Project Grant 2434949

This $499,742 federal Project Grant award from the National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070) aims to develop a novel AI framework for designing safer, non-addictive pain relief medications. The primary objectives are to create an AI system that generates candidate drug molecules capable of enhancing the body's natural pain control mechanisms, and to establish a transparent machine learning model that can extract hidden scientific rules and constraints to improve the reliability of the AI-generated solutions. The award recipient, the Regents of the University of Michigan, will carry out the key research activities, which include developing a generative AI system using conditional variational autoencoders with grammar-based and physics-informed constraints, as well as designing a reinforcement learning-based transparent machine learning model to uncover unknown scientific constraints. This project seeks to address the urgent public health challenge of the opioid crisis by leveraging advanced AI techniques to revolutionize the drug design process and produce effective pain relief medications without the risk of addiction.