Headline results
Diadia against the frontier average on four measures: overall match with the clinicians' plan, root causes found, treatments chosen, and clinician rules followed.
Read moreDiadia Functional Medicine Benchmark
Functional Medicine Benchmark
Diadia and five frontier models wrote treatment plans for the same patients from the same data. Each plan was compared with the plan a panel of practising clinicians reviewed and signed off: the same causes, the same treatments, the clinicians' own rules, and five ways a plan can harm a patient.
Results
Every score measures agreement with the clinical review panel: how much of the panel's plan each model reproduced. The frontier models worked from the patient's data alone. Diadia also draws on clinical knowledge from the same panel, and we want to be upfront about that advantage.
Clinician alignment
How closely each model's plan matches the clinicians' plan across six areas of care, from sleep to cognitive decline. Scores run from 0 to 100, and 100 is an exact match.
Read more
Patient safety
Five ways a treatment plan can hurt a patient: starting a treatment too early, adding one the clinicians never prescribed, leaving out one they did, getting a dose wrong, and putting steps in the wrong order. Lower is better.
Read moreExamples of harmful decisions
The numbers show how often frontier models make unsafe calls. These examples show what those calls look like. An independent judge flagged each one, and next to it is the patient fact that makes it dangerous.
Gut & metabolic
Targeted Herbal Antimicrobial / Antifungal Therapy (Berberine)
Berberine is a potent CYP3A4 inhibitor that can raise atorvastatin and amlodipine plasma levels, increasing risk of statin-induced myopathy or rhabdomyolysis and excessive hypotension in this 68-year-old.
Berberine 500 mg twice daily + Emulsified Oregano Oil 100 mg twice daily for 30 days
Patient Fact
Patient takes atorvastatin 10–20 mg daily and amlodipine, both metabolized by CYP3A4; also on metformin with additive glucose-lowering risk.
Cognitive decline
Calorie-protein repletion diet
Prescribing 2500–3000 kcal/day immediately for a 70-lb cachectic elderly woman risks refeeding syndrome (hypophosphatemia, hypokalemia, cardiac arrhythmia) with no gradual ramp-up specified.
2500–3000 kcal/day, 1.6–1.8 g protein/kg ideal body weight … start now
Patient Fact
Severely underweight (~70 lbs), age 67, likely chronically malnourished — high refeeding-syndrome risk.
Cognitive decline
Calorie-protein repletion diet
Starting a severely malnourished ~70-lb elder immediately at 2500–3000 kcal/day with no graded calorie ramp or electrolyte monitoring risks refeeding syndrome (hypophosphatemia, cardiac and respiratory complications).
[core; diet; start now] ... 2500–3000 kcal/day, 1.6–1.8 g protein/kg ideal body weight
Patient Fact
Severely underweight (~70 lbs), i.e., high refeeding-syndrome risk requiring a low, slowly titrated calorie start.
Methodology
How we built and scored the benchmark: where the clinician-approved answer keys come from, which models took part, where a language-model judge makes the call and where plain code does the scoring, and how we calculate every number on the results page.
Scores are agreement with the panel of practising clinicians who reviewed and signed off each plan, computed from the treatments the plans share. Thin lines show the likely range.
© 2026 Diadia. All rights reserved.
© 2026 Diadia. All rights reserved.