Interested in hosting a ValiDATAthon? Partner with us

Get in touch

Words worth knowing

Plain English. No jargon.

valiDATAthon

A focused, time-boxed session where clinicians blindly evaluate AI tools against real Indian clinical scenarios using a validated rubric.

LLM

A large language model: an AI trained on vast text that answers, summarises, and reasons in language.

Benchmarking

Testing AI against a standard set of cases so performance can be compared fairly and tracked over time.

EWG

Clinicians and specialists who design scenarios, rubrics, and quality standards for their field.

HypMOOVE

A MOOVE built to test a specific clinical hypothesis about model behaviour.

retro-MOOVE

A MOOVE constructed from previously seen, real-world clinical encounters.

IRR

A measure of how consistently different expert evaluators score the same items.

Model Card

A document describing a model's intended use, performance, and limitations — including PrAImaan results.

MOOVE

The structured case format clinicians evaluate in a valiDATAthon.

ABDM

Ayushman Bharat Digital Mission: national infrastructure for secure, interoperable health records.


Common questions, honest answers


Evidence and outputs

Coming 2026

PrAImaan Validation Protocol v1

The methodology and rubric design behind the valiDATAthon, including IRR approach.

Download →
Coming 2026

India Health-AI Evidence Report

First aggregate findings across early valiDATAthon sessions and specialties.

Download →
Coming 2026

Model Card Framework

How PrAImaan documents model performance for Indian clinical contexts.

Download →

Less paperwork. More patient time.

Drag to compare documentation time, before and after voice scribing.

Before voice scribing

  • ~2 hrs/day on documentation
  • Notes written after clinic hours
  • Less eye contact with patients

After voice scribing

  • Documentation drops sharply
  • Notes drafted during the visit
  • More time facing the patient