WO2016141473A1

WO2016141473A1 - System and method for quantifying flow of movement and changes in quality of movement

Info

Publication number: WO2016141473A1
Application number: PCT/CA2016/050247
Authority: WO
Inventors: Ladan A. MAHABADI; Doina Precup
Original assignee: Tandemlaunch Inc.
Priority date: 2015-03-10
Filing date: 2016-03-08
Publication date: 2016-09-15

Abstract

A system and method are provided for encoding physical movement, quantifying changes in fundamental characteristics of physiological movement such as fluidity of movement in complex bio-physiological systems. Such changes in characteristics of physiological mobility can be monitored over time, during the course of regiments, progression of systematic changes, or daily routine activities. The system utilizes various modules and interactions between these modules for encoding movement in a general, adaptive manner, quantifying contextual changes in fundamental characteristics of movement such as its fluidity, and comparing movements for a particular agent or across a population of agents over time.

Description

SYSTEM AND METHOD FOR QUANTIFYING FLOW OF MOVEMENT AND CHANGES IN

QUALITY OF MOVEMENT

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Patent Application No.

62/130,731 filed on March 10, 2015, the contents of which are incorporated herein by reference.

TECHNICAL FIELD

[0002] The following relates to systems and methods for quantifying flow of movement and changes in quality of movement.

DESCRIPTION OF THE RELATED ART

[0003] Although there exist systems that measure particular features of movement such as number of steps and duration of movement during a particular period of time, such systems have been found to be lacking in quantifying the flow of movement.

BRIEF DESCRIPTION OF THE DRAWINGS

[0004] Embodiments will now be described by way of example only with reference to the appended drawings wherein:

[0005] FIG. 1 is a block diagram of a thematic movement analyzer system;

[0006] FIG. 2 is a block diagram of a configuration for the system shown in FIG. 1 ;

[0007] FIG. 3 is a block diagram of another configuration for the system shown in FIG. 1 ; and

[0008] FIG. 4 is a flow chart illustrating example computer executable instructions for monitoring movement.

SUMMARY

[0009] A method and system are herein described which use thematic structures of movement to quantify flow of movement for mobile agents, and build a comparative graphical representation of movement used for temporal analysis of movement for a single or multiple users over time, in the context of daily or regimental physical movement.

[0010] In one aspect, there is provided a method of monitoring movement of an agent, the method comprising: using spatio-temporal measurements from one or more sensing devices to determine at least one movement associated with the agent; generating a movement profile based on the at least one movement; and using the movement profile to monitor the movement of the agent.

[0011] In another aspect, there is provided a system for measuring movements of an agent, the system comprising: an agent module to enable spatio-temporal measurements to be obtained from a sensing device; a shape module to determine at least one movement associated with the agent; a movement profile generation module generating a movement profile based on the at least one movement; and a comparison module for using the movement profile to monitor the movement of the agent.

DETAILED DESCRIPTION

[0012] It has been recognized that typical physical movement is not random, and there is a semantic relationship between consecutive shapes. That is, the progression of shapes in time serves a particular purpose for the agent (e.g., human, animal, object, etc.) moving in space over some period of time.

[0013] For instance, there is a theme (e.g., walking or running) that underlies the evolution of the shapes resulting in the movement of an agent from its initial state (e.g., from a stationary pose in location A) to another state (e.g., to a stationary position in location B which is a number of steps away from the initial position A). In other words, typical movements are coherent, and random movements are considered to be an extreme case, wherein there is little or no semantic relationship in the progression of shapes.

[0014] The system described herein constructs a dynamically adaptive notion of activity for each user using topic modeling (e.g., see Hoffman, Bach et al. 2010) and compares various movement profiles by extracting movement themes from the local movement graphs (i.e., movement profiles constructed by a movement generator, and a dictionary adaptive developed by a particular module). Based on themes extracted, a measure of fluidity is determined. The system described herein is particularly advantageous in the unique application of these principles to the context of movement, in order to encode movement, activities and fluidity of movement.

[0015] That is, the systems and methods described herein recognize that topic modeling frameworks can be applied to sensor data in order to discover movement profiles.

