Rep. Matt Rosendale, R-Mont., urged the Department of Veterans Affairs on Thursday, Feb. 15, 2024, to obtain patient consent before using artificial intelligence as a clinical tool for the diagnosis and treatment of medical conditions. (U.S. Air Force)
WASHINGTON — Rep. Matt Rosendale urged the Department of Veterans Affairs on Thursday to obtain patient consent before using artificial intelligence as a clinical tool for the diagnosis and treatment of medical conditions.
Vast amounts of patient data is already culled from medical records and “fed into AI models” under development for the Veterans Health Administration, he said.
“You’re utilizing AI, and you’re not disclosing it to the veterans. You’re not giving them a choice. And that is dangerous. It truly is. It’s dangerous and it’s dishonest,” said Rosendale, R-Mont., chairman of the House Committee on Veterans’ Affairs subpanel on technology modernization.
He emphasized a process needs to be implemented so veterans learn how their health care information might be used in AI applications and they can give their prior approval.
“This is an important topic and one that the VA must get right,” said Rosendale, who spoke during a hearing on health that looked at AI applications in VA hospitals and clinics.
Rosendale asked VA officials at the hearing when they planned to put a notification and consent procedure in place for veterans who are patients.
Charles Worthington, the VA’s chief technology officer, said the issues are being addressed in an ethics group at the Veterans Health Administration that is seeking to understand the best approach to take.
The VA panel is likely to issue recommendations, though Worthington could not provide details Thursday on when its work would be complete.
“This is a new topic, and I don’t believe we have a specific timeline,” he said. “But we are very aware of this as a priority.”
But Rosendale did not seem satisfied.
“I will reiterate. This needs to be a high priority. You’re utilizing AI and the veterans need to be aware of that — and they need to give consent,” he said.
Worthington described the use of AI as a “generational shift” in how computers are used in health care settings. Data from veterans’ medical records is used in algorithms for new AI tools that fulfill a variety of health care uses.
AI models are deployed to scan medical images and other data to help doctors with early identification of health care risks and the detection of diseases.
Rep. Mariannette Miller-Meeks, R-Iowa, chairwoman of the House VA Committee’s subpanel on health, said while AI holds “great promise, the reality is that it is a new, developing technology. We are still figuring out what is possible and practical.”
She said a potential use for AI is in the diagnosis of sepsis in hospital patients. Sepsis, which is an infection that spreads throughout the body, is life-threatening without proper diagnosis and immediate treatment.
Worthington said there are about 100 examples of AI uses that are either under development or operational at VA health care settings.
“VA is incorporating AI technology into veterans’ health care to enhance diagnostic accuracy and efficiency, and to predict cancer risks and adverse outcomes,” he said.
Worthington said the VA is taking a measured approach in scaling AI-related solutions to ensure the applications are safe, secure and effective.
Health care applications can also include computerized scans of medical records for identifying veterans at risk of suicide and computerized tools to detect pre-cancerous growths in patients undergoing colonoscopies.
The VA uses software called REACH-VET to scan medical records and identify veterans who might be at risk of suicide.
“Since its inception in 2017, the initiative has successfully identified over 117,000 at-risk veterans,” Worthington said. “The clinical program has been associated with increased attendance at outpatient appointments.”
He said regular outpatient health care for veterans at risk of suicide often results in less need for more intensive hospital stays and emergency room visits.
But he also said the VA will need to provide technical information that explains to patients how the effective development of AI depends on the collection of their data.
“We’re going to need to do work to make that understandable to veteran patients — that they can understand how the VA is using their data in creating those AI models,” Worthington said.
