CERSI Collaborators: Molly Jeffery, PhD (Mayo Clinic) (Site PI); Joseph Ross, MD, MHS (Yale) (Site PI); Barbara Barry, PhD (Mayo Clinic) (PI); Jennifer Ridgeway (Mayo Clinic) (Co-PI) Jennifer Miller (Yale) ( Co- PI)
FDA Collaborators: Anindita Saha, BSE (CDRH), Aubrey Shick, MS (CDRH), Lauren Rodriguez (CDRH), Matthew Diamond, M.D., Ph.D (CDRH), Nooshin Kiarashi, Ph.D., (CDRH)
Project Start: September 1, 2021
Regulatory Science Challenge
Artificial Intelligence (AI), broadly defined as the science and engineering of making intelligent machines, especially intelligent computer programs, is increasingly used in health care to improve prevention, diagnosis, treatment, and maintenance of health conditions (https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-software-medical-device). Trustworthy and transparent approaches to informing patients and providers about AI are needed to address the unique challenges of communicating how AI and Machine Learning (ML) enabled devices and software will be used in patient care. Patient decision-making, safety, and trust in AI/ML enabled medical devices hinge on successful communication of how AI tools are developed, their performance, and how they impact care.
Project Description and Goals
This study seeks to understand provider and patient perspectives on transparency and trust in AI/ML-enabled software for heart monitoring and diagnoses in cardiology. To achieve this objective, investigators will conduct focus groups of providers and patients to ask about their information needs, their views on information transparency, and what information would lead to earning trust in AI/ML-enabled medical devices in cardiology. Analysis of focus group data and a systematic literature review will reveal a set of core information needs for AI/ML-enabled cardiac software and medical devices labeling. These results will inform the design of conceptual prototypes for AI/ML-enabled medical device labels. Label prototypes will be refined through patient interviews and interactions with digital label prototypes. Refined label prototypes will used in a web survey with patients to validate information needs, assess how label information impacts trust, and evaluate comprehension. The outcome of this project will be a documented approach to provider and patient informed labeling of AI/ML-based software for cardiac monitoring and diagnostics to inform regulatory science.
Goal 1: Understand provider and patient information needs for transparency and trust in AI/ML-enabled Software
Goal 2: Create and iteratively refine software conceptual label prototypes for AI/ML-enabled software based on patient and provider information needs
Goal 3: Validate information needs, assess trust, and evaluate patient understanding of conceptual AI/ML label prototypes