Upload your AncestryDNA raw file today for a 608-SNP trait report that runs on your own machine. Wearable integrations — Garmin, Oura, Whoop, Apple Health — and an opt-in path to anonymized aggregate research that shares the proceeds back with you are in development.
Early private beta · No spam · One-click unsubscribe
Where this actually stands. OmniHealth is early. What works today is the genetics tooling — a parser and a 608-SNP analyzer that run locally on a file you supply. Wearable integrations, accounts, the hosted service, and the research programme are all still being built. Everything on this page is marked for what it is: shipped, in progress, or planned. We would rather under-promise on a page about privacy than discover later that we described software that didn't exist.
Your wearable's app shows you yesterday's steps and last night's sleep score. The vendor keeps the raw signal, monetizes it through partnerships you don't see, and gives you a 1-decimal "readiness" number in return.
DNA tests sell your raw genome to pharma companies for research you'll never hear about. Health-data brokers aggregate everything from steps to cycle data and sell it to insurers, employers, and advertisers — sometimes anonymized, often not.
You own your data. Today the genetics analysis runs locally on a file you supply — nothing is uploaded, because there is no account to upload it to yet. As wearable integrations land, the same principle governs them: heart-rate variability patterns, training-load trajectories and sleep-stage transitions are computed for you, not harvested from you.
The research programme is a design on paper, not a running system. When it ships, and only if you opt in, derived statistical aggregates (never raw records, never identifiers) will contribute to a community insights pool. It is being designed around a k-anonymity threshold of k ≥ 50, a documented differential-privacy budget, and revenue sharing with contributors. Until it exists, nothing is aggregated and nothing is shared. The default is private.
Shipped: local-first analysis — the genetics tooling reads a file on your machine and writes results beside it. Specified, not yet built: encryption at rest, TLS in transit, per-user key isolation for sensitive streams, access logging, and a self-hosted option for the eventual service. Those are the requirements we are building to, not features you have today.
HRV trajectories, sleep-stage transitions, and training-load curves are time series with non-trivial geometry. OmniHealth will apply the same Riemannian methods we publish in peer-reviewed venues — and report plainly where they beat flat-vector baselines and where they don't. This is an open research direction, not a settled result.
Planned. Opt in to the community insights pool and your derived aggregates would earn a share of any research revenue they generate, with a dashboard showing how data was used, which studies accessed it, and what you earned. None of this is built yet, and the pool does not open until the privacy machinery above is real.
One layer works today. The rest is the roadmap, labelled honestly.
Analyze your raw AncestryDNA file across 608 SNPs in 36 trait categories — physical traits, metabolism, fitness, sleep, cognitive, pharmacogenomics, longevity. Runs locally during the beta; the file never leaves your machine. 23andMe format is planned — the parser currently reads AncestryDNA's export layout only.
Direct integration via Garmin's Connect Developer Program, pending credentials. The OAuth flow is scaffolded and FIT-file decoding is partial — generic records parse, HRV and sleep-stage extraction do not yet. Any integration will comply with Garmin's no-raw-redistribution license.
Unified wearable integration, most likely through an aggregator API. Faster to connect than direct integrations; normalized data covers the broadest wearable ecosystem.
Track biomarkers over time with genetic context. Optimal ranges, not just "normal." Alert on out-of-range values.
Macro/micro tracking, mapped to genetic metabolism markers. Find nutritional gaps you didn't know you had.
Symptoms, mood, custom metrics. Run n=1 experiments and find correlations across every data type you've connected.
The commitments the product is being built around. Where something isn't implemented yet, this page says so.
No advertiser auctions, no health-data brokers, no insurer or employer disclosures. This one is not a roadmap item — it is a standing commitment that applies now and applies to whatever we build later.
Today nothing is shared at all: the genetics tooling runs on your machine and we hold no health data. When the opt-in research programme exists, what leaves your account will be statistical — aggregates passing a k-anonymity threshold of k ≥ 50 and a documented differential-privacy budget. Raw heart-rate, sleep, HRV, or genotype values will never leave your account.
Full data export in JSON and CSV, and complete account deletion, are hard requirements for the hosted service — not upsells, and not features to be buried. During the beta there are no accounts to delete: results live in files on your machine, so deleting the file deletes the data.
Right now the only thing we collect is the email address you send to the waitlist. When wearable connections arrive, each will ask for the minimum scopes that fit the insights you want, revocable individually.
For the full picture, read our Privacy Policy and Terms of Service. We are building to GDPR, CCPA/CPRA, and Washington's My Health My Data Act — including the private right of action it grants you against us if we get it wrong.