Scholarlink

<p>In-person motor and cognitive assessments for Parkinson’s disease(PD) face accessibility, scalability, and geographical diversity challenges. We aimed to address these by integrating large datasets, machine learning (ML), and videoconferencing. We developed the Motor and Cognitive Videoconference(MaC-VC) protocol, allowing non-experts to remotely administer the MDS-UPDRS III and MoCA tests. In this cross-sectional study, we administered MaC-VC to 145 participants from 60+ geographical locations and compared the results with a large (<i>n</i> = 1264), expert-rated, in-person assessments from the PPMI dataset. The abridged, online-feasible MDS-UPDRS III accounted for 95% of the variance in complete MDS-UPDRS III scores. When comparing early-versus-advanced PD in each dataset (In-Person/Online), we observed consistent significant trends in four measures: MDS-UPDRS-III, MoCA, disease-duration, and sex. Using a bidirectional cross-dataset-validation technique, ML classifiers yielded high classification performance both within-dataset and between-datasets(AUCs&gt;0.9), demonstrating predictive power across diverse populations. These findings support the feasibility and generalizability of MaC-VC, paving the path for accessibility, scalability, and geographical diversity in PD assessments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Accessible assessment of motor and cognitive symptoms in Parkinson’s disease: integrating large datasets, machine-learning, and videoconferencing

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月7日

查看全文
  • Avigail Lithwick Algon,
  • William Saban

摘要

In-person motor and cognitive assessments for Parkinson’s disease(PD) face accessibility, scalability, and geographical diversity challenges. We aimed to address these by integrating large datasets, machine learning (ML), and videoconferencing. We developed the Motor and Cognitive Videoconference(MaC-VC) protocol, allowing non-experts to remotely administer the MDS-UPDRS III and MoCA tests. In this cross-sectional study, we administered MaC-VC to 145 participants from 60+ geographical locations and compared the results with a large (n = 1264), expert-rated, in-person assessments from the PPMI dataset. The abridged, online-feasible MDS-UPDRS III accounted for 95% of the variance in complete MDS-UPDRS III scores. When comparing early-versus-advanced PD in each dataset (In-Person/Online), we observed consistent significant trends in four measures: MDS-UPDRS-III, MoCA, disease-duration, and sex. Using a bidirectional cross-dataset-validation technique, ML classifiers yielded high classification performance both within-dataset and between-datasets(AUCs>0.9), demonstrating predictive power across diverse populations. These findings support the feasibility and generalizability of MaC-VC, paving the path for accessibility, scalability, and geographical diversity in PD assessments.