LSINW: AI in Drug Discovery
Each year, Life Science Washington brings together industry experts in the life sciences to present the latest insights and developments to attendees at Life Science Innovation Northwest. This year’s virtual sessions introduced common themes related to the role AI can play in drug discovery and development, what problems AI could help solve moving forward, and how collaborations can drive better outcomes in the future.
AI can assist at almost every phase in drug discovery
AI and machine learning can be a valuable tool in many different domains, including drug discovery and development. According to Dr. Kim Branson, Senior Vice President and Global Head of AI/ML at GSK, AI can be used at almost every single phase in the process of drug discovery.
In Dr. Branson’s Keynote Address – Embracing AI as a Tool for Drug Discovery, he shared that due to the exponential nature of data generation, AI methods are becoming more widely used in the industry. Among the applications for machine learning in drug discovery, Dr. Branson identified that AI can be used to collect data, process data, and derive causal relationships as well as to aid in designing future experiments efficiently. This poses AI as a valuable tool for early discovery.
Data scientists need to frame the right problem
While AI can be a vital tool in drug discovery and development, it can only bring value if it is being used to solve the right problem. Jeff Elton, CEO of ConcertAI, Lara Mangravite, President of Sage Bionetworks, and Julie Sunderland, Managing Director of Biomatics Capital examined how AI can be successfully brought to clinical use in a panel on Machine Learning and Drug Discovery.
Sutherland reminded the audience that AI is a tool that can be applied to solving a problem and that scientists should work to identify a clinically relevant problem first.
“Don’t think about hiring AI, think about what dataset and problem you are trying to solve. How do you get the capabilities to work through that data?” Sutherland says.
Dr. Elton echoed Sutherland, emphasizing that scientists should focus on framing a valuable question and identifying potential health outcomes and benefits for patient populations before moving forward with data. According to Dr. Elton, bringing together the subject matter expertise of data scientists and epidemiologists is crucial to effectively solve clinically relevant problems.
AI is only as good as the data
The role that data plays in developing AI models cannot be overstated. Access to data is a current limitation of early drug discovery and development. Sutherland shared that data is often fragmented and variable, making it difficult to pull strong individual datasets that are representative of populations.
Dr. Mangravite emphasized that, when working with data, scientists should actively work to understand what is and is not a sensible thing to do with it. To overcome this challenge, Dr. Mangravite advocates for teams bringing in subject matter experts from different domains to leverage their expertise and gain a total understanding of appropriate uses for data.
The true value of AI in drug development will take years to uncover
Acknowledging the hype of AI, Sutherland expressed that she is both “optimistic and skeptical” about the potential benefits that AI can bring to the industry. While Sutherland is optimistic that AI will bring us toward better and more efficient development, her skepticism stems from how data scientists and life science experts still work in silos.
Sutherland emphasized that we need to bring together those in the life sciences with AI experts to drive us toward a revolution in the industry. It is only when we leverage the subject matter expertise of both professions that we will be able to improve disease management.