Lead Data Scientist @ Novo Nordisk | Integrated Omics in Clinical Trials | Computational Systems Biology | Data Science I Applied ML/AI | Strategy & Innovation | Mentor | Thought Leader | Scientific Advisory Board Member
Nice read navigating through hype & promises around role of AI in drug discovery. There is a lot of potential if used in the right way in the right phase in different stages of drug development gates be starting from • Target ID, validation • Lead Seleciton • Lead optimization • Preclinical • Clinical Trial followed by • feedback mechanism for Drug Repurposing However, something to realize that it is not a magic bullet (will not be a button on your robot or super computer 😉) & if we are talking about a drug compound that pill is not insilico. 😉Validation & translatability will be key, btw, blind tests are a thing in AI models while measuring performance & we still know there are some caveats around biases in such steps. How we feed such or loop it back, with longitudinal designs of data from translational validations & clinical trials using Reinforcement learning from human feedback (RLHF) might be interesting to learn in next years. This is where AI might push certain frontiers but it will still be taking time to reach that precision specifically given this part isn’t just a modeling exercise but engineering as well to leverage full potnetial of AI. Having said that, still hope to see what we can learn from our failures to build some AI-based Ph III approved drugs in near future. All it needs is a couple of approvals. 😃
