Frontier AI for overlooked patients

Replica Health is an AI research lab that builds and ships software for chronic conditions the tech industry has passed over. We find patient populations that are too small for big tech to serve — and work to bring them state-of-the-art AI, ASAP.

How we work

A repeatable playbook for underserved conditions

1

Find the gap

We look for conditions where patients face daily, data-rich problems — but where the population is small enough that mainstream digital health never shows up.

2

Build the engine

Standardized datasets, high-speed simulators, and rigorous validation pipelines, built once and reused wherever possible.

3

Ship fast

AI-accelerated R&D and on-device machine learning take us from idea to working product in months.

The engine

Bringing frontier AI to people who use medical devices

Every initiative runs on a similar toolkit: AI assisted R&D pipeline, standardized datasets, accelerated simulation, sim-to-real validation, and when possible, models that run on the user's hardware, rather than in the cloud.

Explore the platform
Colorful multiplying cubes illustration
MetaboNet illustration
Open research

Research that moves the whole field

We publish. MetaboNet — the "ImageNet for diabetes" — is the largest standardized Type 1 diabetes dataset, free for researchers — and the foundation for MetaboNet-Bench, a glucose-forecasting benchmark built in collaboration with Stanford. GluPredKit standardizes blood-glucose prediction research. Open tools bring better care to patients sooner, whoever builds it.

2,030
patient-years of data
3,135
subjects across studies
21
datasets consolidated
See our research

Stay in the loop with Replica Health

Sign up to follow our progress and to learn about updates to our work.