EEFSA-SECM: an enhanced ensemble feature selection and stacking ensemble classifier to detect Parkinson's disease.

Area of Science:

• Neurology
• Computer Science
• Biomedical Engineering

Background:

• Parkinson's disease (PD) diagnosis is challenging due to subtle, early symptoms often missed by standard methods.
• Machine learning (ML) can identify speech-based biomarkers imperceptible to human analysis.
• Early detection of PD is crucial for effective patient management and treatment.

Purpose of the Study:

• To develop and evaluate an Enhanced Ensemble Feature Selection Algorithm (EEFSA) for improved Parkinson's disease classification.
• To combine filter, wrapper, and embedded feature selection methods to identify optimal speech-based biomarkers.
• To enhance classification performance and reduce computational time in PD detection using ML.

Main Methods:

• Proposed an Enhanced Ensemble Feature Selection Algorithm (EEFSA) integrating multiple feature selection strategies.
• Applied EEFSA to reduce high-dimensional feature sets from two audio-based Parkinson's disease datasets.
• Developed a Stacking Ensemble Classifier Model (SECM) using logistic regression as the meta-classifier.

Main Results:

• EEFSA effectively reduced feature dimensionality from 46 to 20 (Dataset-1) and 754 to 40 (Dataset-2).
• The SECM achieved high classification accuracies of 86.67% (Dataset-1) and 89.95% (Dataset-2).
• EEFSA-driven dimensionality reduction improved classification accuracy, reduced training time, and minimized overfitting compared to individual classifiers.

Conclusions:

• The EEFSA-SECM framework offers an efficient and effective method for Parkinson's disease classification.
• This approach demonstrates strong performance in terms of accuracy, training/testing times, and AUC scores.
• The proposed method represents a significant advancement in leveraging ML for early and accurate PD diagnosis from speech.