PADG-Pred: Exploring Ensemble Approaches for Identifying Parkinson's Disease Associated Biomarkers Using Genomic...

Area of Science:

• Biomedical Informatics
• Genomics
• Computational Biology

Background:

• Parkinson's disease (PD) is a neurodegenerative disorder impacting motor function, with complex genetic and environmental causes.
• Accurate and timely diagnosis of PD is crucial for effective treatment, yet current diagnostic methods face reliability challenges.
• Identifying reliable biomarkers for PD is essential for early detection and intervention strategies.

Purpose of the Study:

• To introduce PADG-Pred, a novel computational tool for identifying Parkinson's disease-associated biomarkers.
• To enhance the accuracy and interpretability of PD diagnosis through advanced genomic profiling and machine learning.
• To develop a robust predictive model that surpasses existing diagnostic tools in reliability and performance.

Main Methods:

• Utilized genomic profiling data for feature extraction via multiple statistical moments.
• Employed a diverse range of ensemble classification techniques, including XGBoost, Random Forest, Light Gradient Boosting Machine, Bagging, ExtraTrees, and Stacking.
• Applied rigorous validation procedures, assessing metrics like accuracy, specificity, sensitivity, and Mathew's correlation coefficient.

Main Results:

• The PADG-Pred model, specifically PADG-RF, demonstrated high performance across validation tests.
• Achieved consistent accuracy rates of approximately 91% on an independent dataset.
• Showcased superior cross-validation accuracy, reaching ~94% for 5-fold and ~96% for 10-fold validation.

Conclusions:

• PADG-Pred offers a robust and interpretable approach for identifying Parkinson's disease biomarkers.
• The high accuracy achieved by PADG-RF highlights its potential as a valuable tool in PD diagnostics.
• This genomic and machine learning-based strategy represents a significant advancement in the pursuit of reliable PD detection methods.