Artificial Intelligence in Healthcare: Drawing Inspiration From the Aviation Industry
As a recent 2022 report by the International Air Transport Association shows, the fatality risk in the aviation industry is very low, at 0.11. The highly-regulated aviation industry may hold the key to regulating artificial intelligence in health care, according to scientists at MIT. Assistant professor at MIT’s Department of Electrical Engineering and Computer Science (EECS) and Institute of Medical Engineering Sciences, Marzyeh Ghassemi, along with Julie Shah, Professor of Aeronautics and Astronautics at MIT, are interested in the transparency challenges in AI models.
Based on these interests, they have recruited a team of researchers across various institutions to conduct a research project. This project has been accepted to the Equity and Access in Algorithms, Mechanisms and Optimization Conference. One of the primary focuses of this research project is to create frameworks to manage the potential risks of AI deployments in health care settings.
Drawing parallels between the path of aviation and AI, researchers emphasize the importance of increasing transparency in AI models. They believe that the rigorous and comprehensive training required to become a commercial airline captain could serve as a training model for medical professionals using AI tools in clinical settings. Furthermore, they propose a reporting system for unsafe health AI tools, similar to the one the Federal Aviation Agency (FAA) uses for pilots.
Citing a 2023 report published by the World Health Organization, the researchers point out that one in every 10 patients is affected by medical errors in high-income countries. They emphasize that healthcare workers often fear reporting medical errors due to expected negative consequences. This new research project looks to address these issues and suggests the creation of an independent auditing authority, similar to the National Transportation Safety Board, to ensure the safety and efficiency of health AI systems.
The paper goes on to propose that government agencies such as the FDA, FTC, and more, should regulate the use of health AI to ensure equity. Despite existing regulations, there is a need to create a new independent auditing entity to assess the impact of technology, specifically AI. According to co-author Zach Harned, who recently graduated from Stanford Law School, a challenge of emerging technology is having technological development outpace regulation, but given the importance of AI in health care, there have been continued regulatory efforts, recent executive orders, and new regulations in the EU.
Following the logical order in the article, the three potential headings could be:
1. The Role of AI in Health Care
2. Parallels between Aviation and AI
3. Proposals to Regulate Health AI