Machine Learning-Base Drug Repurposing for Amyotrophic Lateral Sclerosis (ALS)

<p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="first" dir="auto" id="d825523e87">This study presents a machine learning-based framework for drug repurposing with open source datasets, leveraging cheminformatics and generative modeling to identify compounds with potential therapeutic relevance for neurodegenerative diseases such as Amyotrophic Lateral Sclerosis (ALS). A Random Forest classifier was trained on molecular fingerprints derived from known ALS drugs and structurally diverse non-ALS compounds, achieving an overall accuracy of 91.2% despite class imbalance. To explore a deep learning framework, a Generative Adversarial Network (GAN) was developed and trained to produce novel drug-like fingerprints. All ten generated samples were predicted as ALS-targeting by the classifier. Structural similarity analysis revealed resemblance to approved drugs, with a Tanimoto similarity of 0.214. Dimensionality reduction via PCA and t-SNE showed clustering between generated and real ALS-related fingerprints. These results support the utility of combining classification and generative models for computational drug repurposing in neurodegenerative disease research.