Development of a depression in Parkinson's disease prediction model using machine learning

Area of Science:

• Neurology
• Geriatrics
• Psychiatry

Background:

• Depression in Parkinson's disease (DPD) significantly impacts patient quality of life.
• Early diagnosis and identification of DPD predictors are essential for effective management.

Purpose of the Study:

• To develop a predictive model for DPD using support vector machines.
• To identify key predictors including sociodemographic factors, health habits, PD symptoms, sleep disorders, and neuropsychiatric indicators.

Main Methods:

• Analysis of 223 patients aged 60+ with Parkinson's disease (PD).
• Depression assessment using the Geriatric Depression Scale (30 items).
• Support vector machine model developed using PD motor signs, REM sleep behavior disorders, and neuropsychological tests.

Main Results:

• Late motor complications, specifically levodopa-induced dyskinesia, emerged as the most significant risk factor for PD motor symptoms.
• Functional weight analysis highlighted the influence of motor symptoms on DPD prediction.

Conclusions:

• Developing customized screening tests for early DPD detection is necessary.
• Continuous monitoring of high-risk groups, identified through the predictive model, is vital for maintaining emotional health in PD patients.