基于心率变异性与机器学习的养老护理人员疲劳分类
打开文本图片集
中图分类号：TP391.4 文献标志码：A 文章编号：1671-024X（2025）04-0044-08
1.00 210 4 5
Fatigue classification of service staff in elderly care based on heart rate variability and machine learning
ZHANG Xin1，MA Shuai²，OU Zongkun²，PENG Cheng³，WEI Ran4 （1.Collge ofElderly Welfare，China Civil Afairs University，Beijing 102600，China；2.ScholofElectronicand InformationEnginering，Tiangong University，Tianjin3O387，China；3.SchoolofLifeSciences，Tiangong University，Tianjin 300387，China；4.CollgeofRehabilitation Technology，China Civil Afairs University，Beijing 1026O，China）
Abstract：Aimingatthe limitations ofthesubjective scalemethod inthe fatigueasessmentofelderlycaregivers，this study proposes afatigue classification method for elderlycaregivers basedon heart rate variability（HRV）and machine leaming. We collected 736 h of ECG data from 78 caregivers through wearable devices，used Butterworth filtering and Pan-Tompkins algorithm for preprocessing，systematically extracted a total of 42 feature parameters in the timedomain，frequency domain，and nonlinear domainof HRV，and used Pearson corrlationcoefficient to filter18 key features to construct afatigue classification model based on XGBoost.The experimental results show thatthe model performance is significantlyimproved byfeature dimensionalityreduction，and theclassification accuracy is increased from O.78 to O.91，among which the accuracy of heavy fatigue classification reaches 0.99; compared with the traditional models such as SVM，KNN，etc.，XGBoost demonstrates the optimal nonlinear fittingability after feature streamlining.Further studies showed that the classification performance of2 min shorttime window ECG data was optimal with an accuracy of O.91，which was significantly beter than the 4-10 min ECG dataaccuracy of O.61-O.69，respectively.This method achieves eficient andaccurate monitoring of nursing fatigue，and provides wearable smart technology support for the prevention of occupational cardiovascular and cerebrovascular diseases and the optimization of elderly service management.
Key words： heartrate variability（HRV）；machine learning;electrocardiogram（ECG）； feature dimension reduction； XG-Boost；fatigue classification
随着我国老龄化进程的加速，养老需求日益增长，使得养老机构护理人员面临着日益繁重的工作压力，其身心健康状况不容乐观。（剩余13385字）
