metrics derived from the objective measurements and the subject’s performed movements, or the average difference between the test widget 128 and the subject-control cursor 130 over the entire duration of the test. The assessment application 112 also computes the percentage of time the cursor 130 is on, or in close vicinity of the test widget 128. This includes both undershoots and overshoots — the proportion of time spent behind or ahead of the test widget 128 respectively.

[0085] Other parameters that are derived, stored and used as part of the calculation may include the smoothness of movement, a parameter that quantifies jerkiness of movement based on the integration of the quadratic sum of the third derivative of the spatial coordinates traced by the subject normalized against the same quantity traced by the target.

[0086] The system 100 and corresponding assessment application 112 enables healthcare professionals to standardize clinical evaluations and make data-driven decisions when designing rehabilitation regimens. The remote performance tracking functionality of the system 100 encourages compliance to prescribed home therapy and allows for progress tracking and rehab plan adaptation in between appointments. The system 100 thus enables clinicians to better attend to individuals with limited access to physical therapy clinics. Consistently executed therapy exercises and targeted rehab plans ensure shorter recovery periods while improved patient outcomes increase the quality of life of impaired individuals and will ultimately reduce the heavy socioeconomic burden associated with the global healthcare problem of musculoskeletal and neuromuscular impairment.

[0087] By using machine learning (or artificial intelligence) to classify test subjects into “impairment” categories based on their test results, a variety of machine learning models included in the assessment application can be trained with a data set consisting of multiple parameter variables. And by applying a statistical classification algorithm on multiple movement parameters, e.g., corresponding to a variety of difficulty levels, enables the system 100 to accurately define a level of accuracy ratio expressed as a fraction of 100 (%) or the indicative equivalent thereof.

[0088] By applying and performing the same or similar ISM methods (non- automated or automated) of the assessment application 112, the assessment application 112 can be utilized to assess and diagnose a range of other manifestations of conditions affecting the muscular,skeletal, and nervous system of individuals. By using feature ranking capabilities of the assessment application 112 to identify the specific features that were most important to the classification ruling enables for more efficient identification.

[0089] The resulting comparison model obtained from the classification algorithm of the assessment application 112 enables users to accurately determine the classifications based primarily on the primary features by ranking them accordingly.

[0090] The assessment application 112 enables for the objective classification of subjects by the level or type of their impairment. Continuing data collection facilitates for the development of the outcomes of the assessment application 112 to support the creation of new objective musculoskeletal or neuromuscular assessments / diagnostic applications by re-applying the fundamentals of the ISM objective musculoskeletal assessment process by applying new datasets.

[0091] Fig. 7 provides a simplified flow chart for how the method of classification operates within the system 100. As depicted, the present disclosure relates to a computer-implemented method for classifying impairments and generating rehabilitation strategies for a human subject, comprising the steps of (1) obtaining characteristic data from an individual using the tracking device 102 and assessment application 112, (2) processing the characteristic data corresponding to the tracking device 102, (3) comparing the characteristic data from the individual to a baseline of characteristic data, and (4) classifying the characteristic data as normal (e.g., healthy) or impaired, wherein an impaired classification is defined as relating to one or more impairments identified from previously obtained characteristic data from a group of individuals, (5) further classifying impaired classification results into subgroups having one or more specific conditions or characteristics, (6) recommending a rehabilitative or exercise program based on the classification of impairment and / or subgroup classification, (7) presenting the rehabilitative or exercise program to the individual via the assessment application 112 (and optionally display 122), and (8) dynamically modifying the rehabilitative or exercise program based on the individual’s progress during or after completion of the rehabilitative or exercise program.

[0092] The ISM technology of the assessment application 112 can be described as cloud-based Internet of Healthcare Things (IOHT) application connected to a wired or non-wired device. The wired or wireless recording devices may include the following: computerized tomography (CT-Scan); computers of various sizes and operational capabilities; high-energyelectromagnetic radiation X-ray; inertial measurement unit (IMU); IT tablets of all shape, sizes and computing power; microwave diathermy; mobile phones; Fit Bits; optical sensors; radar operated devices using radio waves to determine the distance angle and radial velocity of individuals; thermal imaging; and video cameras or devices that are capable of recording musculoskeletal movements of individuals.

