A Highly Reliable Ultrasonic Sensor-Based Bed-Exit Detection System for the Elderly With Early Notification

This letter presents a real-time bed-exit alarm system using ultrasonic sensors, infrared (IR) sensors, and an accelerometer. The ultrasonic sensors detect the elderly in bed; the IR sensors detect the motion of the elderly across the gap at the end of the bed; and the accelerometer detects the bed rail being lowered. To achieve early notification, one-against-one support vector machine classifiers are implemented to recognize the early stages of bed-exit events with a sensitivity of 97.7% and a precision of 93.8%. The proposed system is verified with ten elderly subjects and provides an average lead time of 11.4 s for real bed-exit events, indicating that the proposed system has the ability to issue early warnings to caregivers for immediate assistance.