Diagnosing Parkinson’s Disease Using Multimodal Physiological Signals

Parkinson’s disease (PD) is the second most common neurodegenerative disease after Alzheimer’s disease. Due to the complex etiology and diverse clinical symptoms, it’s difficult to accurately diagnose PD. In this study, we applied multimodal physiological signals, which include electroencephalography (EEG), electrocardiogram (ECG), photoplethysmography (PPG), and respiratory (RA), to classify PD and healthy control (HC) based on a multimodal support vector machine (SVM). Our experiments achieved an accuracy of 96.03%. Besides, we performed statistical analysis on the four types of physiological data of the PD group and the HC group. Results showed that the EEG of non-dementia PD patients had a significant decrease in high-frequency power, and the high-frequency energy distribution of the normalized PPG signal increased compared with HC. The current study suggests that combining the physiological information of multiple models and machine learning methods could improve the diagnosis accuracy of PD disease and be a potentially effective method of clinical diagnosis.