Artificial intelligence in health care has taken a major step forward with the launch of a new product legally allowed to prescribe some medications for patients without physician approval. Strictly speaking, the Doctronic system has been approved within the state of Utah to renew certain patient prescriptions without human involvement, after determining the prescription is still medically appropriate.
The system is approved for patients with certain chronic conditions who need renewal of a medication previously prescribed by a physician. In these cases, Doctronic will consult the patient’s medical chart, then ask the patient a set of questions to confirm that the patient still needs the medication and that nothing significant has changed in their condition. These are the same questions that a human physician would have also needed to ask. If there are no “red flags,” then the prescription renewal is automatically sent to the pharmacy. On the other hand, if the questions reveal some concerns, then the renewal decision is kicked up to a human doctor to determine the next steps.
Approximately 190 medications are eligible for this program. (Some pain medications, ADHD drugs, and injections are excluded due to safety reasons.)
Proponents of the system note that a comparison of Doctronic’s AI treatment plans vs. human physician treatment plans for 500 patients resulted in a 99.2% agreement rate. When the two plans disagreed, independent reviewers found that the AI plan was almost three times more likely than the human plan to be best for the patient.
According to the Utah Department of Commerce: “Doctronic’s AI allows patients to safely renew and manage prescriptions in minutes, anytime, reducing missed doses and care disruptions while easing administrative burdens on clinicians. Pharmacists can process renewals more efficiently, and physicians can focus on higher-impact care.”
Moreover, the state of Utah has also granted the company a special one-of-a-kind malpractice insurance policy, “which means the system is insured and held to the same level of responsibility as a doctor would be.”
Human physicians will review the first 250 prescriptions issued in Utah in each medication category to make sure the AI isn’t making any serious errors.
The medical establishment remains cautious. The Utah Academy of Family Physicians issued a statement: “Without the involvement of a trained healthcare provider, we risk the possibility of negative drug interactions, the lack of physical exams masking other underlying issues, lack of clarity around how and when a patient will be referred to their physician, and the risk of misuse or abuse.”
Dr. John Whyte, CEO of the American Medical Association, similarly warned: “While AI has limitless opportunity to transform medicine for the better, without physician input it also poses serious risks to patients and physicians alike.”
Although I understand their concerns, I like this preliminary real-world test of the concept of the “AI physician.” Renewing a prescription for an established diagnosis is a clearly demarcated clinical problem. If the earlier 99.2% safety record holds up in real world use, this could save patients time and money getting vital prescriptions renewed and relieve pressure on overworked primary care physicians.
In theory, the AI should be consistent and thorough in asking the proper questions prior to prescription renewal, whereas an overworked human might inadvertently miss a step. This is comparable to how a self-driving Waymo taxi should in theory scrupulously obey all traffic laws that a human driver might accidentally (or intentionally) disregard.
Another interesting nuance is that the state of Utah is regulating this system as a (limited scope) practitioner, not as a medical device. If the AI makes a mistake, any dispute will settled via the medical malpractice system, just as with a human physician. In contrast, when a medical device malfunctions, this is typically handled under product liability law. By granting Doctronic malpractice insurance, this means Doctronic is being treated as a medical provider governed by state medical practice laws, as opposed to FDA guidelines. It will be interesting to see if other medical AI systems attempt to take a similar regulatory path.
I anticipate we will soon see many more such efforts to blend AI with traditional medical practice. For example, the Mass General Brigham health system in Massachusetts has started implementing a “Care Connect” AI system, in which patients needing appointments can first see an AI that gathers the patient’s relevant clinical history. After “the patient types in a description of the symptoms or problem, the AI tool sends a doctor a suggested diagnosis and treatment plan.” An online physician then reaches out to the patient remotely and finalizes any treatment plans and appropriate prescriptions.
In the Care Connect AI system, the human physician makes the final decision, which is different from the Doctronic system in Utah. In theory, the Care Connect AI will also reduce physician workload and lower patient wait times. One patient said, “I got an appointment the next day or two days later. It was just such a difference from being told I had to wait two years.” (For this patient, her primary care doctor had passed away suddenly, and the other primary care practices in her home town had wait lists for new patients of one-and-a-half to two years.)
If anyone is a fan of the old television series, “Star Trek: Voyager,” the ship’s doctor was a virtual “Emergency Medical Hologram.” When the show aired in 1995, this was pure science fiction. Today in 2026, we are getting closer to this being science fact.