Topic modeling frameworks have previously been used in text mining and natural language processing (NLP), where inputs are words, sentences, documents, and corpus of documents. Topic modeling discovers multiple themes or subjects in the documents and assigns a probability to each theme or subject based on the occurrence frequency of words.

One example of topic modelling is Latent Dirichlet Allocation (LDA). The system described herein applies topic modeling, such as LDA, to sensor data rather than language processing. For example, movement primitive extraction can be performed using LDA applied to wearables. Sensor data is collected using multiple interconnected sensors or an array of sensors that, when attached to a body, allows the recording of multi-point movements on a body. The sensor data would then be used to infer shapes, movements, and movement profiles; and compare those to an ideal or "gold" standard.

[0016] As used herein, an "agent" can be considered any physical entity or system whose movements in space can be tracked over time. The system described herein characterizes physical movement of any such agent over time. The system also

dynamically learns what movement means on both a global (e.g., population level) and a local level (e.g., a particular agent during some interval of time). Related applications include various health monitoring applications. The system also rigorously characterizes movements using latent thematic structures in the context of physical activities but independent of any labeled activity, context or time.

[0017] The thematic structure latent in movement data can be used to build probabilistic generative and predictive models which can be used to model physical movement as the evolution of a single or a collection of points from one position in the 3-dimensional space to another, in some particular period of time.

[0018] The system also provides methods and apparatus for encoding physical movements, quantifying changes in fundamental characteristics of physiological movement such as fluidity of movement in complex bio-physiological systems - such as human beings and animals, or in robotic systems that can move. Such changes in characteristics of physiological mobility can be monitored over time, during the course of regiments, progression of systematic changes (e.g., particular mobility-affecting pathologies or mechanical wear and tear), or daily routine activities.

[0019] One example use of the described system is by medical practitioners to monitor and quantify patients' movement recovery over time. Another example is for trainers or professional athletes to establish standards or targets that can be used as categories to monitor training individuals' progress towards a higher grade in the standard.

[0020] The system described herein utilizes various modules and interactions between these modules for: 1 ) encoding movement in a general, adaptive manner, 2) quantifying contextual changes in fundamental characteristics of movement such as its fluidity, and 3) comparing movements for a particular agent or across a population of agents over time. [0021] Potential venues of application benefiting from the adaptation of this system include but are not limited to:

[0022] · A shape sensing device (e.g., placed on the back, arms or legs) to track shape evolutions in short or long periods of time to build probabilistic generative and predictive models. Such models can be used to detect sequences of shapes ordered in time that are thematically related (i.e., activity recognition), quantify various characteristics of movement such as dexterity, fluidity and complexity, devise a causal model of activities and

occurrences of critical episodes (e.g., pain incidents), and quantify recovery in terms of variations in complexity, dexterity and fluidity of movement for a user overt various windows of time.

[0023] · A wearable system used to monitor variations in the characteristics of movement of athletes during training or competition. For instance, in the context of running, this system can be used to monitor the quality of movement and provide a causal view of the occurrences of boundary cases (e.g., exhaustion, muscle pain, etc.); in the context of skiing, it can monitor movement, competitive technique and improvements in a larger context of daily training and physical activity; or in the case of swimming, this invention can monitor the variation in the range of movement to model changes over time.

[0024] · A system that can categorize users' movements and physical fluidity by comparing movement profiles to existing "gold" standards pre-computed for various segmentations of a particular agent population. For instance, in the context of physical rehabilitation, patients are categorized according to the severity of their immobility (e.g., severely immobile, intermediate, advanced). This system can categorize a new patient and adapt this categorization in the course of the patient's recovery journey. As another example, consider athletic training: This system can categorize athletes based on their movement distance to particular representative athletes for each category (e.g., novice, intermediate, professional). Finally, consider the case of monitoring degradation in movement (e.g., neurodegenerative pathologies that affect mobility such as Parkinson's disease). In this example, this invention can categorize and monitor changes in patient classification as low, medium or high risk of deterioration in mobility.

