基于MLP神经网络算法分析双任务下老年肌少症人群步态姿势特征  

Analysis of Gait Posture Characteristics of People with Sarcopenia under Dual-Task Based on MLP Neural Network Algorithm

作　　者：王岑依 梁计陵 王国栋[1] 陆阿明[1] WANG Cenyi;LIANG Jiling;WANG Guodong;LU Aming(Soochow University,Suzhou 215021,China;Wuhan Sports University,Wuhan 430079,China.)

机构地区：[1]苏州大学,江苏苏州215021 [2]武汉体育学院,湖北武汉430079

出　　处：《中国体育科技》2024年第5期50-57,共8页China Sport Science and Technology

基　　金：中国博士后科学基金面上项目(2023M742546);国家资助博士后研究人员计划项目(GZC20231899)。

摘　　要：目的:运用3层多层感知机(multilayer perceptron,MLP)神经网络算法,评估双任务范式下老年肌少症步态姿势。方法:选取老年肌少症与健康老年人群受试者各30名。两组受试者被要求分别在单任务、认知-动作双任务、动作-动作双任务3种模式下进行步行动作测试,Vicon红外运动捕捉系统进行数据采集,采用双因素方差法筛选老年肌少症人群步态姿势特征指标,并运用MLP神经网络模型进行权值分析以度量各特征指标。结果:1)相比健康老年组,老年肌少症组在同任务步行下步速、步长、步宽、步态周期、步态变异性指数等25个指标存在显著性差异(P<0.05);2)相比单任务步行,老年肌少症组在双任务下步频、步态周期、支撑期、支撑期百分比变异指数等18个指标存在显著性差异(P<0.05);3)共计14个步态姿势指标存在组间与任务间显著性差异(P<0.05),并纳入老年肌少症步态姿势特征权值分析,显示数据集特征权值的排序前五位分别为步长、步频、右双支撑百分比、步长变异性和左双支撑百分比。结论:老年肌少症人群相比健康老年人在双任务步行中表现出更高的步态变异性,姿势控制能力下降。老年人群健康管理人员可通过关注老年肌少症人群双任务下的步态姿势控制特征表现,优化肌少症人群姿势控制调整策略,降低老年人群跌倒风险。Objective:Multilayer Perceptron(MLP)neural network algorithm was used to evaluate the gait posture of sarcopenia under the dual-task paradigm.Methods:A total of thirty subjects were selected from the sarcopenia group and the healthy older group.The two groups of subjects were asked to complete the gait test in three modes:single-task(ST),cognitive-motor dual-task(CMDT)and motor-motor dual-task(MMDT),respectively,and the Vicon infrared capture system collected the data.Two-way ANOVA was used to screen the gait and posture characteristics of people with sarcopenia,and the weight analysis was performed by MLP neural network model to measure each characteristic index.Results:1)Compared with the normal aged group,there were significant differences in 25 indexes such as stride speed,stride length,stride width,gait cycle,and gait variability index in the sarcopenia group under the same task(P<0.05).2)Compared with ST walking,18 indices such as step frequency,gait cycle,support period,and percentage variation index of support period were significantly different in the sarcopenia group under the dual task(P<0.05).3)A total of 14 gait posture indices had significant differences between groups and between single and dual tasks(P<0.05),and the weight analysis of gait characteristics in sarcopenia was included.The results showed that the top five feature weights in the dataset were stride length,stride frequency,right double support percentage,stride length variability,and left double support percentage.Conclusions:Individuals with sarcopenia exhibited higher gait variability and decreased postural control during dual-task walking compared to healthy older adults.Health managers of the elderly population can optimize the postural control adjustment strategy of the sarcopenia population and reduce the fall risk of the older population by paying attention to the gait and postural control characteristics of people with sarcopenia during dual-task walking.

关 键 词：老年肌少症 神经网络 双任务 步态 

分 类 号：G804.2[文化科学—运动人体科学]