Congress Confronted With Pros and Cons of Health Care AI

WASHINGTON — A congressional committee took a cautious step Wednesday into the prospects for artificial intelligence to improve health care.

The benefits could be great but so could the risks to privacy, according to medical personnel who testified before a House Energy and Commerce subcommittee.

“It represents an incredible opportunity for our country,” said Rep. Anna Eshoo, D-Calif.

Artificial intelligence refers to computerized machine learning that could take over many of the tasks now performed by human doctors and nurses. It offers the potential for enhanced patient care, streamlined bureaucratic procedures and better diagnostic accuracy.

It could track many patient health characteristics often overlooked now, including caloric intake, sleep patterns and brain functions.

It also could trample the privacy of patients if the information gets into the wrong hands.

“It’s important to remember that every decision comes with a cost,” said Rep. Brett Guthrie, R-Ky., chairman of the subcommittee on Health.

Insurance companies, for example, could use private health information to deny coverage for patients who need it.

The privacy challenge explains one of the principles that guides legislation lawmakers are developing to control AI in the health care industry, according to lawmakers.

In other words, “How are we ensuring this data is secure,” Guthrie said.

Without good regulation and guidance on how to use artificial intelligence in health care, “risks will dominate,” said David Newman-Toker, a Johns Hopkins University School of Medicine neurology professor.

In addition to privacy concerns, inaccurate algorithms and data that control AI could lead to misdiagnosis, Newman-Toker said.

The Johns Hopkins Armstrong Institute Center for Diagnostic Excellence reported in July that about 371,000 patients die and 424,000 are permanently disabled each year because they are incorrectly diagnosed.

Artificial intelligence could dramatically decrease the death and disability rate from misdiagnosis but only if it is used appropriately, Newman-Toker said.

“We should make sure we’re not relying on faulty data sources,” he said.

Medical experts at the hearing warned that the technology so far has raced ahead of the government’s efforts to regulate it effectively.

In the past five years, AI advances have been “akin to going from locomotives to powered flight,” said Benjamin Nguyen, a product manager for the consumer digital health company Transcarent.

According to the U.S. Food and Drug Administration more than 100 drug and biologic applications submitted in 2021 included artificial intelligence and machine learning components.

Last month, the FDA added 171 medical devices to its list of artificial intelligence and machine learning-enabled medical devices.

You can reach us at [email protected] and follow us on Facebook and Twitter