基于人工神经网络模型的老年人体质综合评价与探索  

Comprehensive Evaluation and Exploration of Physical Fitness in Elderly Based on Artificial Neural Network Model

作　　者：张频 陈铭 蔡天盼 龙静文 吴磊[1] ZHANG Pin;CHEN Ming;CAI Tian-pan;LONG Jing-wen;WU Lei(School of Public Health of Nanchang University,Jiangxi Province Key Laboratory of Preventive Medicine,Nanchang 330006,China)

机构地区：[1]南昌大学公共卫生学院、江西省预防医学重点实验室,南昌330006

出　　处：《南昌大学学报（医学版）》2020年第4期89-93,共5页Journal of Nanchang University：Medical Sciences

基　　金：国家自然科学基金(81560550,81360446);国家级创新创业训练项目(201801012)。

摘　　要：目的探讨基于人工神经网络(artificial neural network,ANN)模型用于老年人体质综合评价的价值。方法采用整群抽样,抽取南昌市部分社区老年人进行体质调查。构建基于ANN的老年人体质评价模型,并对他们的体质进行综合评价。结果最终完成所有测试指标的有效体质量表计1739份,其中男性718人,女性1021人。男性ANN模型体质权重前3位为:握力(0.160)、肺活量(0.143)、坐位体前屈(0.106);女性前3位为:肺活量(0.117)、选择反应时(0.105)、坐位体前屈(0.092)。构建ANN模型评价男性和女性体质的ROC曲线面积分别为0.935、0.907。结论初步构建的老年人体质神经网络模型准确率较高,可应用于老年人体质的个性化评价。Objective To explore the value of artificial neural network(ANN)model in comprehensive evaluation of physical fitness in elderly people.Methods Cluster sampling was used to investigate the physical fitness of the elderly in some communities of Nanchang city.The elderly’s physical fitness evaluation model based on ANN was constructed,and their physical fitness was evaluated comprehensively.Results A total of 1739 people completed the effective physical fitness scale,including 718 males and 1021 females.The top three of constitution weight of ANN model were grip strength(0.160),vital capacity(0.143)and sitting-posture body anteflexion(0.106)in males,and vital capacity(0.117),reaction time(0.105)and sitting-posture body anteflexion(0.092)in females.The ROC curve area of ANN model for evaluating constitution was 0.935 in males and 0.907 in females.Conclusion The ANN model is accurate for personalized evaluation of physical fitness in elderly people.

关 键 词：体质评价 人工神经网络模型 老年人 南昌 江西 

分 类 号：R161.7[医药卫生—公共卫生与预防医学]