Experts: Regulate Healthcare Artificial Intelligence Based on Patient Outcomes

A proposed regulatory strategy for artificial intelligence in healthcare settings is compared to regulatory standards for new drugs.

A trio of experts is proposing companies that produce artificial intelligence (AI) tools in healthcare should be regulated based on their ability to achieve benefits in patient outcomes.

Last year, President Joe Biden issued an executive order to set guidelines for AI policy. The executive order established the U.S. Health and Human Services Department as the lead agency in setting AI policy, including a directive for the agency to set up a task force to create a plan for responsible AI use.

In a recent Viewpoint article published by JAMA, three experts from the University of California San Diego argue that AI in healthcare should be regulated based on the ability of AI tools to generate positive changes in patient outcomes.

"We believe AI regulatory assessments should be grounded in clinical evidence regarding how patients feel, function, or survive in rigorously designed studies, such as randomized clinical trials, which is consistent with regulatory standards applied to new drugs that also require a net clinically meaningful improvement in patient outcomes compared with a placebo," the experts wrote.

Healthcare regulators already have the ability to draft rules based on clinical outcomes, the experts wrote.

"For instance, electronic health records require federal certification under the Health Information Technology for Economic and Clinical Health Act," the experts wrote. "Rule makers can use this avenue to require that any AI tools seeking to integrate or embed within an electronic health record be evaluated with clinical end points. With the clear goal of accessing the healthcare marketplace under regulatory approval, many companies and health centers may be given appropriate incentive to register trials that evaluate the potential of AI tools to improve patient outcomes. This regulatory approach ensures that AI technology is deployed primarily for the benefit of patients. Otherwise, industry may focus on revenue-generating opportunities and not patient outcomes."

AI tools improving patient outcomes

The top clinical leaders at Allegheny Health Network (AHN) and Providence say their health systems are using AI tools that have a positive impact on patient outcomes.

The primary area where AHN is using AI tools to impact patient outcomes is AI technology linked to the health system's electronic medical record, Epic, says Donald Whiting, MD, chief medical officer of AHN and president of Allegheny Clinic.

"We are using AI to look for sepsis risk, look for ICU readmission risk, and predict length of stay," he says. "So, we are using AI to scour the medical record to look for predictors of things where we can intervene."

AHN is also using an AI tool that uses 3D imaging to scan for skin abnormalities, Whiting says.

"This improves skin cancer screenings, and we are using this technology in primary care offices so they can screen patients instead of them having to go to a dermatology office," he says. "The AI compares the images to a set of images in a database for cancer diagnoses and creates a score to quantify risk to determine whether the patient needs to go to a dermatologist or skin cancer expert for further evaluation."

Hoda Asmar, MD, MBA, executive vice president and system chief clinical officer at Providence, says the health system has been making strides to improve patient care and boost efficiency for clinicians. "There is incredible potential for AI to transform the healthcare space, and I'm encouraged by the positive impact we have seen so far at Providence," she says.

One AI tool being utilized at Providence that is improving clinical outcomes for patients is Trial Connect, which is using AI to match patients with more than 2,500 ongoing research trials.

Clinical leader perspective on healthcare AI regulation

Asmar says it is a good idea to regulate healthcare AI tools based on their ability to drive positive patient outcomes.

"It is going to be critical to have a balanced approach toward AI in healthcare where ethical and governance standards are established not to stifle progress but to make sure we are prioritizing the positive impact on patient outcomes," she says. "Providence is committed to developing and deploying AI in ways that best serve and support our clinicians, positively impact patient outcomes, and provide the means to help us give more access to our communities, especially the poor and vulnerable."

Regulating healthcare AI tools is a necessary step in the development of this technology, Whiting says. "There is a fear of anything new, especially something like AI that has potentially unlimited capabilities. Finding the right utilization of AI intelligently and establishing the right guardrails is going to be important."