New Frontiers of AI for Drug Discovery and Development

NeurIPS 2023 Workshop @ New Orleans, Louisiana, USA

Room 242 @ Ernest N. Morial Convention Center

December 15, 2023

Roundtable Discussions

The Opportunities and Challenges of Integrating ML into Clinical Trials

Facilitator: Karen Sayal (GSK)
Description: Clinical trials are central to the drug development pipeline and bridge preclinical drug discovery with routine clinical care. All medicines developed for patients undergo extensive evaluation in clinical trials to ensure they are safe and efficacious for patients. Designing and running clinical trials is an intensive, long and expensive process. The advances driving ML could be applied to multiple components of clinical trial design in order to accelerate the pace of drug discovery. The tasks best placed to benefit from ML are: patient identification and trial recruitment, safety monitoring, biomarker discovery, and generalising clinical trial findings to the real-world clinical context.

There are challenges. Clinical trials are highly-regulated to safeguard patient safety. Therefore, ML-driven improvements must be designed to complement, not replace, existing trial workflows and fulfill regulatory requirements. Data scale and data quality is a critical bottleneck. Clinical trials recruit sample numbers below the traditional thresholds required for training deep learning models. New algorithmic approaches will be required, which can either (a) achieve acceptable performance in the low-sample regimen, or (b) better integrate real-world and clinical trial datasets. The workshop discussion will be a platform to discuss potential solutions to these algorithmic challenges.

From Drug Discovery to Drug Design: AI-driven Generative Chemistry

Facilitator: Bülent Kiziltan (Novartis)
Description: Drug discovery has been going through a rapid transformation where AI/ML plays an increasingly important role in accelerating and enhancing molecule design. Generative methods have been used to augment computer aided drug design pipelines that inform chemists on the molecular properties for more than a decade. One primary objective has been to accelerate the molecule design process. With many approaches (e.g., topological/geometric deep learning, LLMs) becoming mature for technological implementation, in this roundtable we will discuss the current state of AI-enabled drug discovery and explore the potential of the newest AI/ML methods to transform the domain into AI-driven drug design.

AI/ML Based De Novo Design for Biologics

Facilitator: Haoda Fu (Eli Lily)
Description: AI/ML-Based De Novo Design for Biologics revolutionizes the field of biopharmaceuticals by harnessing artificial intelligence and machine learning to innovate and expedite the design of novel biologic drugs. This transformative approach fosters the creation of therapeutics with enhanced specificity and efficacy, enabling rapid, informed decision-making in drug development. It encapsulates a future-forward vision of biopharmaceutical research, promising advancements that stand to redefine therapeutic design paradigms and propel the industry into a new era of scientific excellence and medical discovery.

Collaboration Between AI Experts and Domain Specialists

Facilitator: Quanquan Gu (UCLA & ByteDance)
Description: Given the role and potential of AI/ML in drug discovery, it is critical to develop the skills for effectively collaborating in interdisciplinary teams. We will strategize ways to bridge the gap between computational approaches and experimental validation. We will also discuss the emerging technologies and trends in AI that will likely impact drug design, particularly their anticipated advacements and potential implications.