Privacy and progress should be the same thing.
Synthetix Health is a research and engineering lab in Stockholm.Our mission is to make the world’s health data usable, and no patient vulnerable.
The evidence medicine needs is already collected. It just can’t be used.
01The constraint
Every hospital visit, prescription and lab result becomes a record. Together those records hold the answers to questions medicine has not yet asked. But they sit inside institutions and national borders, protected by laws that exist for good reason.
So research waits. A feasibility question that should take an afternoon takes months of agreements. Whole populations are absent from the datasets models are trained on, because the data that describes them cannot travel.
02The old answer is ending
For two decades the workaround has been de-identification. Strip the names and hope enough was removed. It was never a guarantee, and regulators have noticed. Draft European guidance now asks for evidence that identification is effectively precluded, and the guidance written for health data access bodies examines differential privacy as a method that can show it.
The field is moving from promises to proofs. Data that cannot carry its own proof will stop moving.
03A standard, not a tool
Medicine became trustworthy when it was made to a standard. The prescription put a guarantee between the compound and the patient. Made to spec, labelled honestly, traceable end to end. Strangers could trust a pill they would never see made.
We are building that standard for health data. Synthetic cohorts generated under formal ε-differential privacy, each with a certificate that states its privacy budget, its measured utility and its contraindications, and a hash-chained audit trail an auditor can independently recompute.
04Honesty as a design principle
Synthetic data is built for specific questions and unsafe for others. Depending on the measured privacy–utility trade-off, it can support pipeline validation, cohort sizing and model benchmarking. Under strict person-level bounds, it is not suitable for safety-signal detection, rare-disease cohorts, or as primary evidence in a regulatory submission.
We print that on every certificate. A standard that only lists strengths is marketing.
05The long horizon
From 2029 the European Health Data Space opens secondary use of health data across the Union. Access bodies will need to decide what a safe release looks like. That definition does not yet exist. We intend to help write it, in public, with the institutions that will apply it.
How we decide, when it is hard.
Prove, don’t promise.
Every privacy claim is a number with a method behind it. If we cannot state the bound, we do not make the claim.
Print the contraindications.
Each dataset says what it must not be used for. Saying no to the wrong use is part of the product.
Records stay home.
The engine runs inside the institution. Raw patient records never cross its boundary; only certified cohorts do.
Regulation is a design input.
We read the guidance before we write the code, and we design for the strictest jurisdiction a cohort will touch.
Publish what we can.
Our regulatory analysis and our measured results are public. What we protect, we protect through filings, not silence.
Build for every population.
A standard that only works for the best-documented patients is not a standard. We build supply where data is scarcest.
Built from the filings up.

- BasedStockholm, Sweden
- FilingIN 202641026065
- InternationalPCT/IN2026/051623
- LinkedInin/gauravraj7
Synthetix Health was founded and engineered by Gaurav Raj Chattarki, who designed the platform, wrote the engine and authored the patent filings behind it. Reach him directly at gaurav@synthetixhealth.com.
The company is now headquartered in Stockholm and assembling a founding team of privacy researchers, machine-learning scientists and clinical data engineers who want their work to be provable and used in medicine.