Detecting falls with multiple wearable motion sensors
Inventors
Filatov, Denis • Zaveri, Hitten
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11937917-B1
Publication Date
2024-03-26
Expiration Date
Abstract
In an example system, multi-sensor motion data that reflects the motion of a user over a period of time is received as recorded by a plurality of motion sensors on wearable devices, specific changes are determined in the data by, firstly, quantifying it using a two-dimensional data transform; secondly, extracting an anomalous area; and thirdly, calculating a number of the anomaly's properties, and the results are input into a machine learning model to detect that the user fell. The machine learning model processes the anomaly's properties and evaluates the current state of the user's activity by classifying the properties against a state space previously calculated by analyzing historical activities of daily living. Depending on a two-dimensional transform implemented, the machine learning model detects when the user falls, as well as potentially allows predicting that the user will suffer a fall in advance of the actual event, aiming at solving the stroke prediction problem.
Core Innovation
A wearable device worn by a user includes a motion sensor with a first motion sensor worn on a first wrist and a second motion sensor worn on a second wrist. The motion sensor collects acceleration data indicating a first wrist acceleration and a second wrist acceleration, where the acceleration data relates to motion data based on the user’s activities over a period of time, and the wearable device communicates with a server.
One or more properties of the motion data are calculated by calculating a two-dimensional data transform of the motion data. Using the two-dimensional data transform, an area is determined where the transform’s values exceed a threshold, and one or more properties of the area are calculated.
The user fell is determined using a machine learning model executed on a processor of a cloud service based on the one or more properties of the motion data. The machine learning model was previously trained on historical motion data, and the determination is transmitted to the server, a caretaker of the user, and an emergency system.
Claims Coverage
The independent claims are clm-00001, clm-00008, and clm-00015, each covering determining whether a user fell using wrist-worn wearable motion sensors, a two-dimensional transform with threshold-exceeded area properties, and a cloud-executed machine learning model trained on historical motion data, followed by transmitting the determination to specified recipients. Across these independent claims, three main inventive aspects are expressed.
Wrist-worn dual motion sensors providing acceleration-based motion data to a server
A wearable device worn by a user includes a motion sensor with a first motion sensor worn on a first wrist and a second motion sensor worn on a second wrist. The motion sensor collects acceleration data including data indicating a first wrist acceleration and a second wrist acceleration, and the wearable device communicates with a server.
Two-dimensional transform with thresholded transform-exceeding area and area property calculation
One or more properties of the motion data are calculated by calculating a two-dimensional data transform of the motion data, determining an area where the transform’s values exceed a threshold, and calculating one or more properties of the area.
Cloud-executed machine learning fall determination based on historical training and transmitting the determination
A machine learning model executed on a processor of a cloud service determines that the user fell based on the one or more properties of the motion data, the machine learning model was previously trained on historical motion data, and the determination is transmitted to the server, a caretaker of the user, and an emergency system.
The independent claims consistently center on wrist-worn motion sensing, transform-based calculation of area properties, and cloud-executed machine-learning fall determination with transmission of the result to specified recipients.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?