By Niven R. Narain, Ph.D., CEO & President, BPGbio, and Forbes Councils Member
AI has been hailed as the Lionel Messi of the pharma industry’s productivity problem. But just like star players with high salaries don’t necessarily lead to more game-winning goals, nearly $60 billion in investments in companies claiming their AI can find the next blockbuster therapy have yet to translate into new drug approvals.
The problem isn’t AI itself—it’s the approach most companies have taken to its use. The data inputs used to train models, how in silico predictions are validated and the questions we ask AI models all contribute to the industry’s collective failure to realize the promise of AI in drug discovery and development.
Better Data Inputs For Solid AI Predictions
The underlying issue for pharma companies using AI lies in the quality of the inputs and how AI models are used. Think of AI as the teamwork and coaching that enable players to reach their full potential. Most AI models are built using publicly available datasets or limited representations of patient biology, and many are designed only to identify basic correlations. These approaches are always going to be insufficient for providing unbiased answers because they aren’t designed to make big-picture inferences.
(continues)
For the full Forbes article, please visit AI Can Give Pharma More Shots On Goal, But Robust Data And Validation Are Key To Drug Approvals
