科技创新与应用2024，Vol.14Issue(13)：60-63,4.DOI:10.19981/j.CN23-1581/G3.2024.13.015
失智老人徘徊行为识别算法
摘要
Abstract
In order to detect the wandering behavior of the elderly people with dementia and realize intelligent nursing,a two-stage multi-granularity indoor wandering behavior pattern detection algorithm framework based on UWB(Ultra Wide Band)motion trajectory is proposed.In the first stage of the algorithm framework,the motion trajectory of the elderly is screened to determine whether it is wandering behavior,and in the second stage,fine-grained classification is carried out to determine which specific wandering mode(ring,round-trip,ring-round,and ring-round-round).Comparing the two stages,in the first stage,comparing 1D-CNN,LSTM and LR algorithms,1D-CNN is the best,with an accuracy of 96.1%and a recall rate of 94.3%.In the second stage,compared with LR,XGBoost,LightGBM,and 2D-CNN algorithms,XGBoost has the best effect,with a Macro-Precision of 97.24%and a Macro-Recall rate of 97.18%.The experimental results show that the proposed wandering detection algorithm framework can effectively identify the wandering behavior of the elderly people with dementia.关键词
徘徊行为/UWB轨迹/CNN/XGBoost/失智老人Key words
wandering behavior/UWB trajectory/CNN/XGBoost/elderly people with dementia分类
信息技术与安全科学引用本文复制引用
吕非非,鲁卫华,崔震..失智老人徘徊行为识别算法[J].科技创新与应用,2024,14(13):60-63,4.基金项目
2017年度北京市重点研发项目(D17110600820000) （D17110600820000）
