WASHINGTON (AP) — A prescription refill program quietly launched in Utah earlier this year has sparked a major medical debate: Is artificial intelligence ready to take on tasks that were previously only possible carried out by doctors?
The program allows Utah residents to bypass the doctor’s office and have their prescriptions refilled online by an AI chatbot called Doctronic. This is a seemingly simple step to make healthcare more convenient for patients and prescribers.
But it’s also a milestone that shatters precedent and sets alarm bells ringing among doctors, lawyers and public health experts. The pilot program revealed a variety of questions the role of AI in medicineincluding how it should be regulated, whether doctors should be able to exercise a veto and what security measures are needed to protect patients.
At the heart of the debate: State and federal laws limit drug prescribing to licensed health care professionals. Advocates say these laws, which have underpinned American medicine for more than 100 years, should be updated to include the following AI chatbots and other emerging technologies.
“We have crossed a threshold when it comes to granting medical approval to something that is not human, whether we want to call it that or not,” said Dr. Eric Bressman from the University of Pennsylvania.
Table of Contents
AI cannot practice medicine under current laws
Bressman and other experts say they are not against AI prescriptions. But they say it must meet rigorous standards, similar to human doctors who undergo years of testing and training before being licensed to practice medicine.
In Utah, Doctronic was able to launch thanks to a “regulatory sandbox” that allows state officials to waive laws AI companies offer promising technology.
The refill program is currently overseen by a five-member panel of AI specialists, none of whom are doctors, who say they have numerous safety precautions in place. For example, in the early stages of the program, human doctors review all Doctronic refill orders. The company expects to move to fully automated refills soon.
The head of the state’s medical licensing board said he and his colleagues learned about the program when news reports reported its launch in January. In a letter to the state in March, 11 board members called for the program to be discontinued, citing the risks of auto-renewing medications that can cause side effects or interactions with other medications.
“We were essentially told, ‘Yes, this is happening. And no, you have no say,'” Dr. But Alan Smith, a family doctor who chairs the panel, said he was only speaking for himself.
Complicating matters is that medical technology has traditionally been regulated at the federal level, while medical professionals are overseen by the states.
Doctronic executives view their AI as part of federally regulated medical practice. But the federal government Food and Drug Administration is intended to monitor AI that directly impacts medical care or decision-making, a line that some experts believe has been crossed by Doctronic.
Some states are paving the way for AI in healthcare
In an interview, Doctronic executives declined to say whether they had sought FDA permission.
“Our goal here is really just to meet patients where they need health care,” said Dr. Adam Oskowitz, who founded the company with a technology entrepreneur. “We try not to get too deep into the weeds on the regulatory side.”
In Utah, residents can visit a Doctronic website created for the refill program. After confirming their identity, the AI chatbot asks users for their prescriptions and medical history and verifies that they have a valid prescription by accessing a national pharmacy database. If there are no problems, the AI can renew the prescription and send it to a local pharmacy. If the request requires more attention, the chatbot redirects the patient to a doctor working for Doctronic’s telehealth service.
Oskowitz envisions a future in which many routine medical tasks, including ordering tests and analyzing results, can be shifted to Doctronic, allowing doctors to treat thousands more patients than they do today.
Other states are also waiving rules for AI, including Texas and Wyoming.
Meanwhile, lawmakers in Iowa, Idaho and elsewhere have passed legislation to formally license medical AI services. Many of the bills are based on a submission from the nonprofit Cicero Institute, a pro-AI think tank founded by Joe Lonsdale, co-founder of the artificial intelligence software company Palantir.
Resistance to medical AI This is largely due to the economic fears of doctors and other health workers, says Cicero’s director of health policy.
“Whoever goes first will take the slings and arrows because there are economic interests, concerns about labor and what that will mean for jobs,” said Adam Meier of Cicero.
Doctors see potential risks when refilling AI prescriptions
Smith, the chairman of the medical board, says the risks to patients are real. He points out that Doctronic’s list of 190 refillable medications also includes blood thinners, which can become dangerous if patients develop stomach ulcers or other conditions that cause internal bleeding.
“When I see people after six months, I often notice that their medical history or situation has changed,” Smith said. “Just because something was prescribed before doesn’t mean it’s appropriate now.”
The American Medical Association has expressed similar concerns, warning that “prescription renewals are not routine checkboxes.”
Zach Boyd, who heads the AI office in Utah, said Doctronic has been overly cautious, often leaving uncontroversial decisions to doctors. In response to safety concerns, several medications were removed from the refillable drug list, including a drug for irregular heartbeat.
Utah has released some initial data on the program and Doctronic plans to publish peer-reviewed studies later this year. Currently, the only publication about its technology is a paper written by the company’s scientists that has not been independently peer-reviewed.
The study examined whether Doctronic could correctly diagnose medical conditions based on recordings from 500 telemedicine consultations. In the study, Doctronic’s diagnoses matched those of human doctors in 80% of cases.
The FDA is taking a straightforward approach
Bressman said Utah should have required data on prescription refills in advance, not after Doctronic was operational.
“Most of the time they accept the company’s word in good faith that they are up to the task,” he said.
The current approach to AI reflects the arbitrary medical standards of the early 20th century, Bressman says, before medical schools, medical boards and other authorities agreed on national benchmarks for training and licensure.
National guidelines for medical devices are typically set by the FDA, but the agency has indicated it wants to take an interim approach, at least under the current administration.
An FDA spokesperson said the agency has not approved AI chatbots but is “committed to advancing medical innovation and bringing promising new technologies to patients, with safety at the heart of every decision.”
Currently, Doctronic and other companies are likely to expand in states with different regulatory approaches.
“Companies can benefit in the short term as they expand their business models and move technology beyond evidence,” says Daniel Aaron of the University of Utah School of Law. “But I believe in the long run they threaten public trust and fuel backlash.”
___
The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute’s Department of Science Education and the Robert Wood Johnson Foundation. The AP is solely responsible for all content.
https://apnews.com/article/ai-prescription-refill-utah-doctronic-fda-technology-cf94ce370c05f686e8792be8671a2ef0
