U.S. senators are seeking to heighten scrutiny around artificial intelligence’s use in healthcare, holding a hearing on the topic Thursday.
Several experts testified at the Senate Committee on Finance inquiry, Artificial Intelligence and Healthcare: Promise and Pitfalls, including two physicians and Siemens Healthineers North America’s head of digital health. Members of Congress are concerned that AI may be perpetuating bias in medicine, amid reports that health insurers are using algorithms to deny care.
“I think most of us here would agree there are many ways this technology can be used to improve healthcare and patient outcomes,” Sen. Ron Wyden, D-Ore., chair of the Finance Committee, said during remarks given Feb. 8. “However, as we increasingly rely on technology like AI to make decisions in every facet of our day-to-day lives, this committee has a responsibility to ensure there are guardrails in place to protect patients, particularly in Medicare and Medicaid, and I do not believe that current laws go far enough to achieve that goal.”
In September, Wyden and other Democrats in the House and Senate reintroduced the Algorithmic Accountability Act. The bill would require companies to conduct impact assessments for effectiveness, bias and other factors when using AI to make critical decisions. It also would create the first public repository of these systems, housed within the Federal Trade Commission, and add 75 FTC staffers to enforce the law.
During his remarks, Wyden pointed to 2019 research authored by Ziad Obermeyer, MD, with the University of California, Berkley. It unearthed racial bias in an AI tool developed by Optum and used by providers across the U.S. to offer care management services. On average, black patients needed to present with worse symptoms than their white counterparts to qualify for the same level of care, the study found.
“Despite my great optimism, I worry that without concerted effort from researchers, the private sector and government, AI may be on a path to do more harm than good in healthcare,” Obermeyer, a practicing emergency physician, said in written testimony submitted to senators.
Meanwhile Finance Committee Ranking Member Mike Crapo, R-Idaho, touted the need for “responsible and ethical” AI deployment, with appropriate safeguards for privacy and security. In cases such as algorithms, applied by insurers to expedite utilization management, “improper denials or delays of needed services warrant government scrutiny.” But in other instances, including AI-related challenges stemming from insufficient provider experience and education, a lighter legislative touch is needed.
“Along these lines, AI highlights the need for adaptability,” Crapo said in his remarks. “One-size-fits-all, overly rigid, and unduly bureaucratic laws and regulations risk stifling life-saving advances and becoming outdated before they are even codified.”
Peter Shen, head of digital health, North America, at Siemens Healthineers also testified during the hearing. He highlighted work that is already underway to improve transparency around imaging AI. Shen cited the American College of Radiology’s new program, which asks AI manufacturers to disclose details about how they created algorithms. Its aim is to aid radiology practices in selecting products tailored to their needs.
Siemens Healthineers was among the first eight vendors to sign up for the initiative, Shen noted. He said the company has sought to build algorithms that are accurate and unbiased, using training data from individuals of different ages, genders, ethnicities and healthcare profiles.
Shen believes concerns around AI bias are currently addressed under existing risk management processes, quality systems, and compliance with requirements from the FDA and other regulators.
“With the rapid acceleration in development and innovation of AI, the need for the regulatory environment to be able to balance safety, effectiveness, as well as update and improve functionality, without hampering innovation and adoption, is critical,” Shen said in his testimony. “While we believe the current regulatory framework is sufficient to support AI innovation, we support the continuation of flexibility in the approval process, as a one-size-fits-all approach could seriously inhibit the potential of AI, as well as efforts to facilitate global harmonization and the development of appropriate international consensus standards.”
You can watch a recording of the hearing, and read testimony from other witnesses, here.