Fall detection systems and methods
View Patent ↗The present invention relates to systems and methods for fall detection. Specifically, embodiments of the present invention are configured to provide fall detection, including fall prediction and fall prevention, in home health care settings for patients with Alzheimer's disease and related neurodegenerative conditions. In addition, embodiments of the present invention provide continuous fall prediction and detection, including alerts to caregivers, while allowing the patient the flexibility to move outside their home.
1. A method for automatic fall detection, comprising the steps of:
configuring a patient-wearable kinematic sensor to continuously capture, sample, filter, and analyze baseline kinematic fall detection data representative of: patient Activities of Daily Living, fall events, and near-fall events, from said patient wearable kinematic sensor, including data representative of patient acceleration, patient orientation, patient angular momentum, and patient altitude;
training one or more machine learning or artificial intelligence algorithm on said baseline kinematic fall detection data to discriminate between patient Activities of Daily Living, patient fall events, and patient near-fall events, and storing the trained configuration;
initializing one or more machine learning or artificial intelligence algorithm in said patient-wearable kinematic sensor with said trained configuration;
configuring said patient-wearable kinematic sensor to transmit a heartbeat signal detectable by either or both of a medication hub or smart-phone;
configuring said medication hub to receive and detect a heartbeat signal sent by said patient-wearable kinematic sensor;
configuring said smart-phone to receive and detect a heartbeat signal sent by said patient-wearable kinematic sensor;
configuring a fall-detection monitoring time period in said patient-wearable kinematic sensor, said medication hub, and said smart-phone; and, for each fall-detection monitoring time period in a countably infinite set of one or more sequential fall-detection monitoring time period, detecting patient fall events and sending alerts, wherein for each said fall-detection monitoring time period in a countably infinite set of one or more sequential fall-detection monitoring time period, detecting patient fall events and sending alerts includes:
transmitting a heartbeat signal from said patient-wearable kinematic sensor;
performing a test at said medication hub to determine if said heartbeat signal from patient-wearable kinematic sensor is detected by said medication hub;
upon a determination said heartbeat signal from said patient-wearable kinematic sensor is detected by said medication hub, sending from said medication hub a response with information identifying said medication hub and configuring said medication hub to receive one or more event indication;
upon a determination said heartbeat signal from said patient-wearable kinematic sensor is not detected by said medication hub, auditing at said medication hub no heartbeat detected; performing a test at said smart-phone to determine if said heartbeat signal from said patient-wearable kinematic sensor is detected by said smart-phone;
upon a determination said heartbeat signal from said patient-wearable sensor is detected by said smart-phone, sending from said smart-phone a response with information identifying said smart-phone and configuring said smart-phone to receive one or more event indication;
upon a determination said heartbeat signal from said patient-wearable sensor is not detected by said smart-phone, auditing at said smart-phone no heartbeat detected;
configuring said patient-wearable kinematic sensor to continuously, for at least a fall-detection monitoring time period, capture, sample, filter, and analyze kinematic test data, including data representative of patient acceleration, patient orientation, patient angular momentum, and patient altitude;
presenting as input said kinematic test data to said one or more machine learning or artificial intelligence algorithm in said patient-wearable kinematic sensor;
monitoring, in said patient-wearable kinematic sensor, one or more output of said one or more machine learning or artificial intelligence algorithm and performing a test in said patient-wearable kinematic sensor of said one or more output of said one or more machine learning or artificial intelligence algorithm to determine if a fall or near-fall event is detected;
upon a determination that a fall or near-fall event is not detected in said patient-wearable kinematic sensor, auditing at said patient-wearable kinematic sensor no fall event;
upon a determination that a fall or near-fall event is detected at said patient-wearable kinematic sensor, performing a test at said patient-wearable kinematic sensor to determine if said heartbeat response was received at said patient-wearable kinematic sensor from said medication hub;
upon a determination that said heartbeat response was received from said medication hub at said patient-wearable kinematic sensor, sending from said patient-wearable kinematic sensor a fall event to said medication hub;
upon a determination that no said heartbeat response was received from said medication hub at said patient-wearable sensor, auditing at said patient-wearable kinematic sensor no said heartbeat response received from said medication hub;
performing a test at said patient-wearable kinematic sensor to determine if said heartbeat response was received at said patient-wearable kinematic sensor from said smart-phone;
upon a determination that said heartbeat response was received from said smart-phone at said patient-wearable kinematic sensor, sending from said patient-wearable kinematic sensor a fall event to said smart-phone;
upon a determination that no said heartbeat response was received from said smart-phone at said patient-wearable kinematic sensor, auditing at said patient-wearable kinematic sensor no said heartbeat response received from said smart-phone;
performing a test at said medication hub to determine if a fall event was received from said patient-wearable kinematic sensor at said medication hub;
upon a determination a fall event was received from said patient-wearable kinematic sensor at said medication hub, sending from said medication hub an alert to caregivers indicating said medication hub as the source of the alert;
upon a determination a fall event was not received from said patient-wearable kinematic sensor at said medication hub, auditing at said medication hub no fall event;
performing a test at said smart-phone to determine if a fall event was received from said patient-wearable kinematic sensor at said smart-phone;
upon a determination a fall event was received from said patient-wearable kinematic sensor at said smart-phone, sending from said smart-phone an alert to caregivers indicating said smart-phone as the source of the alert, said alert including GPS coordinates from said smart-phone; and
upon a determination a fall event was not received from patient-wearable kinematic sensor apparatus at said smart-phone, auditing at said smart-phone no fall event.