Feature selection using metaheuristics to predict annual amyotrophic lateral sclerosis progression.

Area of Science:

• Neuroscience
• Computational Biology
• Medical Informatics

Background:

• Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disease impacting motor neurons, leading to significant functional decline.
• Currently, no curative treatments exist for ALS, making accurate prediction of disease progression and survival critical for patient management.

Purpose of the Study:

• To develop and validate machine learning models for predicting ALS progression and patient survival.
• To identify key factors influencing disease trajectory to improve patient care strategies.

Main Methods:

• Utilized data from multiple cohorts (PRO-ACT, ExonHit, PULSE) comprising 5041 ALS patients.
• Applied various machine learning techniques, including logistic/linear regression (LR) and light gradient boosting machine (LGBM), with feature selection via ANOVA and differential evolution (DE).
• Validated models using 10-fold cross-validation and Kaplan-Meier estimates for patient clustering.

Main Results:

• LR with DE achieved 76.33% accuracy and an AUC of 0.84 for predicting survival.
• Identified five distinct patient clusters using Kaplan-Meier analysis (C-index = 0.8).
• LGBM models accurately predicted ALS Functional Rating Scale (ALSFRS) scores at 3 months with an adjusted R² of 0.764.

Conclusions:

• Machine learning models demonstrate significant potential in predicting ALS progression and survival.
• These predictive insights can enhance the understanding of ALS disease dynamics and support personalized patient care.