‹ Back

ELHS Institute Proposes a Vision to Accelerate Open Clinical AI Science

 

📅 December 15, 2025

 

 

Although generative AI (GenAI) is transforming health care, its full clinical impact may take decades to realize due to persistent inefficiencies in clinical science. At the Chen Institute’s AI Accelerated Science Symposium on October 28, 2025, in San Francisco, ELHS Institute founder Dr. AJ Chen—together with collaborators from Stanford and Harvard Universities—proposed a new vision for Open Clinical AI Science (OCAIS). He also presented a practical framework that delivers free GenAI-based disease prediction services to clinical teams worldwide, including those in low-resource settings.

This approach enables broad participation in the generation and dissemination of clinical evidence for the responsible use of GenAI across all diseases. Widespread adoption of open, responsible GenAI services by physicians can support more timely diagnoses and treatments, facilitate the sharing of synthetic data and fine-tuned AI models, and ultimately accelerate open clinical AI science by an order of magnitude.

Dr. Chen has long recognized the risk of repeating the history of failed AI revolutions in health care. Since 2000, three major technological efforts to transform health care have fallen short. The traditional clinical AI wave effectively stalled around 2011, exemplified by the limitations of the IBM Watson chatbot. The first genomics-driven precision medicine wave collapsed in 2000, and the second wave derailed in early 2025. Dr. Chen believes that the prolonged timeline required for clinical validation was a root cause of these failures. If left unaddressed, the slow pace of clinical science may cause GenAI to follow a similar trajectory.

The OCAIS vision builds on the Learning Health System (LHS) framework proposed by the U.S. National Academy of Medicine (NAM), which seeks to address inefficiencies in knowledge generation and dissemination. However, many mainstream LHS implementations rely on large, complex infrastructures that require substantial resources, limiting scalability. In contrast, ELHS Institute’s prior pioneering work demonstrated that a minimal, ML-enabled LHS unit, focused on a single clinical task, can be implemented by any clinical team while still achieving the two core objectives of LHS: rapid evidence generation and dissemination.

The convergence of GenAI and task-specific LHS units offers a scalable path to shift clinical evidence generation timelines from decades to years, enabling the democratization of high-quality health care for people everywhere.

Source: ELHS Institute

🤖 ELHS GenAI Copilot Platform

🌐 ELHS Institute Website