A great talk from Gilead's Jeremy Zhang on scaling LLMs like OpenAI and Claude to the clinical po...

A great talk from Gilead's Jeremy Zhang on scaling LLMs like OpenAI and Claude to the clinical portfolio at BiotechX.

LLMs have already transformed the way they do clinical development. LLMs are being embedded into the workflow. "A few models instead of many"

Key use cases include data engineering, insight generation, augmenting humans, and ultimately evidence generation.

As so often, stakeholder skepticism from internal stakeholders is a key hurdle for adoption, hence the focus on human augmentation.

Regulatory clarity is also still very much an emerging issue that is not solved.

3 focus areas:

• protocol design
• clinical ops
• downstream analysis

A fantastic use case was presented on analyzing clinical protocol design to get rid of protocol hurdles.

Gilead also used tweaked LLMs to effectively run complex queries on RWE in natural language to generate insights

Finally, automate clinical document generation is currently being scaled for all types of forms from consent to study reports, even creating a dedicated word plugin for that use case.