[0025] Turning now to the figures, FIG. 1 illustrates a thematic movement analyzer system 10 (the "system 10" hereinafter). The system 10 includes an agent module 12 associated with a user or "agent". The agent module 12 includes an action generator 14 and an agent repository 16, which are used to record, store, and transmit spatio-temporal measurements as discussed below. The agent module 12 can communicate with a social repository 18 such as a social network system to encode, store, transmit, and share data associated with the agent to social networks. The agent module 12 also communicates with a shape module 20 to trigger measurements.

[0026] The system 10 also includes a shape module 20. The shape module 20 includes or is connectable or coupled to a wearable sensing device 22 to determine movements and provide data to a movement shape translator 24. The movement shape translator uses sensed shape data to construct curves for generating profiles as discussed below.

[0027] The system 10 also includes a movement profile generation (MPG) module 26. The MPG module 26 includes a global shape dictionary generator 28 for generating representations of movement shapes to be stored in a global shape repository 30. The MPG module 26 also includes a local movement shape generator 32 to create, maintain and update a movement profile for a particular agent.

[0028] The system 10 also includes a comparison module 34 to compare and contrast properties of movement profiles for agents (i.e. same agent over time), or between agents for various applications.

[0029] While the comparison module 34 can be responsible for generating

comparisons, it can be appreciated that once a comparison has been generated, and it is determined that a particular agent is closer to a representative agent of a particular type (e.g., an athlete of a certain expertise), a reporter/assessment/recommender module (not shown) can be used to report back to the user that they're close to a particular

representative group. In some implementations, such a module can be used to recommend a set of exercises or movements that could further improve the agent's status. For instance, if the comparison module 34 determines that certain shapes are missing from the current agent compared to the representative agent, then perhaps exercises that contain those shapes could be recommended.

[0030] Another example is in a rehab application. It has been observed that not all patients have the same level of mobility difficulty, and typically not all are equally fit.

Observing that a patient's physical aptitude can be categorized as novice, intermediary and fit, various applications can be constructed. For example, the comparison module 34 can be used to compare an agent's (in this case patient's) initial assessment to three representatives 1 . Novice, 2. Intermediary, and 3. Fit. Next, the comparison module 34 can be used to determine where the patient stands, and adjust the physical regimens according to their initial assessment. The comparison module 34 can therefore continuously compare the patient to the representatives to dynamically determine the physical aptitude of the patient. The comparison module 34 can also be used to determine the efficacy of the rehab physical regimens based on comparison of the patient at the starting point to a later point in time.

[0031] The configuration of the system 10 shown in FIG. 1 is for illustrative purposes only and it will be appreciated that the modules can be configured in various ways to suit particular applications, for example, certain modules can be situated on the "client side" - i.e. the user or agent side, while other modules operate on a server side, e.g., in a cloud-based or other networked environment. FIGS. 2 and 3 illustrate other example configurations and it will be appreciated that these are non-exhaustive.

[0032] FIG. 2 illustrates one example in which each agent includes or otherwise wears or supports or carries the agent module 12 and the shape module 20. The agent module 12 and shape module 20 are coupled to each other, e.g., by connecting a sensing device 22 to software loaded on a smart phone or other wearable device operating to provide the functionality of the agent module 12 and movement shape translator 24. Each agent is communicably connectable to one or more networks 50 such as WiFi, cellular, or other communication networks, to send and receive information to/from a server entity operating to provide the functionality of the movement profile generation module 26 and the comparison module 34. In this configuration, operations requiring interaction with the agent are performed at the agent while other computations, comparisons, and updating/synchronizing occurs at the server entity to offload processing and storage requirements of the agents and their devices.

[0033] FIG. 3 illustrates another configuration in which each agent carries or otherwise supports the agent module 12, shape module 20, MPG module 26, and comparison module 34, e.g., to have access to these modules even when a network connection is unavailable. Each agent can communicate with a global server 52 over one or more networks 50 in order to periodically update the global shape repository 30, configuration settings or files, etc. In this way, the agent can receive feedback and/or instructions from the MPG module 26 and comparison module 34 "on site" without requiring a direct or persistent connection to a server. For example, an athlete that is training or a patient that is being monitored may primarily utilize data concerning itself and thus can benefit from local operations with periodic updates with the global server 52.

