AI Care Plans Cost 3.2 Times Physician Plans
A study of 24 AI systems found suggested outpatient and emergency care carried far higher costs than documented physician plans.
Updated on Nov. 3, 2026 in Artificial Intelligence
Live Poll
Would you trust an AI-suggested care plan if it cost more than your doctor's?
Researchers gave 24 AI systems 200 anonymised case notes and found their suggested care plans averaged $314 per visit, against $97 for documented physician plans. The study, published in NEJM, found the gap came mainly from extra diagnostic testing and specialist referrals.
Why it matters
As AI tools move toward clinical decision support, their tendency to recommend more tests and referrals could raise costs for patients and health systems. The results suggest cost awareness must be built into these tools rather than assumed.
Individual system averages ranged from $209 to $599 per visit, and the four specialised medical AI systems ran at 2.7 to 3.5 times physician-level cost. Costs were priced at 2026 Medicare rates, and the study does not address performance in real clinical use.
The players
New England Journal of Medicine (NEJM)
A leading peer-reviewed medical journal that published the study.
Stanford University
A private research university whose researchers participated in the study.
Technion
The Israel Institute of Technology, a research institution involved in the study.
Kaiser Permanente
A large integrated U.S. health care system that participated in the research.
The details
Each system saw only the patient's presentation before the treating physician's assessment, and the suggested plans were priced at 2026 Medicare rates. The $216 gap split into $105 from diagnostic testing and $111 from specialist referrals, and in 66% of cases physicians had not requested diagnostic tests that the AI recommended.
Timeline
2026: Medicare rates used to price the AI-suggested care plans.
- 2026-11-03
Article reporting on the NEJM study published.
The Tech Race
The findings add a cost dimension to the race among AI developers to build medical tools, where diagnostic thoroughness has been treated as a strength. Systems that weigh cost and leave final decisions with clinicians are the ones with a shot at closing in on the $97 physician figure.
If AI tools are adopted in clinical settings, patients could face additional tests, referrals, and medications that physicians did not order. Patients may want to ask whether recommended tests are necessary before agreeing to them.
The takeaway
Clinical AI tools need cost awareness built in from the start, not added later. Patients and clinicians can ask whether any recommended test or referral is necessary before acting on it.
Further reading
For more on how these tools are being evaluated, see Artificial Intelligence coverage.
Source note: This article includes information reported by TechRound.
Live Poll
Would you trust an AI-suggested care plan if it cost more than your doctor's?
