Industry Track

Abstract

As artificial intelligence reshapes the delivery, administration, and governance of healthcare, the federal response has never been more consequential — or more complex. In this fireside chat, Jesse Isaacman-Beck, Director of AI Policy and Strategy at the U.S. Department of Health and Human Services (HHS) and senior leader within the Office of the National Coordinator for Health Information Technology (ONC), offers an inside look at how the federal government is navigating this pivotal moment.

Drawing on his work developing the HHS AI Strategy, leading ONC's efforts on data interoperability, and engaging the broader health informatics community through HHS's recent Request for Information, Isaacman-Beck will share the department's vision for a health system where data flows freely, technology evolves responsibly, and AI genuinely expands access and affordability of care. He will offer candid early insights from community feedback on how stakeholders — from health systems to researchers to patients — want to see federal AI priorities implemented in practice.

The conversation will explore what it means to govern AI in a rapidly shifting policy landscape, how ONC interoperability mandates intersect with the emerging demands of AI-enabled workflows, and where the federal government sees the greatest opportunities — and the most pressing risks — in the years ahead. The session will close with an open audience Q&A, offering ICHI participants a rare opportunity to engage directly with federal leadership on the policies shaping their research, their institutions, and their patients.

Date & Time: June 2, 2026, 8:15–8:35 AM CT

Abstract

Large Language Models, (LLMs) are extensively being used by the healthcare profession. These systems are aiding the clinician in decision making by analyzing unstructured text, medical literature, and various guidelines to provide precise and personalized treatment in line with the P9 concept that we outlined in our earlier work. However, there is a need for rigorous measurement and evaluation of these systems. This talk delves into the emerging field of AI metrology, which seeks to establish standardized methods for assessing the accuracy, reliability, and trustworthiness of AI models in general, LLMs in particular. After providing a brief overview of LLMs use in health care, we explore state-of-the-art evaluation metrics designed to assess health care chatbots from an end-user perspective These metrics encompass an evaluation of language processing abilities, impact on real-world clinical tasks, and effectiveness in user-interactive conversations. Finally, we discuss ongoing work in Agentic AI and the specific metrics required to evaluate these architectures.

Abstract

Today, medical discovery advances one clinical trial at a time, each taking years to execute and often costing $100 million or more. As we enter the era of precision health in which we recognize that “one size doesn't fit all” and thus try to tailor treatments for each individual, continuing on today's discovery processes is clearly not sustainable. The confluence of technological advances and social policies has led to rapid digitization of multimodal, longitudinal patient journeys, such as electronic health records (EHRs), imaging, and multiomics. Our overarching research agenda lies in advancing multimodal generative AI to learn the language of patients and create a virtual patient world model as digital twin for forecasting disease progression and treatment response. This enables us to synthesize population-scale real-world evidence from hundreds of millions of patients and accelerate medical discovery through AI-powered virtual clinical trials, in deep partnerships with real-world stakeholders such as large health systems and life sciences companies.

Abstract

Artificial intelligence in healthcare is often presented through model performance, but real-world impact depends on much more than accuracy alone. In ambulatory cardiac monitoring, the challenge is not simply detecting arrhythmias from ECG signals; it is building AI that can operate reliably across noisy, heterogeneous, real-world data, fit into clinical workflows, support expert review, and earn trust from clinicians, patients, and health systems.

In this keynote, Ben Teplitzky will trace the evolution of BeatLogic from an early deep-learning platform for comprehensive ECG annotation to a broader case study in healthcare AI enablement. Early work showed that combining deep learning with high-quality expert-adjudicated data could substantially improve beat and rhythm interpretation in ambulatory monitoring. But that technical milestone was only the beginning. Translating AI into practice required confronting the harder informatics problems: dataset design, annotation quality, edge-case discovery, validation strategy, workflow integration, human oversight, and continuous iteration as the system encountered new conditions in the field.

Using lessons from cardiac rhythm management and broader experience across healthcare AI, this talk will examine what it takes to move from an impressive algorithm to a trusted operational capability. The session will highlight practical strategies for developing clinically meaningful real-world evidence, managing model evolution responsibly, balancing automation with expert judgment, and building systems that are robust not only in development but also in deployment. The goal is to offer a pragmatic roadmap for organizations seeking to create human-centered AI that is safe, scalable, and genuinely useful in healthcare.