Towards a Machine Learning Empowered Prognostic Model for Predicting Disease Progression for Amyotrophic Lateral...

Area of Science:

• Neuroscience
• Biomedical Informatics
• Computational Biology

Background:

• Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disease with unpredictable progression.
• Accurate prediction of ALS progression is crucial for effective patient management and care.
• Existing tools lack the precision needed to forecast disease trajectory reliably.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting ALS progression.
• To accurately forecast the reduction in ALS Functional Rating Scale (ALSFRS) scores over 3 to 12 months.
• To identify the most effective machine learning approach for ALS prognostication.

Main Methods:

• Utilized the Pooled Open Access Clinical Trials (PRO-ACT) database.
• Developed and compared an extensive set of machine learning models, including screener-learner approaches.
• Paired 5 feature selection algorithms with 17 predictive models and 4 ensemble models.

Main Results:

• The study demonstrated promising predictive capabilities using machine learning models.
• A blender-type ensemble model achieved the highest prediction accuracy.
• The best-performing model exhibited significant prognostic potential for ALS progression.

Conclusions:

• Machine learning, particularly ensemble methods, can accurately predict ALS progression.
• The developed models offer a valuable tool for forecasting ALSFRS score decline.
• These findings can contribute to improved patient care and clinical trial design for ALS.