Normalize
Reconcile measurements, timing, and source quality from the devices a person already wears.
Research for predictive health.
Yura is building models that learn what is normal for each person, combine related changes across wearables, and surface a watchpoint when the pattern holds together.
One person. One evolving baseline.
Population averages miss the context inside a single life. Yura starts with the individual, then asks whether several changes are moving together.
Learn
Estimate the person's recent range and how it changes across sleep, recovery, temperature, breathing, and strain.
Watch
Surface a watchpoint only when related shifts persist, with the contributing signals and confidence kept visible.
Not every change is a signal.
Yura looks for coordinated movement across measurements instead of reacting to a single noisy number. The model keeps the drivers attached to the result so a user can see why a watchpoint appeared.
Skin temperatureAbove personal range
Respiratory rateTrending upward
Sleep durationBelow recent baseline
Resting HRSustained change
48–72hExample forecast window with signal drivers
The research program.
The work is organized around usefulness, calibration, and evidence — not a larger dashboard.
Personal baseline modelling
Learn stable individual ranges while accounting for device changes, missing data, time of day, and measurement quality.
Multimodal signal fusion
Test whether related changes across several inputs provide more useful context than any metric viewed on its own.
Prospective evaluation
Measure calibration, false-alert rate, lead time, and user response in real-world cohorts before expanding product claims.
Built to show its work.
Source lineage
Keep the device, timestamp, and data-quality context attached.
Visible drivers
Show which changes moved together and over what period.
Explicit limits
Separate wellness watchpoints from diagnosis, treatment, and emergency care.
Help us build the evidence.
We welcome conversations with researchers, clinicians, data partners, and teams working on prospective validation.