[0034] Further detail regarding the modules shown in FIGS. 1 -3 is provided below. [0035] As indicated above, an agent is a physical system that can (or has equipment or devices that can) record, store and transmit spatio-temporal measurements, over some period of time. Examples include but are not limited to human beings, animals and robots, with an attached wearable sensing measurement devices. Such agents are physically mobile, complex systems that generate thematically related shapes in time.

[0036] The agent module 12 is an input module that processes spatio-temporal measurements from the - internal or external - sensing component therein. The length of the time interval of recordings can vary according to the application utilizing the system 10.

[0037] The agent module 12 maintains an agent repository 16 of states of the agent's movement. The agent's movement profile will be created and updated through the local movement shape generator 32 in the MPG module 26. This repository 16 maintains information that gets updated over time about the agents' movements undertaken, and particular properties of the movements.

[0038] The agent module 12 can also communicate with external systems or repositories that encode, store or transmit information to social networks 18 or repositories. Such social systems 18 may be created and maintained by an external source (e.g., a social network, health club or physical training facility) - to enable agents to view, post, and otherwise socially interact with other agents.

[0039] The agent module 12 therefore maintains the agent repository, triggers the shape module 20, receives agent state information and updates for the agent repository from the MPG module 26, sends queries and receives information from the comparison module 34, and communicates with a social repository 18, which may be maintained by an external source (e.g., some social network site).

[0040] The shape module 20 includes or communicates with a wearable sensing device 22, and includes a movement shape translator 24. The wearable sensing device 22 - shown in a dashed box in FIG. 1 - may be an external sensing module (e.g., a wearable sensing bracelet, device or garment). The movement shape translator 24 constructs a multivariate curve from streaming sensed data. The shapes are forwarded into the MPG module 26.

[0041] The MPG module 26 uses probabilistic graphical models to encode movement, analyze both local and global properties of moves, and extract significant movement themes for each agent's history and profile of movement. [0042] The role of the MPG module 26 is to automatically construct, maintain and update the following:

[0043] 1 ) a global view of possible movement shapes, and

[0044] 2) a localized probabilistic view of movements and activities for each agent.

[0045] The MPG module 26 captures these using a global shape dictionary generator 28, and a local movement shape generator 32 respectively. These generators 28, 32 contribute to, retrieve from, and update shape values stored in a global shape repository 30. The system 10 therefore constructs movement profiles - both at global and local view - with varying scales of granularity.

[0046] The global shape dictionary generator maintains a concise representation of all shapes in a "universe of movement" observed thus far by the system 10, by any agent observed thus far. This component has direct access to the global shape repository 30. As movements from various agents are observed, the collection of possible movements - stored in the global shape repository 30 - expands. The stored dictionary of shapes is multi- resolution. The global shape dictionary generator is particularly advantageous in that it encodes the observed shapes as a new multivariate curve in the dictionary 30, only if no existing shape in the dictionary is close enough to the current shape. That is, the global shape dictionary generator 28 is configured to encode a dictionary of shapes for each resolution of analysis - threshold e and determines whether or not a shape is added to the dictionary of shapes.

[0047] In other words, new movement shapes streaming in will be either represented directly in the dictionary, or they are represented as a neighboring shape already in the dictionary. The global shape dictionary generator 28 has the capability of storing various dictionaries parameterized by a parameter of resolution.

[0048] The local movement shape generator 32 is used to create, maintain and update a movement profile for a particular agent. A movement profile is a graphical model and can include: 1 ) a list of movement shapes observed for this agent, and 2) a set of weighted edges that are updated as movement sequences are observed over time. Each shape in the movement graph is a representative shape of all shapes that are within a threshold determining the resolution granularity. Distances between various shapes are computed using the comparison subcomponent, which uses discrete Frechet distance (Eiter and Mannila 1994). [0049] For every new shape to be added to the global shape repository 30, the local movement shape generator 32 sends an update flag to the global shape dictionary generator 28 to update the shape repository 30. The movement profile generation module builds a movement graph from the set of representative shapes and the transitions between them. Each movement graph has an associated scale of resolution, and a probabilistic graphical model, denoted by G . The local movement shape generator 32 updates this movement graph as moves are observed by the system 10.

