Machine Learning for Early Detection of Cognitive Decline in Parkinson's Disease Using Multimodal Biomarker and...

Area of Science:

• Neuroscience
• Gerontology
• Biomedical Engineering

Background:

• Parkinson's disease (PD) is a prevalent neurodegenerative disorder impacting cognition.
• Cognitive decline (CD) in PD is an early indicator of dementia, necessitating timely risk assessment.
• Predicting CD is vital for proactive intervention and personalized patient management.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for predicting 5-year CD risk in early-stage PD patients.
• To identify key demographic, clinical, and biomarker predictors of CD in PD.
• To assess the efficacy of ML algorithms in early CD risk stratification for Parkinson's disease.

Main Methods:

• Utilized longitudinal data from the Early Parkinson's Disease Longitudinal Singapore study (2014-2018).
• Defined CD as a significant decline in Montreal Cognitive Assessment scores over two consecutive years.
• Applied and compared four ML methods: AutoScore, Random Forest, K-Nearest Neighbors, and Neural Network, using baseline data.

Main Results:

• Key predictors included education, blood pressure, Hoehn and Yahr scale, BMI, and specific biomarkers (phosphorylated tau, total tau, NfL, ST2).
• Random Forest demonstrated superior performance, achieving an Area Under the Curve (AUC) of 0.93 (95% CI: 0.89, 0.97).
• The model effectively identified individuals at high risk for cognitive decline in early PD.

Conclusions:

• Machine learning models can effectively identify early-stage PD patients at high risk for cognitive decline.
• These predictive models support targeted interventions and enhanced management strategies for Parkinson's disease.
• ML-driven risk assessment offers a promising approach for improving patient outcomes in PD.