Artificial Intelligence & Machine Learning
,
Healthcare
,
Industry Specific
Medical Board Urged State to Ditch Pilot, Citing Patient Safety Concerns

The state of Utah is forging ahead with a pilot program that allows patients with chronic conditions to autonomously refill medication prescriptions using an artificial intelligence-powered telehealth platform despite opposition from its state medical licensing board.
See Also: Reduce Cloud Risk in Healthcare with Security by Default
The state in January launched the pilot, “a first of its kind” in the U.S.,” with Doctronic, a New York-based telehealth company that touts itself as being an “AI doctor.”
Under the Utah pilot, the Doctronic AI platform can legally refill 190 common medications for chronic health conditions.
State officials said the effort with Doctronic aims to demonstrate that “safe, well-regulated AI” can improve patients’ medication adherence, prevent hospital visits and reduce healthcare spending with keeping “clinicians at the center of care.”
The state’s medical professional licensing board argued otherwise, urging late last month the pilot’s end in a letter citing patient safety concerns.
“While we support the legislative mandate to explore AI implementation, we also have a stewardship to protect Utah citizens,” the reads.
“Collectively, the board has decades of medical experience across a variety of specialties, positioning us to understand the potential consequences of implementing what may seem like an innocuous task of AI-driven prescription refills.”
Patient medication refills require professional reassessment so that clinicians can decide whether an adjustment to dose or drug choice is needed, and to monitor for side effects and potential new adverse interactions with other drugs that individual is taking, the board said.
Furthermore, “patients who continue refilling medications without assessment may remain on outdated or suboptimal therapy for months or years. There is a reason prescription refills require physician authorization,” the board argued.
Although the licensing panel urged the pilot be ditched, Utah’s Office of Artificial Intelligence Policy responded that the project is addressing the board’s concerns.
That includes requiring “100% human physician review” in phase one of the pilot. That means ensuring that every AI-generated prescription renewal is reviewed and approved by a licensed human physician before transmission to a pharmacy, the office said.
Through all phases of the program, the AI platform will be required to conduct comprehensive clinical assessments “that mirrors human clinical decision-making, including screening for new side effects, analyzing drug interactions,” Utah officials said.
The AI system is also prohibited from handling controlled substances, modifying treatment plans, or initiating new prescriptions, the state said. “It only processes 30, 60, or 90-day renewals for existing prescriptions being used to treat specific chronic conditions. Furthermore, patients must be evaluated by a physician in-person or via telehealth at regular intervals.”
Still, Doctronic on its website said that unlike most other telehealth firms that typically start with a patient first communicating with an actual doctor, Doctronic patients start – and continue – their encounter by chatting with AI, only connecting the individual with a licensed physician “if needed.”
Physicians signing their names to the medical licensing board’s letter to the state did not immediately respond to ISMG’s requests for comment on Utah’s decision to forge ahead with the project.
Doctronic also did not immediately respond to multiple ISMG requests for comment and additional details about the Utah pilot and how its AI platform works.
Attorney Lee Kim, founder of consulting firm Keytera, and former longtime cybersecurity and privacy principal at the Healthcare Information and Management Systems Society, said Utah appears to be the first state that has rolled out AI-enhanced efforts allowing prescription renewals. It likely won’t be the last.
“If a doctor’s office is closed, too busy, or if a certain doctor or other clinician is unavailable, an AI solution can help facilitate the prescription renewal,” she said. “Communication, speed and efficiency can be optimized. But we also need to ensure there are appropriate guardrails for safety purposes,” she said.
AI is an important evolution in medicine, and the industry needs to move with it, Kim said. “But if these systems are not properly designed, tested and implemented, adverse results can happen very quickly,” she said.
Robust identity authentication is also essential, she said. That includes controls that are being used to authenticate the patient, caregiver, or other authorized person requesting the refill. Also, as important are the controls being used to authenticate the individual or entity authorizing the refill, and whether that authorization is automated in whole or in part.
The Utah Office of Artificial Intelligence Policy was launched in May 2024 with authorization “to create regulatory mitigation agreements to encourage beneficial and safe adoption of AI in regulated industries.” That includes healthcare.
Besides the drug prescription pilot, the office has other healthcare related efforts underway.
That includes publishing “best practices” involving the use of AI for mental health therapy.
“This is probably the most extensive authoritative guidance in the U.S. right now on mental health and AI, and the community has appreciated the balanced, practical approach,” Zach Boyd, director of the Utah Office of AI policy told ISMG.
“So far, the feedback from the community has been very positive. No changes are immediately expected,” with that guidance, he said. “The biggest gap that would ideally be filled in the future is probably more specific liability guidance for the practitioners using AI,” he said.
The office is spearheading a project to investigate possible regulatory reforms for healthcare in AI in the state, Boyd said.
“We are currently in the early stages of preparing recommendations, but initial stakeholder feedback suggests possible needs for legislation around issues such as consumer-facing healthcare information, liability, reimbursability, professional conduct rules and autonomous healthcare services.”
