Demo: H2AI: A Framework for Experiential Learning and De-Risking Generative AI in Healthcare

Rohan M. Rani, Jacqueline Sandling, Francis Arellano, Nika Shroff, Yumin Gao, Raj Ratwani

Keywords: generative AI, hackathon, interdisciplinary, clinical translation, ai ethics, healthcare innovation

TL;DR: This paper presents H2AI, a structured, university-based hackathon framework that successfully de-risks and accelerates the development of patient-centered generative AI solutions.

Abstract: The rapid evolution of generative AI presents immense opportunities for healthcare, yet a significant gap exists between technological potential and clinical application. This gap is exacerbated by a lack of supported environments where emerging leaders can apply new technologies to complex health problems. Traditional innovation pipelines are often too slow and siloed, while unstructured events like conventional hackathons frequently fail to produce lasting, clinically relevant solutions. This paper demonstrates the H2AI (Health and AI) hackathon as a working system and replicable framework designed to bridge this gap. Over two years, the H2AI model has evolved a structured, four-pillar approach: 1) curated interdisciplinarity in a university setting, 2) a technology-agnostic platform for agility, 3) patient-centricity by design, and 4) an integrated ethical and safety scaffolding. A longitudinal analysis of outcomes from 2024 to 2025 reveals significant growth in participation and a marked maturation in the technical sophistication of prototypes, evolving from general workflow tools to specialized, multimodal platforms targeting complex diseases like Parkinson’s. By embedding a formal curriculum on trustworthy AI and creating a structured pipeline to clinical translation, H2AI functions as a de-risking mechanism for early-stage innovation. This paper presents H2AI as a scalable model for experiential learning that accelerates the development of responsible, patient-centered generative AI solutions.

Submission Number: 124