Validating AI in Digital Pathology for Clinical Use Beyond the Algorithm

About this Class

Speaker will: 

• Address how robust AI models in digital pathology are limited not only due to the scarcity of large-scale, annotated data sets but also due to the lack of clear framework for clinical translation. • exemplify a deep learning model that can generate synthetic, but plausible, histological stains from unstained tissue.
• Highlight how NIH exploratory grants (e.g., NIH R21) have made it possible for a feasible technology to become a clinically validated, risk-probability tool (i.e., prognostic model)
• Give perspective on both emerging challenges and standards for validating and approving these complex, AI-enabled medical devices intended for clinical use.