Health predictions across 100+ wearables and apps
From the wearable and app data you already receive: sleep schedules, cycle phase, activity patterns, and the readings a device never took.
Run predictions over health data you already receive through Terra API
Different predictions help you understand and predict various events in your users’ health and fitness: activity patterns, menstrual cycle, sleep and lifestyle habits.
Choose your prediction and run it across selected users
Sleep returns 15+ metrics, dailies over 21, activities over 20. Predictions can run on any data type and selected users.
Know how much more you can get, user by user
You understand how much each prediction will increase the health context of your user.
Get confidence scores for each prediction
Confidence for each health event prediction is scored per device, so you always understand how meaningful the new metrics are.
Predictions your product can act on, without building a research team
Each prediction returns a structured output that can either be rendered directly in your app or used as additional context for AI to reason over your users’ data.
Foresee sleep changes
A sleep regularity index from 0 to 100, banded Low, Fair at 65, Good at 75, Excellent at 85, and showing changes that will likely help the user improve, such as closing a 57-minute weekend gap to go from Fair to Good.
Predict cycles
Cycle phase for every day in a range, with period onsets and fertile windows read from overnight physiology alone. The user logs nothing, and a day you have already shown stays fixed. Intended for wellness use only.
Infer missing data
The measurements a device never took, predicted from the channels it did capture and from the user’s own history. Every field is marked as recorded or predicted and carries a confidence tier, so you never show an estimate as a measurement.
A prediction catalogue that keeps growing
Sleep, women’s health and gap-filling are live now. Activity patterns, apnea screening, cycle anomalies, fragmented nights and metabolic health are moving through validation with design partners, and the research behind each one is published as it lands.
Sleep schedule
Generally available. The bedtime and wake time a user should keep, with a consistency score and the single change worth making.
Browse the research
Menstrual cycle
Generally available. Cycle phase for every day, with period onsets and fertile windows predicted from overnight physiology alone.
Browse the research
Missing readings
Generally available. The measurements a device left out, predicted from the channels it did capture, each field marked recorded or predicted.
Browse the research
Activity patterns
In validation. Repairs the activity type a user mis-tagged, reads the training intent behind a run, and surfaces workouts that were never logged at all.
Browse the research
Cycle anomalies
In validation. Flags sustained disruption to a user's established cycle rhythm, with an alert level that escalates the longer it persists.
Browse the research
Sleep apnea risk
In validation. A per-night risk reading and a user-level score from overnight data. Intended for wellness use only.
Browse the research
Fragmented nights
In validation. Regroups the sessions a device split across one night back into the single night they belong to, so a broken night reads as one.
Browse the research
Metabolic health
Metabolic predictions read from weight, nutrition, activity and recovery metrics to indicate energy changes and fluctuations.
Browse the research
A prediction for your use case
Bring the research team a question your data could answer and we will tell you whether it is tractable.
Browse the research
Sleep schedule
Generally available. The bedtime and wake time a user should keep, with a consistency score and the single change worth making.
Browse the researchMenstrual cycle
Generally available. Cycle phase for every day, with period onsets and fertile windows predicted from overnight physiology alone.
Browse the researchMissing readings
Generally available. The measurements a device left out, predicted from the channels it did capture, each field marked recorded or predicted.
Browse the researchActivity patterns
In validation. Repairs the activity type a user mis-tagged, reads the training intent behind a run, and surfaces workouts that were never logged at all.
Browse the researchCycle anomalies
In validation. Flags sustained disruption to a user's established cycle rhythm, with an alert level that escalates the longer it persists.
Browse the researchSleep apnea risk
In validation. A per-night risk reading and a user-level score from overnight data. Intended for wellness use only.
Browse the researchFragmented nights
In validation. Regroups the sessions a device split across one night back into the single night they belong to, so a broken night reads as one.
Browse the researchMetabolic health
Metabolic predictions read from weight, nutrition, activity and recovery metrics to indicate energy changes and fluctuations.
Browse the researchA prediction for your use case
Bring the research team a question your data could answer and we will tell you whether it is tractable.
Browse the researchThe most comprehensive suite of health predictions
Terra predictions target 90% coverage and measure themselves against that: 0.901 on sleep, 0.898 on dailies, 0.893 on activities, calibrated across over 4,300 users.
- Recommended bedtime
- Daily cycle phase
- Value provenance
- Recommended wake time
- Period onset
- Per-field confidence
- Time in bed
- Fertile window
- Predicted sleep fields
- Sleep regularity index
- and more…
