Translational AI in Medicine: Unlocking AI’s Potential in Health Care with Nigam Shah

In this episode of NEJM AI Grand Rounds, Raj Munrai and Andy Beam interview Dr. Nigam Shah, a professor of medicine at Stanford University and the chief data scientist for Stanford Healthcare. Dr. Shah discusses his meandering path into AI, his work on ontologies and knowledge representation, and his role in bringing AI into clinical use safely, ethically, and cost-effectively at Stanford Health Care. The conversation explores the challenges of deploying AI in healthcare, the importance of focusing on productivity gains rather than just efficiency, and the potential of large language models (LLMs) in medicine. Dr. Shah also shares his thoughts on the AI ecosystem, the need for open-source alternatives, and the democratization of data to empower patients.

Outlines

Part 1: Introduction and Background

Part 2: AI Implementation in Healthcare

Part 3: Challenges and Future of AI

Sign in to continue reading, translating and more.

Open full episode in Podwise

NEJM AI Grand Rounds

Part 1: Introduction and Background

Introduction to AI in Healthcare and Nigam Shah's Background

Nigam Shah's Journey into AI and Biomedical Informatics

Postdoctoral Work and the Role of Ontologies in AI

Nigam Shah's Role as Chief Data Scientist at Stanford Health Care

Collaboration with Clinicians and Problem-Solving Approaches

Part 2: AI Implementation in Healthcare

Piloting AI in Clinical Use at Stanford HealthCare

Pilot AI Studies and the Impact on Healthcare Delivery

Efficiency vs. Productivity in Healthcare AI

The Future of Primary Care and the Role of AI

Basics of How LLMs Work

Applications and Limitations of LLMs in Medicine

Training LLMs with EHR Data

The Value Proposition of LLMs in Healthcare

Part 3: Challenges and Future of AI

Lightning Round - Personal Influences and AI in the Bay Area

Lightning Round - AI in Medicine and College Admissions

Concerns About the AI Ecosystem and National Control

The Path Forward for Open Source AI

Experimentation Platforms for Open Source AI

Building Useful Tools for Open Models

National Conversation About Language Models for Research

LLMs as an Interface to Patient Data

Hospitals and Patient Data

Potential Good and Not Good Uses of Models for Patients

Translational AI in Medicine: Unlocking AI’s Potential in Health Care with Nigam Shah

NEJM AI Grand Rounds

Part 1: Introduction and Background

00:03Introduction to AI in Healthcare and Nigam Shah's Background

Introduction to AI in Healthcare and Nigam Shah's Background

02:57Nigam Shah's Journey into AI and Biomedical Informatics

Nigam Shah's Journey into AI and Biomedical Informatics

05:36Postdoctoral Work and the Role of Ontologies in AI

Postdoctoral Work and the Role of Ontologies in AI

08:07Nigam Shah's Role as Chief Data Scientist at Stanford Health Care

Nigam Shah's Role as Chief Data Scientist at Stanford Health Care

11:03Collaboration with Clinicians and Problem-Solving Approaches

Collaboration with Clinicians and Problem-Solving Approaches

Part 2: AI Implementation in Healthcare

12:41Piloting AI in Clinical Use at Stanford HealthCare

Piloting AI in Clinical Use at Stanford HealthCare

15:19Pilot AI Studies and the Impact on Healthcare Delivery

Pilot AI Studies and the Impact on Healthcare Delivery

17:08Efficiency vs. Productivity in Healthcare AI

Efficiency vs. Productivity in Healthcare AI

20:32The Future of Primary Care and the Role of AI

The Future of Primary Care and the Role of AI

22:35Basics of How LLMs Work

Basics of How LLMs Work

25:15Applications and Limitations of LLMs in Medicine

Applications and Limitations of LLMs in Medicine

28:19Training LLMs with EHR Data

Training LLMs with EHR Data

31:30The Value Proposition of LLMs in Healthcare

The Value Proposition of LLMs in Healthcare

Part 3: Challenges and Future of AI

34:15Lightning Round - Personal Influences and AI in the Bay Area

Lightning Round - Personal Influences and AI in the Bay Area

37:02Lightning Round - AI in Medicine and College Admissions

Lightning Round - AI in Medicine and College Admissions

39:55Concerns About the AI Ecosystem and National Control

Concerns About the AI Ecosystem and National Control

41:29The Path Forward for Open Source AI

The Path Forward for Open Source AI

43:47Experimentation Platforms for Open Source AI

Experimentation Platforms for Open Source AI

46:03Building Useful Tools for Open Models

Building Useful Tools for Open Models

48:23National Conversation About Language Models for Research

National Conversation About Language Models for Research

50:27LLMs as an Interface to Patient Data

LLMs as an Interface to Patient Data

51:42Hospitals and Patient Data

Hospitals and Patient Data

53:03Potential Good and Not Good Uses of Models for Patients

Potential Good and Not Good Uses of Models for Patients