Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12575760-B2
Publication Date
2026-03-17
Expiration Date
Abstract
Methods and electronic devices for measuring motion and acoustic signatures of physiological processes of a human subject. The method includes measuring motion and acoustic signatures of physiological processes of a human subject; sending the first feature-related data to a machine learning service; sending the first feature-related data to a machine learning service; and determining a predicted detection of human scratching activity by the machine-learning service by performing a machine-learning operation on the feature-related data.
Core Innovation
The invention is an electronic device and method for measuring motion and acoustic signatures of physiological processes of a human subject using an inertial measurement unit (IMU) mounted on an electronics board in direct mechanical communication with the skin. The device includes a microcontroller unit (MCU) configured to calculate feature-related data from the motion and acoustic signatures and to send the feature-related data to a machine learning service executing a trained and programmed machine language model.
The machine language model performs a machine-learning operation on the feature-related data to generate a predicted detection of human scratching activity. The predicted detection is performed by a first machine-learning operation involving at least one neural network flow within the machine language model, such that the neural network flow processes the feature-related data to output the predicted detection.
The described approach further supports refinement of the predicted detection, including detection of scratching start and end with associated times, calculation of scratch duration, and a conditional prediction of whether the human subject is asleep to gate a second machine-learning operation. The closed-loop device can also provide haptic feedback using a vibratory motor in response to predicted human scratching activity.
Claims Coverage
The document includes two independent claims: one directed to an electronic device and one directed to a method. Across these independent claims, the core inventive content centers on IMU-based motion/acoustic signature measurement, MCU feature-related data calculation, and a trained machine language model performing a neural-network-based machine-learning operation to predict human scratching activity.
IMU-and-MCU device for motion and acoustic signatures to predicted scratching detection
An electronic device includes top and bottom portions and an electronics board with an IMU operably in direct mechanical communication with the skin and a microcontroller unit (MCU) communicatively coupled to the IMU, where the MCU calculates feature-related data from the motion and acoustic signatures and sends the feature-related data to a machine learning service executing on the MCU.
Trained machine language model with neural network flow for predicted scratching activity
The machine learning service includes a trained and programmed machine language model to receive the feature-related data and generate a predicted detection of scratching activity by performing a machine-learning operation including at least one neural network flow, where the predicted detection is performed by a first machine-learning operation using the trained and programmed machine language model.
IMU-only motion and acoustic signatures measuring for machine-learning predicted scratching
A method provides a sensor including an IMU and an MCU co-mounted on an electronics board and including a machine learning service, measures motion and acoustic signatures of physiological processes of a human subject solely through the IMU, calculates a first feature-related data from the motion and acoustic signatures, and sends the first feature-related data to the machine learning service.
Neural-network machine-learning operation on features to determine predicted scratching activity
The method determines a predicted detection of human scratching activity by performing a machine-learning operation using a machine language model of the machine learning service on the feature-related data, where the machine-learning operation involves at least one neural network.
Together, the independent claims require an IMU-based wearable sensor architecture with an MCU calculating feature-related data from motion and acoustic signatures, and a trained machine language model performing a neural-network-based machine-learning operation to generate a predicted detection of human scratching activity from the feature-related data.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?