中文摘要：

针对目前基于模式识别的语音障碍帕金森病诊断可解释性差、可推广性差的问题，本文提出基于多维筛分类器的可视化帕金森病诊断。该分类器具有全程可视化的特点，在保证诊断精度的同时，可以将特征进行可视化表示。可视化的引入不但使操作者了解不同特征对于诊断的重要性，而且可以发现最具诊断价值的特征，有助于简化帕金森病的诊断过程并提高诊断水平。

英文摘要：

The Parkinson＇s Disease （PD） based on speech features in classical pattern recognition is intrinsically deficient in ex- plainable and generation. In this paper, the visual combing classifier （VCC） is introduced to the diagnose of PD. The VCC is characteristic of visualization throughout the procedure. As a result of that, the features are visualized as the necessary precision is guaranteed. Introducing the visualization make users understand each features＇ contribution in diagnosis and emerge the most value features. It is promising to refine the diagnosis procedure and improve the diagnosis precision.