Utilizing a Non-Motor Symptoms Questionnaire and Machine Learning to Differentiate Movement Disorders

Area of Science:

• Neurology
• Digital Health
• Machine Learning

Background:

• Parkinson's disease (PD) is a progressive neurodegenerative disorder impacting quality of life.
• Early diagnosis and treatment of PD are crucial for reducing patient burden and healthcare costs.

Purpose of the Study:

• To evaluate the diagnostic potential of the International Parkinson and Movement Disorder Society's Non-Motor Symptoms (NMS) questionnaire.
• To assess the utility of patient-completed NMS data, collected via a smartphone system, for PD diagnosis.

Main Methods:

• A prospective, single-center study involving 489 participants: PD group, healthy control (HC) group, and differential diagnosis (DD) group.
• Data collection using a smartphone-based system for patient-completed NMS questionnaires.
• Machine learning (ML) based classification models were employed for cross-validation.

Main Results:

• Significant differences in NMS were observed between the PD, HC, and DD groups.
• ML classification achieved high balanced accuracy: 88.7% for PD vs. HC, 72.1% for PD vs. DD, and 82.6% for (PD + DD) vs. HC.
• NMS questionnaire data demonstrated high feature importance for diagnostic classification.

Conclusions:

• The NMS questionnaire is a valuable tool for supporting the early diagnosis of Parkinson's disease.
• Self-administered NMS questionnaires, analyzed with ML, offer a promising approach for objective PD detection and differentiation.