Scholarlink

<p>This research aimed to establish a model predicting mild cognitive impairment in Parkinson’s disease patients (PDMCI) by integrating multimodal indicators. We prospectively collected general demographic data, clinical scales, gait parameters, eye tracking parameters, and neuroimaging parameters from 50 PDMCI patients, 50 Parkinson’s disease patients with normal cognition (PDNCs), and 20 healthy controls (HCs). Support Vector Machine (SVM) classifiers and nested cross-validation were used to evaluate 31 feature combinations. Results demonstrated that the combination of clinical, gait, eye tracking, Diffusion Tensor Image Analysis along the Perivascular Space (DTI-ALPS), and Global Functional Connectivity Density (gFCD) features achieved an average accuracy of 0.9135 and an average area under the curve of 0.9602 on the test dataset. Notably, the combination of eye tracking and gait features also showed superior performance. These findings indicate that multimodal data integrated with machine learning (ML) can effectively distinguish between PDMCI and PDNC patients.</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

Construction of a mild cognitive impairment prediction model for Parkinson’s disease patients on the basis of multimodal data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月18日

查看全文
  • Chunyu Liang,
  • Yili Chen,
  • Yongyun Zhu,
  • Yangfan Zhu,
  • Jieyu Chen,
  • Chenxi Liu,
  • Fang Wang,
  • Xinglong Yang

摘要

This research aimed to establish a model predicting mild cognitive impairment in Parkinson’s disease patients (PDMCI) by integrating multimodal indicators. We prospectively collected general demographic data, clinical scales, gait parameters, eye tracking parameters, and neuroimaging parameters from 50 PDMCI patients, 50 Parkinson’s disease patients with normal cognition (PDNCs), and 20 healthy controls (HCs). Support Vector Machine (SVM) classifiers and nested cross-validation were used to evaluate 31 feature combinations. Results demonstrated that the combination of clinical, gait, eye tracking, Diffusion Tensor Image Analysis along the Perivascular Space (DTI-ALPS), and Global Functional Connectivity Density (gFCD) features achieved an average accuracy of 0.9135 and an average area under the curve of 0.9602 on the test dataset. Notably, the combination of eye tracking and gait features also showed superior performance. These findings indicate that multimodal data integrated with machine learning (ML) can effectively distinguish between PDMCI and PDNC patients.