July 13, 2024
AI Regulation for Health Care

New Approach Needed: Putting Patients First in AI Regulation for Health Care

In a recent study published in JAMA, researchers led by John W. Ayers from the Qualcomm Institute at the University of California San Diego have called for the prioritization of patient outcomes in the regulation of artificial intelligence (AI) in health care. The study highlights the absence of patient outcomes in the White House Executive Order that focuses on developing regulatory strategies for AI in health care.

The research team evaluated the Executive Order’s attention to equity, safety, privacy, and quality in AI health care solutions. However, they found it surprising that patient outcomes, the crucial metric for evaluating health care products, were not mentioned. The goal of medicine, according to Dr. Davey Smith from UC San Diego School of Medicine, is to save lives. Therefore, AI tools in health care need to demonstrate significant improvements in patient outcomes before being widely adopted.

The study emphasizes the consequences of neglecting patient outcomes in regulations, using AI-powered early warning systems for sepsis as an example. Sepsis is a life-threatening condition that affects millions of Americans each year. An evaluation of the most widely adopted AI sepsis prediction model revealed that it failed to identify 67% of patients who developed sepsis. The research team questions whether hospital administrators would have chosen this prediction system if patient outcomes had been thoroughly assessed.

Calling for a revision to the Executive Order, Dr. John W. Ayers emphasizes the need to prioritize patient outcomes when regulating AI products. Similar to pharmaceutical products, AI tools that impact patient care should be evaluated by federal agencies to ensure improvements in patient feeling, function, and survival. The researchers highlight a previous study on AI-powered chatbots responding to patient messages as an example of how patient outcome-centric regulations can be effective.

The study compared standard care to standard care enhanced by AI conversational agents and found differences in downstream care utilization among certain patient populations. However, the researchers note that such studies require appropriate incentives from regulators. By adopting a patient outcomes-centric approach, AI solutions for patient messaging and other clinical applications can truly enhance people’s lives.

While acknowledging that implementing patient outcome-centric regulations may be challenging for the AI and health care industry, the researchers believe that excluding these rules from the Executive Order is a significant oversight. They argue that prioritizing patient outcomes is essential for the safe and effective integration of AI into the health care system.

In conclusion, the study calls for a revision of current regulations to ensure that AI in health care is evaluated based on its impact on patient outcomes. By prioritizing patient outcomes, regulators can promote the development of AI tools that genuinely improve patient care, leading to enhanced lives and better health outcomes.

1. Source: Coherent Market Insights, Public sources, Desk research
2. We have leveraged AI tools to mine information and compile it