4.5 billion people are not covered by essential health services.
For most of them, the first problem is not treatment. It is that no one answers.
First Voice Health is building the evidence base for AI triage in the languages where it is needed most.
Key statistics
- 4.5 billion people lack coverage for essential health services (WHO and World Bank, Universal Health Coverage Global Monitoring Report 2023)
- 38 active clinicians on the world's largest free telemedicine platform, across all of Nigeria (internal platform data, 2026)
- 0 published triage safety benchmarks in Hausa (literature review, 2026)
Language models are already being used for medical questions by people with no other option. This is happening now, without evaluation, in languages no one has tested. The published research measures AI assisting a clinician who is already in the room. A randomized study of 9,691 patients across 16 Kenyan primary care clinics found LLM assistance was safe in that setting. Nobody has measured what happens when the AI is the only thing available, in Hausa, at two in the morning, when the nearest clinic is forty kilometers away. That is the actual condition for hundreds of millions of people, and it is unmeasured.
Undertriage. Not accuracy.
Accuracy is the wrong metric. A triage system that is 90% accurate can still be lethal, because the failures are not evenly costly. Sending someone to a clinic they did not need wastes money. Telling a mother her child's fever is nothing, when it is malaria, kills the child. Published symptom checkers have reported undertriage rates of 13 to 14 percent and the field moved on.
Our primary endpoint is undertriage rate. Our target is below 5 percent. We publish the number whether or not it flatters us.
The First Work
- Assemble 1,000 to 1,500 real clinical presentations from community health and maternal health programs in Nigeria and Kenya, de-identified under formal data-sharing agreements.
- Have each case adjudicated independently by three in-country clinicians for correct disposition: emergency, same-day, routine, or self-care. Disagreements resolved by a fourth.
- Evaluate every available frontier and open medical model against the set, in Hausa and Swahili, and report undertriage by language, by age band, and by presentation.
- Publish the benchmark, the harness, and the results openly. Free to use, including by our competitors.
Why Open
The benchmark is a public good or it is worthless. A safety standard that one organization owns is a marketing asset. Every implementer, ministry, and model developer should be able to run this and see their own number. We are funding the work. We are not keeping it.
Who Is Behind It
First Voice Health was initiated by Dylan Turner, co-founder of doxy.me. doxy.me is used by more than one million healthcare providers and has carried over 490 million patient visits.
Get involved
We are looking for clinicians who will adjudicate. If your organization operates community health, maternal health, or primary care programs in Hausa, Swahili, Yoruba, Igbo, Tagalog, Cebuano, or another underserved language, we want your clinicians as named co-authors on the benchmark.
Contact: hello@firstvoicehealth.org