CDSC AI Validation Framework

Please join us for this month's meeting of the Bio-Data Club. Rodrigo Carvajal and Jorge Barrios, Ph.D. will introduce the CDSC AI Validation Framework, a structured methodology designed to obtain reliable, evidence‑based data from clinical free‑text documents using large language models.

The framework spans patient cohort identification, targeted document selection, AI‑driven data extraction, expert validation, iterative refinement, and ongoing reliability assurance.

As a practical example, we will demonstrate how the framework is applied to extract Lung‑RADS categories and modifiers directly from CT findings documents. This use case illustrates how validated AI methods can support consistent and trustworthy reporting of findings from low‑dose CT lung cancer screening studies while maintaining transparency and clinical confidence.