Private by design
Sanna is being shaped around on-device processing. Health signals stay on the iPhone, without accounts, cloud scoring, or background profiling.
Methodology
Building a readiness algorithmpeople can actually trust.
Sanna's full methodology is not final yet. We are developing the on-device algorithms with beta testers so the score becomes personal, explainable, and useful before it becomes polished.
On-device firstTester-calibrated
Signal before certainty
We focus on patterns that are useful day to day: recovery, sleep consistency, strain, resting trends, and whether today's body state looks typical for you.
Calibrated with testers
Beta testers help us compare the score against lived experience: energy, soreness, stress, sleep quality, and whether the explanation actually makes sense.
What we evaluate
Readiness is a relationship between signals, not a single metric.
The goal is not to diagnose, prescribe, or reduce health to one number. The goal is to summarize whether your recent recovery and strain look supportive, mixed, or worth taking gently.
Sleep duration and regularity
Resting heart rate trend
Heart rate variability trend
Recent activity load
Recovery after strain
Outlier days and missing data
The beta loop
We are using the private beta to learn where the algorithm is helpful, where it is too confident, and where the explanation needs to be clearer.
01
Baseline
The app learns what normal looks like for each tester instead of treating a population average as the target.
02
Explain
Every readiness estimate needs a short, readable reason. If a score cannot be explained clearly, the model needs more work.
03
Compare
Tester feedback helps us find gaps between the algorithmic signal and how the body actually felt that day.
04
Tune
We adjust weighting, confidence, and edge cases while keeping the computation local to the device.
Current status
Honest limits are part of the methodology.
The model is still being validated across different routines, wearable data quality, sleep schedules, and training loads. A readiness score is only useful if it respects uncertainty.
That is why Sanna will surface confidence, explain what changed, and avoid pretending that one daily score can capture the whole body. The methodology will become more specific as beta evidence becomes stronger.