From Motion to Cognition: Machine Learning Techniques for Parkinson's Disease Classification Using Multi-Source Data

This research explores the application of machine learning algorithms to improve early diagnosis and personalized treatment of Parkinson's Disease (PD), with an emphasis on cognitive decline and response to therapy. By analysing multimodal data from motion sensors during the instrumented Timed-Up-and-Go (iTUG) test and EEG recordings, we aim to accurately identify motor status due to dopaminergic treatment and cognitive impairments in patients with PD. Our study also focuses on the techniques of data analysis, including data cleaning and correlation methods, to enhance the accuracy and reliability of the machine learning models. Our findings indicate that Artificial Neural Networks (ANNs) excel in classifying cognitive states during specific phases of the iTUG test. Furthermore, K-Nearest Neighbors (KNN) and Random Forest algorithms demonstrate high accuracy in processing EEG data, effectively distinguishing patients affected by PD from healthy patients. These results underscore the potential of machine learning in enhancing the diagnostic and therapeutic processes for PD.