Deriving insights into motion of an object through computer vision
Inventors
Kruszewski, Paul Anthony • Zhang, Wenxin • Lacroix, Robert • Russell, Ryan
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Assignees
MemberHinge HealthHinge HealthHinge Health delivers a musculoskeletal care platform that combines digital physical therapy, individualized exercise programs, in-person clinical services, and multidisciplinary care for conditions such as pain, mobility loss, pelvic health, gastrointestinal disorders, and menopause-related symptoms. The platform utilizes remote monitoring, motion analysis, and evidence-based care, prioritizing validated clinical outcomes, data security, and integration with benefits systems for employers and health plans.
Hinge Health delivers a musculoskeletal care platform that combines digital physical therapy, individualized exercise programs, in-person clinical services, and multidisciplinary care for conditions such as pain, mobility loss, pelvic health, gastrointestinal disorders, and menopause-related symptoms. The platform utilizes remote monitoring, motion analysis, and evidence-based care, prioritizing validated clinical outcomes, data security, and integration with benefits systems for employers and health plans.
Publication Number
US-12329515-B2
Publication Date
2025-06-17
Expiration Date
Abstract
Introduced here are computer programs that are able to generate computer vision data through local analysis of image data (also referred to as “raw data” or “input data”). The image data may be representative of one or more digital images that are generated by an image sensor. Also introduced here are apparatuses for generating and handling the image data and computer vision data.
Core Innovation
The invention relates to computer-vision architectures that generate computer vision data from digital images of an environment in which an individual is situated over an interval of time. A camera generates the digital images, and a processor applies analysis of the digital images with a neural network to produce outputs for each digital image, where the neural network outputs a pose of the individual so that serialized poses are established over the interval of time.
The generated computer vision data is encoded into a portable data structure for communication to a second apparatus. The second apparatus decodes the portable data structure for analysis of the computer vision data, and determines a health status of the individual. In related implementations, the system uses series of outputs representing spatial position, populates them into a data structure, transmits the data structure to a first computing device, and assesses health of the individual in real time.
The disclosed dataset is representative of information related to the individual learned through analysis of the digital images, including musculoskeletal state. Representations described include serialized poses and additional pose- and anatomy-related information, such as two-dimensional locations of one or more joints, three-dimensional rotation of one or more joints, location and size and/or shape of one or more muscles, thermal maps representing the surface of an individual’s body, and voxel-based volumetric representations.
Claims Coverage
The independent claims are clm-00001, clm-00010, and clm-00025, and collectively define four inventive features: generating computer vision data from camera images using a neural network or model, encoding and transmitting portable data structures for analysis with a graphics processing unit, assessing health in real time for an individual, and multi-apparatus capture with decoder-based aggregation. Across the claims, the output representations include serialized poses/spatial position and musculoskeletal/thermal/volumetric information.
Neural-network serialized pose generation encoded as portable computer vision data
The apparatus includes a camera configured to generate digital images over an interval of time, and a processor configured to generate computer vision data by analyzing the digital images with a neural network, where the neural network outputs, for each digital image, a pose of the individual so as to establish serialized poses over the interval of time, and encodes the computer vision data into a portable data structure.
GPU-based decoding for health status determination
The apparatus includes a communication module configured to communicate the digital images and the portable data structure to a second apparatus with a graphics processing unit, where the digital images are stored and the portable data structure is decoded for analysis of the computer vision data and determination of a health status of the individual.
CPU-GPU distributed health assessment with real-time action
The method includes acquiring, by a second computing device, a series of digital images generated by a camera in rapid succession, applying a model to produce a series of outputs representative of information regarding a spatial position of the individual, populating the series of outputs into a data structure transmitted to a first computing device, assessing health of the individual in real time based on the computer vision data in the data structure, and performing an action based on the health.
Multi-apparatus dataset generation with decoder-based aggregation
A system includes a plurality of imaging apparatuses, each having an image sensor producing digital images over an interval of time and a central processing unit configured to generate a dataset representative of information learned through analysis of the digital images and populate it into a data structure exiting the imaging apparatus; a processing apparatus includes a decoder and communications interface to receive the data structures, and a graphics processing unit provides the data structures to the decoder to obtain a plurality of datasets corresponding to different imaging apparatuses, and assesses health of the individual by examining the plurality of datasets.
Taken together, the independent claims focus on distributed computer-vision processing that converts image series into encoded portable data structures or datasets containing serialized pose/spatial-position information, transmits them for decoding on a graphics processing unit, and then determines or assesses health of an individual in real time, optionally using multiple imaging apparatuses and decoder-based aggregation.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?