Biosignal integration with vehicles and movement

Inventors

Low, Philip

Assignees

Neurovigil Inc

Abstract

The present disclosure relates to methods and system for acquiring and analyzing biosignals or physiological signals of a person sitting in a vehicle and predicting (in real-time) time-varying attention, engagement level or alertness level using the biosignals. The biosignals may be acquired using one or more clusters of electrodes together with a wearable user device or from a sensing device that is embedded in the seat of the vehicle. In some embodiments, the biosignals may be utilized to predict restedness level of the subject, to monitor or predict physiological state of the subject, to detect a distress situation and to adapt a vehicle control accordingly. In some other embodiments, the biosignals can be transformed into communication, for example, speech signals or instructions for the vehicle. One or more actions may be triggered based on the analysis of the biosignals including engaging the person, generating alerts, or adapting the vehicle control.

Core Innovation

The invention provides a computer-program product that accesses one or more physiological signals of a subject sitting in a vehicle, where the signals are collected by a physiological data acquisition assembly comprising a sensing device and one or more clusters of electrodes with at least an active electrode. For a given time interval corresponding to a scanning window that segments the physiological signals, the system computes a set of features derived from one or more frequency bands of the physiological signals, including features derived using component analysis from a spectrogram or a normalized spectrogram of the frequency bands.

The system predicts one or more health metrics by using the set of features with one or more machine learning models, where the health metrics correspond to measures indicating a health status of the subject. It determines an extent to which the predicted health metrics are deviated from a corresponding baseline value and, based on the determination, triggers one or more actions that include outputting the one or more health metrics, generating an alert signal to the subject or a caregiver, or adapting a vehicle control.

The invention further predicts a state for each of a plurality of time intervals based on other physiological data collected over a period of time, where the state corresponds to one or more sleep stages or an awake state. Based on predicting the state, the system determines a sleep pattern, a relative frequency of a particular state, or a duration of a particular state, and predicts a restedness level using a restedness prediction model.

In addition, the invention supports real-time control based on biometric patterns by accessing other physiological signals of the subject or another subject sitting in a mobility-aid device, a pod, or a vehicle. It computes another set of features from the other physiological signals for a given time interval, determines a biometric pattern, and generates one or more instructions to control behavior of the pod, the mobility-aid device, or the vehicle by controlling one or more motors or actuators, where the instructions include a driving instruction, a movement instruction, or a latching instruction.

Claims Coverage

The partial content includes four independent claims. Across them, the main inventive features are organized around accessing vehicle-situated physiological signals with an electrode-cluster sensing assembly, time-windowed frequency-band feature computation including spectrogram-based component analysis features, machine-learning prediction of health metrics and baseline deviation determination, and triggering actions including alerts and vehicle control adaptations; additional independent claim coverage extends to sleep-state and restedness prediction and to real-time biometric-pattern-based instruction generation for pod, mobility-aid, and vehicle control.

Across the independent claims, the core coverage centers on computing scanning-window frequency-band-derived features, including spectrogram-based component-analysis features, predicting health metrics via machine learning, comparing predicted metrics to baseline, and triggering outputs, alerts, and vehicle control adaptation; additional coverage adds sleep-stage and restedness prediction for vehicle movement control and real-time biometric-pattern-based instruction generation for controlling a pod, mobility-aid device, and/or vehicle via motors or actuators.

Stated Advantages

Documented Applications