Design and implementation of a fall detection system for elderly patients
10.3969/j.issn.1672-8270.2024.02.030
- VernacularTitle:老年患者跌倒检测系统的设计与实现
- Author:
Min ZHANG
1
;
Huan ZHANG
;
Xiaojuan SHI
;
Zhuowen LIANG
;
Na ZHANG
Author Information
1. 空军军医大学西京医院骨科 西安 710032
- Keywords:
Machine vision;
Medical care;
Fall detection;
YOLOv5 algorithm
- From:
China Medical Equipment
2024;21(2):157-161
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To design a fall detection system for elderly patients to solve the problem of elderly patients failing to detect accidental falls in time and to improve the efficiency of medical care.Methods:Based on real-time stream transmission protocol(RTSP),combined with YOLOv5 and Kalman algorithms,a fall detection system for elderly patients was designed by using Vue and Flask technologies.A visual background system management was established,and a unified management platform was provided for medical staff through comprehensive processing of multiple video streams to realize the autonomous detection and alarm of human fall behavior.30 healthy volunteers who underwent fall testing at Xijing Hospital of Air Force Medical University in 2020 to 2022 were selected and divided into normal walking group,squatting group and falling group according to the simulated behavioural categories,with 10 in each group.The fall detection performance was evaluated using two evaluation indicators:detection accuracy and detection speed to verify and determine whether the fall detection system for elderly patients can meet the requirements of timely and accurate fall detection and alarm.Results:The overall fall detection rate of the normal walking group,the squatting group and the falling group can reach 29 frames per second,and the accuracy rate can reach 95.24%.and the system can respond to the fall alarm in time.Conclusion:The fall detection system for elderly patients can assist medical staff to promptly detect and deal with the occurrence of falls,improve the efficiency of fall detection for elderly patients,and meet the real-time detection and alarm of fall behavior for elderly patients.