[0050] The MPG module 26 determines and maintains likelihoods of various movements, for each agent. This is achieved by the local movement shaper generator 32, which assigns asymmetric weights from moving from shape _έ to 5) for each movement profile graph. The stochastic weight generation performed by the local movement shaper generator 32 determines such transitional stochastic weights. The system 10 uses this determination to automatically update the stochastic weights - for each particular agent - as physical activities are observed over time.

[0051] As such, the local movement shaper generator 32 can be configured to: 1 ) maintain and update shapes (i.e., vertices of a movement graph) for each agent, and 2) update and maintain stochastic weights between the shapes (i.e., stochastic transitional weights). In other words, the local movement shaper generator 32 builds a graphical representation of movement (i.e., a movement profile) for each agent, communicates these to the global and local storage modules shown in FIG. 1 , and keeps updating the stochastic weights as more movements are observed.

[0052] The movements are stochastic because we the weights on the edges are used to calculate flow - as a function of the likelihood that a particular shape is followed by its most immediate past neighbor and the difficulty of the move. This is done so that rigid movement is characterized as movements that are not likely to be seen immediately after each other and that their Frechet distance is large (e.g., imagine a choppy, rigid sequence of movements).

[0053] The idea of applying graphical models to represent physical movement shapes - for individual agents or populations thereof - to study the local and global characteristics of such networks is achieved using the system 10 described herein. The system 10 can therefore study temporal changes in movement while using such graphs as underlying representations.

[0054] The comparison module 34 uses the various methods described herein (e.g., the comparison of movement and movement profiles relying on the graphical model described above, and structural themes extracted using topic modeling techniques) to compare and contrast various properties of the movement profiles for the same agent over time, or different agents that are to be compared for some underlying common purpose. Observed movements of an agent (described as shapes) are treated as a repository and topics are extracted. These themes are compared to his/her or other agents' themes using the below- described KL divergence measures. The comparison module 34 quantifies variations in a movement profile generated by the local movement shape generator 32 and learns thematic patterns from a movement profile by exploring its movement history, and characterizes fundamental movement properties such as fluidity and range of movement.

[0055] As an example application, given gold standards for multiple categories such as those used by professional athletes, healthy individuals, etc. the comparison module 34 can categorize a movement profile as such.

[0056] The comparison module 34 can be used to: 1 ) analyze changes in the movement graph at varying resolutions of granularity, 2) use information divergence measures to compare the likelihood of movement profiles for different agents, and 3) utilize diffusion kernels (Kondor and Lafferty 2002) and Kullback-Leibler divergence measures (Kullback and Leibler 1951 ) on movement graphs to compare different movement themes extracted for a particular agent - or in comparison with another one.

[0057] In the latter case, the comparison module 34 uses topic modeling applied to the graphical encoding of movement (i.e., movement profiles constructed by the local movement shape generator 32 and global shape dictionary generator 28.

[0058] The comparison module 34 can measure characteristics such as the number of shapes undertaken in a particular period of time, the maximal range of distance between shapes (i.e., range of movement), etc. It can construct a dynamically adaptive notion of activity for each user by extracting movement themes from the local movement graphs (i.e., movement profiles constructed by the local movement shape generator 32 and the dictionary adaptive developed by the global shape dictionary generator 32.

[0059] The comparison module 34 uses latent thematic structures of shape

progressions observed captured in local and global patterns in movement graphs to automatically learn about quality and changes in movement. It automatically extracts system thematic patterns in movement - which may or may not directly relate to particular physical activities (e.g., walking) - and explores the history of movement to determine changes in fluidity, complexity and mobility for each user. [0060] Turning now to FIG. 4, an example flow chart is shown which illustrates operations performed by the system 10 in a typical measurement and comparison process.

