About the Seminar

May 13, 2026 9:00am—10:30am

How can federal agencies and policymakers assess the ways in which financial sponsorship shapes the evidence base used to inform regulation, treatment coverage, and public health decision-making? As scientific output grows at an exponential scale, existing oversight tools such as disclosure notices struggle to keep pace with the volume and complexity of potential conflicts of interest in health research. This talk introduces InfluenceMapper, a large language model (LLM)–based tool designed to extract, structure, and analyze disclosure information from scientific publications at scale. The presentation describes the development of the tool and presents empirical findings regarding patterns of sponsorship bias across areas of health research. It also examines how such tools can support agency missions related to research integrity, evidence evaluation, and public trust. The talk concludes by reflecting on the broader opportunities and challenges posed by the use of LLMs and artificial intelligence in science and public policy, including issues of transparency, validation, governance, and responsible deployment to strengthen evidence oversight.

Links

• Youtube Link