Prana, a living brain.

Prana is building a model of how illness moves through India.

It learns from triage. A patient describes what is wrong in their own words, a doctor reviews where the model sent them, and we record what happened next. Insurers, pharma companies and public health teams can use that record to see an illness rising before claims and prescriptions show it.

01Doctors are scarce

India doesn't have enough doctors, and most of them work in cities. A patient in a hospital outpatient queue often gets a few minutes. Plenty of people describe their symptoms first to a pharmacist, or to no one.

Prana takes that first conversation, in Hindi or English, and sends the patient to the right department. Doctors review its routing and correct it. Prescribing and signing stay with them, as Indian telemedicine rules require.

02The loop

When a doctor corrects a route, the correction becomes training data, and so does the patient's follow-up. We retrain on those cases.

patientin their words modelroutes doctorjudges the route outcomefollow-up, claim modellearns retrained on reviewed cases
Fig. 1. How a case becomes training data.

03One patient, as the model sees it

Each case is a sequence: what the patient reported, where they were sent, what the doctor decided, and what happened later. With enough cases we can compare districts and age groups.

day 0 day 7 day 40 state "सीने में भारीपन, दो दिन से" action general medicine verdict doctor: wrong desk, cardiology outcome "better" outcome claim: procedure
Fig. 2. A made-up example. The patient said their chest had felt heavy for two days. They were sent to general medicine, a doctor moved them to cardiology, and the follow-up and the claim were added later.

04What other records miss

Hospital software stores diagnoses and insurance claims store bills. Neither has what the patient said before diagnosis, a doctor's view of where they were first sent, or anyone who didn't file a claim.

weeks of warning symptoms at triage claims filed time
Fig. 3. Illustrative shape, not data. Symptoms show up in triage before they show up in claims.

05Who uses it

Prana answers two kinds of question. Some are about one patient or one claim and need an answer immediately. Others are about groups, such as which illnesses are rising in which districts.

One caseGroups
InsurersDoes this claim match what the patient came in with? Is this admission warranted?What to cover, how to price it, where to grow
PharmaWhich consenting patients may be eligible for a trialWhich diseases to work on, where patients are, how long diagnosis takes
GovernmentClaim checks for public insurance; triage in primary careEarly signals for disease surveillance; district-level burden

06How we work

  1. We check consent each time patient data is used. Patients who say no get the same care.
  2. Patient data is stored and processed in India.
  3. We keep each hospital's records separate.
  4. What the model says to a patient is logged before the patient hears it. Clinicians review those logs.
  5. We pay doctors to review and label clinical cases.
  6. Buyers receive aggregate reports only.

07Built on ABDM

ABDM, India's digital health mission, gives each person an ABHA health ID and each facility a registry entry. Prana adds doctor-reviewed triage at the first point of contact.

94.87 crABHA health IDs created
5.36 lakhhealth facilities registered
As of 20 July 2026. Source: Ministry of Health and Family Welfare, via PIB.