Method and system utilizing pattern recognition for detecting atypical movements during physical activity
Methods, systems and devices are provided for utilizing user movement data obtained from one or more wearable sensors during physical activity to compare individualized changes overtime, for example typical versus atypical movement patterns, with subgroup analyses for assessing changes between other users in order to develop an assessment of movement for, for example, tracking injury risk, performance, and/or rehabilitation. The movement information may comprise multi-sensor, high dimensional datasets. Techniques are provided for integrating human movement data from one or more wearable sensor with one or more additional data sources to define an individualized movement profile of a user's movements. The user or another individual may be notified when the user's movements deviate from this individualized movement profile.
1. A device comprising:
at least one computer processor;
computer memory in communication with the at least one computer processor and for storing computer executable instructions, which when executed by the at least one computer processor cause the at least one computer processor to perform operations comprising:
receive individualized movement profile information for a user, wherein the individualized movement profile information defines a subspace in which movements of the user are considered typical for the user, and wherein the individualized movement profile information corresponds to external condition information for providing context to the individualized movement profile of the user;
receive new movement information of the user relating to a physical activity from one or more wearable motion sensors;
receive new external condition information associated with the new movement information;
determine if the movements of the user during the physical activity are typical for the user by determining if the movements according to the new movement information are located within the subspace;
generate, in response to determining that the movements of the user during the physical activity are typical for the user, a first output indication signal;
identify, in response to determining that the movements of the user during the physical activity are not typical for the user, a subgroup from among a plurality of subgroups that most closely corresponds to the new movement information, wherein each subgroup consist of movement information of users all sharing on one or more predetermined characteristics, wherein each subgroup defines a subspace associated with the one or more predetermined characteristics of the users of the subgroup such that the movements of the users of the subgroup according to the movement information are located within the subspace of the subgroup, wherein the subspace of the subgroup comprises a multivariate threshold boundary such that the movements of the users of the subgroup according to the movement information are located within the multivariate threshold boundary of the subgroup, and wherein at least of the determining if the movements of the user during the physical activity are typical for the user, and the identifying a subgroup from among the plurality of subgroups that most closely corresponds to the new movement information is based on the received new external condition information;
determine if the identified subgroup is the same or different as a predefined subgroup associated with the user;
generate, in response to determining that the identified subgroup is the same as the predefined subgroup, a second output indication signal; and
generate, in response to determining that the identified subgroup is different than the predefined subgroup, a third output indication signal.
2. The device of claim 1 , wherein the subspace of the individualized movement profile information is a multivariate subspace, wherein the individualized movement profile information comprises a multivariate threshold boundary defining an area or region of the multivariate subspace, and wherein the determining if the movements according to the new movement information are located within the subspace involves determining if the user movements are located within the multivariate threshold boundary.
3. The device of claim 1 , wherein the identifying a subgroup from among the plurality of subgroups involves comparing the new movement information to movement information of users of at least some of the plurality of subgroups using a supervised machine learning process.
4. The device of claim 1 , wherein the movement information of users of a subgroup comprises a plurality of values where each of at least some of the values is an average of movement segments of a particular user of the subgroup.
5. The device of claim 1 , further configured to generate or modify the individualized movement profile information for the user by receiving movement data representing a plurality of individual movement segments of the user from one or more wearable motion sensors, and using an unsupervised machine learning process on the received movement data to define an expected multivariate range of movement values for the user.
6. The device according to claim 5 , further configured such that the unsupervised machine learning process generates a model based on the received movement data describing values and multivariate relationships to be considered typical for the received movement data, and the unsupervised machine learning process then defines the subspace in which movements of the user are considered typical based on the generated model.
7. The device of claim 1 , further configured to amalgamate the new movement information into the individualized movement profile information of the user in response to determining that the movements of the user during the physical activity are typical for the user, and saving the amalgamated individualized movement profile information in the memory.
8. The device of claim 1 , wherein the external condition information comprises information relating to at least one of terrain, route, weather, season, or time of day.
9. A method comprising:
receiving, by at least one computer processor, individualized movement profile information for a user from a database, wherein the individualized movement profile information defines a subspace in which movements of the user are considered typical for the user, and wherein the individualized movement profile information corresponds to external condition information for providing context to the individualized movement profile of the user;
receiving, by at least one computer processor, new movement information of the user relating to a physical activity from one or more wearable motion sensors;
receiving, by at least one computer processor, new external condition information associated with the new movement information;
determining, by at least one computer processor, if the movements of the user during the physical activity are typical for the user by determining if the movements are located within the subspace;
generating, by at least one computer processor, in response to determining that the movements of the user during the physical activity are typical for the user, a first output indication signal;
identifying, by at least one computer processor, in response to determining that the movements of the user during the physical activity are not typical for the user, a subgroup from among a plurality of subgroups that most closely corresponds to the new movement information, wherein each subgroup consist of movement information of users all sharing on one or more predetermined characteristics, wherein each subgroup defines a subspace associated with the one or more predetermined characteristics of the users of the subgroup such that such that the movements of the users of the subgroup according to the movement information are located within the subspace of the subgroup, wherein the subspace of the subgroup comprises a multivariate threshold boundary such that the movements of the users of the subgroup according to the movement information are located within the multivariate threshold boundary of the subgroup, and wherein at least one of the determining if the movements of the user during the physical activity are typical for the user, and the identifying a subgroup from among the plurality of subgroups that most closely corresponds to the new movement information is based on the received new external condition information;
determining, by at least one computer processor, if the identified subgroup is the same or different as a predefined subgroup associated with the user;
generating, by at least one computer processor, in response to determining that the identified subgroup is the same as the predefined subgroup, a second output indication signal;
generating, by at least one computer processor, in response to determining that the identified subgroup is different than the predefined subgroup, a third output indication signal.
10. A non-transitory computer-readable storage medium storing instructions that when executed by at least one computer cause the computer to perform operations comprising operations according to claim 9 .
11. The method according to claim 9 , wherein the subspace of the individualized movement profile information is a multivariate subspace, wherein the individualized movement profile information comprises a multivariate threshold boundary defining an area or region of the multivariate subspace, and wherein the determining if the movements according to the new movement information are located within the subspace involves determining if the user movements are located within the multivariate threshold boundary.
12. The method according to claim 9 , wherein the identifying a subgroup from among the plurality of subgroups involves comparing the new movement information to movement information of users of at least some of the plurality of subgroups using a supervised machine learning process.
13. The method according to claim 9 , wherein the movement information of users of a subgroup comprises a plurality of values where each of at least some of the values is an average of movement segments of a particular user of the subgroup.
14. The method according to claim 9 , further comprising generating or modifying the individualized movement profile information for the user by receiving movement data representing a plurality of individual movement segments of the user from one or more wearable motion sensors, and using an unsupervised machine learning process on the received movement data to define an expected multivariate range of movement values for the user.
15. The method according to claim 14 , wherein the unsupervised machine learning process generates a model based on the received movement data describing values and multivariate relationships to be considered typical for the received movement data, and the unsupervised machine learning process then defines the subspace in which movements of the user are considered typical based on the generated model.
16. The method according to claim 9 , further comprising amalgamating the new movement information into the individualized movement profile information of the user in response to determining that the movements of the user during the physical activity are typical for the user, and saving the amalgamated individualized movement profile information in a computer memory.
17. The method according to claim 9 , wherein the external condition information comprises information relating to at least one of terrain, route, weather, season, or time of day.