AI Care Plans Cost 3.2 Times Physician Plans
A study in NEJM finds the gap is driven by extra testing and referrals, and cost-aware prompts can narrow it, though this remains research-stage.
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?
In a study published in NEJM, researchers gave 24 AI systems 200 anonymised outpatient and emergency room case notes and found their suggested care plans averaged $314 per visit, against $97 for documented physician plans. This is a research-stage finding from a published study, not a product recommendation or deployed clinical practice.
Why it matters
The gap points to a design tension in medical AI: systems built to eliminate diagnostic blind spots tend to recommend broader testing and referrals, which raises costs. Cost-aware prompting narrowed the gap, suggesting that how these systems are instructed shapes their spending.
Costs were priced at 2026 Medicare rates and covered diagnostic tests and specialist referrals, so the figures reflect a specific billing basis. The $216 gap split into $105 from diagnostic testing and $111 from specialist referrals, and individual system averages ranged from $209 to $599.
The players
New England Journal of Medicine (NEJM)
A leading peer-reviewed medical journal that published the study comparing AI-suggested and physician-documented care plans.
Technion
Israel's technology research university, one of the research institutions involved in the study.
Stanford
Stanford University, a research institution involved in the study with strong programs in clinical AI and machine learning.
Tel Aviv University
Israeli research university, one of the institutions involved in the study.
Kaiser Permanente
Large U.S. integrated healthcare system and research institution involved in the study.
The details
Each AI system saw only the patient's presentation before the treating physician's assessment, then proposed a care plan that was compared against what the physician actually documented. The systems were programmed to eliminate diagnostic blind spots, and that objective led them to recommend wider testing: in 66% of cases, physicians had not requested diagnostic tests the AI suggested. The four specialised medical AI systems ran at 2.7 to 3.5 times physician-level cost.
Timeline
2026: Medicare rates used to price the AI-suggested care plans
November 3, 2026: Article published reporting on the NEJM study
The Tech Race
The study tests 24 systems, including four specialised medical AI tools, against the same documented physician baseline, which allows a direct cost comparison across the field. The specialised systems ran at 2.7 to 3.5 times physician cost, placing cost-efficiency alongside accuracy as a measure developers will need to address.
Patients and clinicians are the groups most directly affected, since the study suggests AI tools that are not cost-aware could raise the cost of routine visits. Developers and health systems evaluating these tools for clinical use should check how cost is weighted in the prompts they deploy.
The takeaway
Watch whether the cost-reduced prompts hold up in follow-up studies that also measure diagnostic accuracy, since cheaper care plans only matter if they remain clinically sound. The next concrete check is any published replication or outcomes study tied to the NEJM findings.
Further reading
For more on how these systems are being built and tested, see the Artificial Intelligence section.
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?
