中国医学装备2026，Vol.23Issue(1)：11-17,33,8.DOI:10.3969/j.issn.1672-8270.2026.01.002
基于步态特征与机器学习的单侧膝髋关节疾病识别模型构建
Construction of identification model based on gait characteristics and machine learning for disease of unilateral knee and hip joints
叶冬梅 1吴文达 2崔均健 2李岱熹 2宋泱宏 1袁儒非1
作者信息
- 1. 大连大学附属中山医院康复医学科 大连 116001
- 2. 大连恒锐科技股份有限公司研发部 大连 116085
- 折叠
摘要
Abstract
Objective:To construct a best identification model for diseases of unilateral knee and hip joints,which facing to two kinds of scenarios including clinical screening and diagnosis for disease,so as to effectively identify the diseases of unilateral knee joint and unilateral hip joints.Methods:An identification model for disease of unilateral knee and hip joints was constructed through gait characteristics combined with machine learning.Participants were recruited from Affiliated Zhongshan Hospital of Dalian University between September 2024 and March 2025,which included 40 healthy persons,41 patients with diseases at unilateral knee joint,and 24 patients with diseases at unilateral hip joint.Everspry Gait Health Assessment System was adopted to measure 20 features of gait parameters including ratio of the load on left side of plantar,ratio of the load on right side of plantar,ratio of single support period of left side,and ratio of single support period of right side.At the same time,the Decision Tree,Multilayer Perceptron(MLP),Random Forest,Adaptive Boosting Algorithm(AdaBoost),and Support Vector Machine(SVM)were adopted to establish identification model for diseases of unilateral knee and hip joints on the basis of the above features.The gait features of healthy persons and patients with disease at unilateral knee or hip joint were trained,and the effects of different models in identifying diseases of unilateral knee or hip joint were compared.Results:The optimal model of identifying heathy persons and patients with disease at unilateral knee or hip joints was the AdaBoost algorithm,which identification accuracy was 80%.The optimal model of identifying healthy persons,patients with disease at unilateral knee joints,and patients with disease at unilateral hip joints was SVM model,which identification accuracy was 67%.Conclusion:The identification model based on gait characteristics and machine learning for disease of unilateral knee and hip joints can effectively identify the disease of unilateral knee joints,and disease of unilateral hip joints,which can provide conveniences for clinical screening and diagnosis for diseases of unilateral knee and hip joints.关键词
步态特征/机器学习/膝关节/髋关节Key words
Gait analysis/Machine learning/Knee joint/Hip joint分类
医药卫生引用本文复制引用
叶冬梅,吴文达,崔均健,李岱熹,宋泱宏,袁儒非..基于步态特征与机器学习的单侧膝髋关节疾病识别模型构建[J].中国医学装备,2026,23(1):11-17,33,8.