[0061] At 100 the agent module 12 obtains spatio-temporal measurements from the wearable sensing device 22 and triggers the shape module 20 at 102 to translate the measurements into shape movements at 104 (e.g., as discussed above). The shape module 20 then provides translated data to the MPG module 26 at 106, which is used by the MPG module 26 at 108 to generate one or more local movement shapes at 108, i.e., to determine movements associated with the agent, and to update and maintain stochastic weights between the shapes as discussed above. At 1 10 the MPG module 26 checks the particular shapes with the global shape dictionary, e.g., to determine if the repository 30 needs to be updated at 114 by making a determination at 1 12. The movement profile for the agent is then sent at 1 16 to the comparison module 34. The comparison module 34 receives the movement profile at 1 18 and performs a comparison at 120, which is specific to the particular application (e.g., for health or athletic monitoring etc.). Based on the comparison, updates may be generated (e.g., feedback etc.) and sent at 122, which are received by the MPG module 26 at 124 and by the agent module at 126.

[0062] Therefore, there is provided a system 10 for concisely representing and monitoring changes in physical movement (agent-based and population-based) using latent thematic structures therein. The system 10 builds an efficient representation of movement for each user over time.

[0063] The system 10 can provides contextual, user-customized quantification of changes in characterizing features of movement, such as fluidity.

[0064] The system 10 can encode, maintains and stores a history of movements for a particular or a population of agents, over time.

[0065] The system 10 can encode a concise multi-resolution representation of physical movement at the local layer of individual agents, and global layer of populations of agents.

[0066] The system 10 provides a quantifiable mechanism to categorize physical movement into predefined profiles (e.g., professional, healthy, in recovery, etc.) and can encode defining characteristics to belong to a category of agents (e.g., humans, robots, animals, etc.).

[0067] The system 10 quantifies what it means for a particular subset of the population of agents to move (e.g., human beings of a particular age group, human groups of a particular pathology, or agents following a particular physical regimen), and provides for modules to compare movement profiles of the same agent over time, or different agents.

[0068] The system 10 can also facilitate the transmission of movement profiles for the same or different agents to external sources (e.g., a coach or a physical professional, etc.) and social specialized networks (e.g., a social network for runners, or for back pain patients, etc.).

[0069] The system 10 can also be used to devise a causal model of activities and occurrences of critical episodes (e.g., self-reported pain incidents).

[0070] The system 10 can use multivariate shapes as the atomic unit of physical movement to capture dynamical evolution of physical moves over time, and can maintain a persistent repository of multi-resolution shapes over time, both for a particular agent as well as a population of agents.

[0071] The system 10 can be configured to include an agent module 12 that stores, maintains and initiates queries about the movement of an agent; a shape module 20, which generates or receives multivariate temporal representation of shape-based movements in time; a movement profile generation module 26, that detect sequences of shapes ordered in time that are thematically related (i.e., activities), quantifies fundamental characteristics of movement such as fluidity and range of movement, and can be used to generate categorization profiles (e.g., gold standards); and a comparison module 34, which compares movements of a particular agent over time with the agent's, compares movements of a particular agent with a population or a subset therein, past history, uses categorization profiles based on characterizing features to compare and monitor progress towards milestones, analyzes the latent dynamics of a nonlinear underlying generative process, quantifies how various movement themes connect and change over time, and quantifies variations in mobility and fluidity of movement over time.

[0072] The system 10 can be configured to maintain a repository of universal shapes and can communicate with external repositories for the purposes of movement profile transmission or online social sharing of partial or complete analysis of movement profiles.

References:

[0073] Eiter, T. and H. Mannila (1994). Computing discrete Frechet distance. Rapport technique num. CD-TR. 94: 64.

[0074] Hoffman, M., et al. (2010). Online learning for latent dirichlet allocation.

Advances in Neural Information Processing Systems. [0075] Kondor, R. I. and J. Lafferty (2002). Diffusion kernels on graphs and other discrete input spaces. ICML.

[0076] Kullback, S. and R. A. Leibler (1951 ). On information and sufficiency. The Annals of Mathematical Statistics: 79-86.

