NeuroRepurpose: An In Silico Approach to Repurposing Drugs for Multiple Sclerosis Treatment

Multiple Sclerosis affects nearly 2.9 million people worldwide. Recent studies have identified STAT3 as a key factor in the disease, making it a prime target for drug development. However, most STAT3 inhibitors are extremely expensive, and creating cheaper alternatives could take another decade to refine and bring to market and the public. To address this, I developed a neural network designed to differentiate between STAT3 inhibitors and non-inhibitors. I trained the model using experimentally verified inhibitors sourced from PubMed. After clustering the data and applying feature selection, I optimized the model for accuracy. Once the model was trained, I screened FDA-approved drugs to identify potentially repurposable treatments. The top 9 candidates were subjected to molecular docking simulations to test their binding affinity for STAT3. Finally, I conducted ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) predictions to assess their suitability as viable treatments, and tested the compound in an artificial neural tissue environment to ensure accuracy. This approach revealed Celecoxib as the most viable treatment, and could revolutionize Multiple Sclerosis treatment plans for the better.