基于多特征融合的早期帕金森病步态特征识别方法
2024-04-13李一宽



本文引用格式:李一宽.基于多特征融合的早期帕金森病步态特征识别方法[J].自动化与信息工程,2024,45(1):55-60.
LI Yikuan. Gait feature recognition method for early Parkinson's disease based on multi feature fusion[J]. Automation & Information Engineering, 2024,45(1):55-60.
摘要:帕金森病是一种常见的神经退行性疾病,目前尚无可治愈的手段,只有通过药物治疗来缓解发病周期。因此,针对早期帕金森病的研究具有重要意义。针对早期帕金森病步态数据较少、识别准确率较低的问题,提出一种基于多特征融合的早期帕金森病步态特征识别方法。该方法对比传统的单一特征识别方法,识别准确率更高,可为早期帕金森病临床诊断提供一种辅助手段。
关键词:早期帕金森病;步态数据;特征融合;分类识别
中图分类号:TP391.41 文献标志码:A 文章编号:1674-2605(2024)01-0009-06
DOI:10.3969/j.issn.1674-2605.2024.01.009
Gait Feature Recognition Method for Early Parkinson's Disease Based on Multi Feature Fusion
LI Yikuan
(School of Mechanical and Vehicle Engineering, Chongqing Jiaotong University, Chongqing 400074, China)
Abstract: Parkinson's disease is a common neurodegenerative disease, and there is currently no cure available. Only drug treatment can alleviate the onset cycle. Therefore, research on early Parkinson's disease is of great significance. Aiming at the problem of limited gait data and low recognition accuracy in early Parkinson's disease, a multi feature fusion based gait feature recognition method for early Parkinson's disease is proposed. Compared with traditional single feature recognition methods, this method has higher recognition accuracy and can provide an auxiliary method for early clinical diagnosis of Parkinson's disease.
Keywords: early Parkinson's disease; gait data; feature fusion; classification recognition
0 引言
帕金森病(Parkinson disease, PD)是一種与年龄相关的神经退行性疾病。据统计,目前全球的PD患者有700多万人,我国的PD患者近300万人,约占全球PD患者总数的一半[1]。根据专家预测,2030年我国的PD患者将达到500万人[2]。
目前,临床上针对PD的治疗尚无可治愈的手段。因此,早发现早治疗,延缓病人的发病周期显得尤为重要[3]。PD的诊断主要采用评估量表法(H&Y评估量表、Webster帕金森分级量表、UPDRS评估量表等)和影像学法,这两种方法主观性较强,对医师个人能
力要求较高。根据调查发现,PD的误诊率达10%~ 50%[4]。
为提高PD诊断的高效性和可靠性,许多学者将人工智能技术应用于PD诊断。LITTLE等[5]利用音高周期熵(pitch period entropy, PPE)的声调测量方法,提取10个声调高度不相关的特征向量,并采用核支持向量机进行分类,实验结果表明,该方法能够很好地区分PD患者和健康人群。……