[0077] For simplicity and clarity of illustration, where considered appropriate, reference numerals may be repeated among the figures to indicate corresponding or analogous elements. In addition, numerous specific details are set forth in order to provide a thorough understanding of the examples described herein. However, it will be understood by those of ordinary skill in the art that the examples described herein may be practiced without these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the examples described herein. Also, the description is not to be considered as limiting the scope of the examples described herein.

[0078] It will be appreciated that the examples and corresponding diagrams used herein are for illustrative purposes only. Different configurations and terminology can be used without departing from the principles expressed herein. For instance, components and modules can be added, deleted, modified, or arranged with differing connections without departing from these principles.

[0079] It will also be appreciated that any module or component exemplified herein that executes instructions may include or otherwise have access to computer readable media such as storage media, computer storage media, or data storage devices (removable and/or non-removable) such as, for example, magnetic disks, optical disks, or tape. Computer storage media may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data. Examples of computer storage media include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by an application, module, or both. Any such computer storage media may be part of the system 10, any component of or related to the system 10, etc., or accessible or connectable thereto. Any application or module herein described may be implemented using computer readable/executable instructions that may be stored or otherwise held by such computer readable media. [0080] The steps or operations in the flow charts and diagrams described herein are just for example. There may be many variations to these steps or operations without departing from the principles discussed above. For instance, the steps may be performed in a differing order, or steps may be added, deleted, or modified.

[0081] Although the above principles have been described with reference to certain specific examples, various modifications thereof will be apparent to those skilled in the art as outlined in the appended claims.

Claims

Claims:

1 . A method of monitoring movement of an agent, the method comprising:

using spatio-temporal measurements from one or more sensing devices to determine at least one movement associated with the agent;

generating a movement profile based on the at least one movement; and using the movement profile to monitor the movement of the agent.

2. The method of claim 1 , wherein the at least one movement is modeled as one or more shapes, and the movement profile is generated using the one or more shapes.

3. The method of claim 1 or claim 2, wherein the movement of the agent is monitored by comparing the movement profile to an ideal standard.

4. The method of claim 1 or claim 3, wherein the movement of the agent is monitored by comparing the movement profile to a previously determined movement profile.

5. The method of any one of claims 1 to 4, wherein generating the movement profile comprises generating at least one local movement shape, checking a global shape dictionary to determine if the at least one local movement shape is a new shape, and updating a shape repository if so.

6. The method of any one of claims 1 to 5, further comprising outputting information associated with a regimen according to the monitoring.

7. The method of claim 6, wherein the regimen corresponds to any one or more of a fitness regimen, rehabilitation regimen, and a therapy regimen.

8. The method of any one of claims 1 to 7, further comprising providing sensor data to a social repository.

9. The method of any one of claims 1 to 8, wherein topic modeling is applied to sensor data associated with the spatio-temporal measurements such that the sensor data corresponds to words, shapes extracted from the sensor data corresponds to a string of words, the at least one movement corresponds to a document, a movement profile corresponds to a corpus of documents, and characteristics of movements or quality of movements or themes of movements extracted from the movement profile corresponds to a topic or theme of a set of documents.

10. The method of any one of claims 1 to 9, wherein the method is performed by one or more modules included in a wearable device.

1 1 . A computer readable medium comprising computer executable instructions for performing the method of any one of claims 1 to 10.

12. A system for measuring movements of an agent, the system comprising:

an agent module to enable spatio-temporal measurements to be obtained from one or more sensing devices;

a shape module to determine at least one movement associated with the agent; a movement profile generation module generating a movement profile based on the at least one movement; and

a comparison module for using the movement profile to monitor the movement of the agent.

13. The system of claim 12, wherein the at least one movement is modeled as one or more shapes, and the movement profile is generated using the one or more shapes.

14. The system of claim 12 or claim 13, wherein the agent module and the shape module are located at a client device, and the movement profile generation module and the comparison module are located at a server device.

15. The system of claim 12 or claim 13, wherein all of the modules are located at a client device, with ideal standard data being stored at a server device, the server device being communicable with the client device.