AI for Biomedical Sciences

The workshop highlighted the burgeoning landscape of this field, with over 4,000 companies globally engaged in AI for biomedical sciences. Despite significant advancements, successfully navigating drugs through clinical trials remains difficult, with a notable proportion of both AI-designed and human-designed drugs failing during this crucial phase. Singapore aims to capitalize on its strengths in data quality, expertise, and crucially the diversity of its gene pool, to carve out a competitive niche in AI-driven drug discovery. 

A number of challenges were discussed, including limitations in AI models to leverage multi-modal/ multi-omic datasets and to generate valid hypothesis for experimental planning. Improving computational resources and addressing the need for increased model explainability were identified as crucial areas for future development. It was mentioned that efforts are also underway to optimize AI models and datasets, with a focus on improving prediction accuracy and exploring AI's potential in identifying novel drug targets from genomic data. Overall, the workshop underscored the promise of AI in biomedical sciences, the challenges it faces, and the opportunities for Singapore to lead the way in this critical domain.