[0093] The disclosed system 100 adds versatility to rehabilitation for both impaired individuals (e.g., patients) and practitioners. The system 100 advantageously offers remote capability to accommodate telemedicine rehabilitation applications. Patients may carry out clinically verified assessment tests in the comfort of their own homes. This, in turn, reduces the number of face-to-face clinic visits needed to ensure therapeutic effectiveness. The system 100 also adds versatility for the patient by reducing the cost associated with outpatient physical therapy. The cloud storage capability (e.g., network 124) allows clinicians to remotely access patient data, easily review sequential therapy sessions to assess patient progression or regression and modify the therapy regimens in response to patient performance. This versatility for patients to perform their therapy at home and the ability for clinicians to track these at-home therapy sessions may also improve patient compliance. The regular, consistent movement of an impaired joint in a controlled manner leads to better patient outcomes. Thus, when patients perform their therapy more frequently (i.e., as a consequence from implementing a simplified, accessible system 100) there is an increased likelihood that their therapy will remediate their impairment and relieve their pain.

[0094] It is to be understood that even though numerous characteristics and advantages of various embodiments of the present disclosure have been outlined in the foregoing description, together with details of the structure and function of various embodiments thereof, this detailed description is illustrative only, and changes may be made in detail, especially in matters of structure and arrangements of parts within the principles of the present disclosure to the full extent indicated by the broad general meaning of the terms in which the appended claims are expressed.

Claims

CLAIMS1. An Individualized Signature Mapping (ISM) system (100) for identifying deviations and abnormalities from musculoskeletal movements in an individual, the system (100) comprising: a tracking device (102) arranged for monitoring biometric characteristics of the individual, the biometric characteristics being preselected for further processing and classifying as relating to one or more impairments or normal conditions; a first computing device (106) having an assessment application (112), a processor (114), an interface (118) for communicating with the tracking device (102), and one or more hardware storage devices (116); wherein the one or more hardware storage devices (116) store instructions that are executable by the system (100) to: initiate an assessment test on the assessment application (112); obtain characteristic data from the individual using the tracking device (102), the characteristic data being defined as neuromuscular and / or musculoskeletal data specific to an individual; compare the characteristic data from the individual to a baseline of characteristic data stored on a secure cloud-based network (124), the characteristic data from the individual corresponding to a real-time performance of the individual on the assessment test and the baseline corresponding to previous performances of a group of individuals on the assessment test of the assessment application (112); and classify the characteristic data as normal or impaired based on the compared characteristic data from the individual to the baseline; wherein an impaired classification is defined as relating to one or more impairments identified from previously obtained characteristic data from the group of individuals.

2. The ISM system (100) according to claim 1, wherein the processor (114) generates an objective diagnostic assessment based on classified characteristic data.

3. The ISM system (100) according to claim 2, wherein the assessment application (112) generates the objective diagnostic assessment using data analytics and comprises one or more machine learning algorithms to evaluate range of motion, proprioception, sensorimotor, and neuromuscular control related to the classified characteristic data.

4. The ISM system (100) according to claim 3, wherein the first computing device comprises a monitor for displaying the assessment application of the first computing device.

5. The ISM system (100) according to claim 4, wherein the assessment application generates a specialized movement procedure test for the individual to perform using the tracking device (102).

6. The ISM system (100) according to claim 5, wherein the assessment application (112) includes a deep artificial neural network (ANN) to automatically evaluate an impairment level based on performance on the assessment test.

7. The ISM system (100) according to claim 5, wherein the assessment application (112) includes a deep artificial neural network (ANN) to automatically recommend a rehabilitation or exercise strategy using the classified characteristic data.

8. The ISM system (100) according to claim 1, wherein the tracking device (102) includes a first sensor unit (104) attachable to the individual and further including a second processor (114b) for calculating displacement of the tracking device (102) based on measurements obtained by an inertial measurement unit (IMU) housed within the tracking device (102).

