Can We Trust AI Output? A Trustworthy AI Perspective for HEOR and RWE
Author(s)
Xiaoyan Wang, PhD, IMO Health, Rosemont, IL, USA, Mitchell Higashi, PhD, Associate Chief Science Officer, ISPOR, Lawrenceville, NJ, USA, Rachael Fleurence, PhD, NIH, Washington, DC, USA, Hua Xu, PhD, Yale University School of Medicine, New Haven, CT, USA and Ganhui Lan, PhD, Pfizer, New York, NY, USA
PURPOSE:
This session aims to explore the critical issue of trust and delve into the core tenets of trustworthy AI for HEOR in the era of generative AI. It will also shed light on the evolving status, significance, and challenges associated with validating AI-generated outputs, particularly in HEOR applications from the perspectives of technology, academic health systems, and the biopharmaceutical industries.DESCRIPTION:
The integration of AI into healthcare and life sciences has surged, especially with the advent of generative AI technologies. Establishing a framework for trustworthy AI within HEOR is crucial, requiring a comprehensive exploration of its applications and implications. The session, moderated by Dr. Higashi from ISPOR, will begin with the significance of building trust in AI models for HEOR. Dr. Xu from Yale will provide the latest advancements in generative AI and LLMs, presenting his research on trustworthy AI in addressing hallucination of LLMs and literature discovery. Dr. Wang from Tulane/IMO will present her research on RWE and clinical trials, focusing on the evaluation of key validation metrics for AI tools (12’). Dr. Lan from Pfizer will discuss leveraging AI and automation to enable continuous generation and monitoring of RWE and value within pharmaceutical settings. Dr. Fleurence from NIH will examine the pillars of trustworthy AI, drawing upon the frameworks proposed by entities including the National Academy of Medicine and the Coalition for Health AI to inform the trustworthy AI framework in HEOR. The session will include a comprehensive review of the requisites for building trustworthy AI frameworks for various real-world use cases in HEOR and RWE, emphasizing their reliability in conducting robust investigations and providing confirmatory evidence crucial for payer and regulatory decision-making. Attendees are encouraged to actively engage with the panelists through digital polling and a Q&A segment, fostering a dynamic exchange of perspectives and insights.Conference/Value in Health Info
2024-11, ISPOR Europe 2024, Barcelona, Spain
Topic
Methodological & Statistical Research