Recent Advances in AI-Assisted Methods and Software for Modern Biopharmaceutical Development

AI tools for large language modeling (LLM) have had a rapid and accelerating impact on scientific practice. In biopharmaceutical development, LLMs are already used to develop standard forms and reports, summarize literature, write computer code, draft technical reports, and many other functions. Now FDA has announced they too will scale use of a generative AI pilot (and their own internal AI chatbot, Elsa) to assist scientific reviews. However, AI tools require training with both data and domain expertise, and they remain prone to mistakes that must be spotted and corrected by humans. This session features four speakers working on different aspects of AI-assisted biopharm development. Speaker 1 (Haoda Fu, Amgen) is AI/ML enterprise lead for a major pharmaceutical company. He will review the historical progression of AI and its growing role in pharmaceutical research, highlighting how AI-driven methodologies are revolutionizing drug discovery and development, and lay out a vision for a thoughtful balance between leveraging AI's capabilities and maintaining rigorous scientific and ethical standards. Speaker 2 (Krishna Padmanabhan, Madrigal Pharma) will then describe BACTA-GPT, a custom fine-tuned large language model (LLM) that translates natural language instructions into executable R code for Bayesian adaptive clinical trial design. His talk will share effective prompting strategies, and evaluate the new tool's performance in supporting adaptive trial simulations. Speaker 3 (Dan Zhao, Servier) will turn to the subject tumor MRI imaging, where AI enhances the workflow from acquisition optimization and automated segmentation to quantitative feature extraction, ultimately enabling prediction of tumor aggressiveness and treatment response. Her talk will include radiogenomic prediction of molecular subtypes, differentiation of pseudoprogression from true progression, and integration of imaging biomarkers into prognostic models. Speaker 4 (Raviv Pryluk, PhaseV Trials) is CEO of a young technology company developing AI-augmented software for clinical trial design and analysis, including automated literature search and subgroup discovery. His talk will introduce a novel ensemble estimation approach that leverages causal ML methods to enhance the detection and assessment of heterogeneity in adaptive trials. The framework uses conformal prediction to assess uncertainty in its ML estimates for finite samples, facilitating a more nuanced understanding of how different patient subgroups respond to treatments. Finally, the Discussant (J. Jack Lee, formerly FDA, currently UT MD Anderson Cancer Center) will comment on the presentations and offer his views on the routine incorporation of AI into regulatory science and biopharm development more broadly. The proposed session demonstrates how cross-sector statistical communities are actively shaping trustworthy uses of LLMs in biopharmaceutical development. Speakers from competing research teams will share concrete methods for integrating domain expertise, rigorous validation, and human oversight to convert AI's speed into reproducible evidence and equitable decisions. The session embodies JSM2026's "Communities in Action" theme, and helps translate innovation into societal benefit for patients and public health. The session should appeal to a large collection of biostatisticians and trialists hoping to use modern AI-assisted adaptive, ML, and causal inference tools on a routine basis.