9. The ISM system (100) according to claim 1, wherein the first computing device (106) includes a graphical user interface (120) operatively coupled to the processor (114) and arranged for interacting with the individual by means of a display (122).

10. The ISM system (100) according to claim 7, wherein the assessment application (112) is arranged to dynamically adjust and update the recommended rehabilitation or exercise strategy based on the performance of the individual during and or after completion of the one or more assessment tests.

11. The ISM system (100) according to claim 1, wherein the assessment application (112) includes feature ranking capabilities to identify specific features from the characteristic data as being most important to impairment classification.

12. A method of assessing and classifying musculoskeletal- or neuromuscular-related impairments of an individual using an Individualized Signature Mapping (ISM) system (100) including a first computing device (106) having one or more processors (114b), a tracking device (102), one or more storage devices (116), and interactive assessment application (112), the method comprising: collecting characteristic data from a group of individuals using the tracking device (102) and assessment application (112), the characteristic data being defined as neuromuscular and / or musculoskeletal data specific to the individual; creating an artificial neural network (ANN) to automatically evaluate an impairment level based on recorded characteristic data from an individual’s performance on assessment test implemented using the assessment application (112); training the ANN of the system (100) using characteristic data collected from assessment tests completed by the group of individuals; validating the ANN of the system (100); updating the assessment application to implement the ANN to compare characteristic data from the individual to a baseline of characteristic data stored on a secure cloud-based network (124), the characteristic data from the individual corresponding to a real-time performance of the individual on the assessment test and the baseline corresponding to previous performances of a group of individuals on the assessment test of the assessment application (112); andclassifying the characteristic data from the real-time performance of the individual as normal or impaired based on the compared characteristic data from the individual to the baseline.

13. The method according to claim 12, wherein during the step of training the ANN of the system (100), the ANN is arranged to: extract features identified as corresponding to normal or impaired conditions from a newly provided input, determine whether the features match those of individuals having one or more known impairments; and indicate matching results at an output of the neural network.

14. The method according to claim 13, wherein the matching results indicating one or more impairments are further classified into subgroups corresponding to one or more impairments and / or characteristics.

15. The method according to claim 12, further comprising the step of generating a classification layer determining rehabilitation needs of the individual and suggesting certain rehabilitative exercises based on a predicted impairment level and / or group biometric characteristics of each cluster identified in the classification layer.

16. The method according to claim 15, further comprising the step of specifying a level of difficulty for the rehabilitative exercises and numbers of repetitions.

17. A method for assessing and classifying musculoskeletal or neuromuscular-related impairments of an individual on a first computing device (106) having one or more processors (114b) and one or more storage devices (116), the method comprising the steps of: transmitting characteristic data from a tracking device (102) attached to an individual to the first computing device (106);providing, on an output display (122) and by means of an assessment application (112), an assessment test arranged for being completed by the individual with the tracking device (102) for monitoring performance by the individual during completion of the assessment test; comparing characteristic data from the individual to a baseline of characteristic data stored on a secure cloud-based network (124), the characteristic data from the individual corresponding to a real-time performance of the individual on the assessment test and the baseline corresponding to previous performances of a group of individuals on the assessment test of the assessment application (112); and classifying the characteristic data as normal or impaired based on the compared characteristic data from the individual to the baseline.

18. The method according to claim 17, further comprising the step of using an artificial neural network (ANN) with the assessment application (112), automatically evaluating an impairment level of the individual.

19. The method according to claim 17, further comprising the step of using an artificial neural network (ANN) with the assessment application (112), generating a rehabilitation or exercise strategy using the characteristic data obtained by the tracking device (102), and presenting the rehabilitative or exercise program to the individual using the assessment application (112) by means of the display 122.

20. The method according to claim 19, wherein the rehabilitation strategy is generated based on both movement information of the tracking device and patient reported data from at least one questionnaire, the patient reported data being manually input into the processor.

Images