The science
Four questions, computed.
Every plan comes down to four questions: Am I doing it? Is it working? Is anything off? Why? Kinosis answers each one every day from the person's own data. Here is what each one measures and how.
Protocol Adherence
Am I doing it?
What it measures
How closely what you do matches your plan, item by item, every day: medications, supplements, training, sleep, meals, habits.
How
Each item of the plan is checked against what you log and what your wearables sync. A reminder confirmed from the notification counts the moment you tap it.
Protocol Adherence blends two baselines: 50% your own rolling three-week average and 50% evidence-based targets (such as 7.5 hours of sleep or 150 minutes of Zone 2 a week). It rises as you improve against both.
Adherence is shown per item, so the one that slips is visible, not hidden in an average.
30–50%
Non-adherence
Patients not taking prescribed medication as directed, rising to 80–100% over extended follow-up. Chapman & Chan, Frontiers in Pharmacology, 2025.
About half of patients with chronic conditions don't take long-term therapy as prescribed. WHO, Adherence to long-term therapies, 2003.
Protocol Efficacy
Is it working?
What it measures
Whether each marker the plan should move is moving: weight, blood pressure, glucose, HRV, resting heart rate, sleep, lab results, symptoms.
How
For each marker, Kinosis takes the readings of the last 90 days from wearables, lab reports and what you log, and fits a trend line.
The trend becomes a direction (improving, steady or worsening) only when it is clear of the marker's normal day-to-day noise, and the value is placed against population percentiles.
Every marker is shown on its own, with its value and direction. There is no black-box composite.
Signals
Is anything off?
What it measures
A marker that stays off target with a flat or wrong-way trend, and, for people followed by a clinic, missed-dose streaks, recurring symptoms, out-of-range lab results and patients who go quiet.
How
Signals are deterministic rules. A marker signal needs several readings over months before it fires, so one bad week doesn't raise a flag.
When the plan includes something aimed at that marker and it's being logged, the signal says the marker isn't responding; when nothing is being done about it, the bar to fire is higher.
Rules decide what fires; the language model only explains it. Kinosis flags; it never diagnoses.
Correlations
Why?
What it measures
What moves together in one person's data: sleep and next-day energy, late meals and HRV, stress and flares, air quality and symptoms.
How
A correlation engine runs across hundreds of dimensions in 13 categories over a rolling 60-day window, with time lags, and surfaces a pattern only at |r| ≥ 0.4.
A second layer looks for evidence-based relationships that correlation alone can miss with little data: sleep and mood, exercise and energy, meal timing, day-of-week rhythms, supplements and symptoms.
Every log is stamped with time and place, so the daily environment (UV, air quality, temperature, humidity) is part of the picture.
Published science
Built on published science.
Kinosis does not invent biology. It gives the protocols that researchers and clinicians write a way to be followed and measured every day.
Mitochondrial function and Zone 2
Zone 2 training for metabolic flexibility, fat oxidation and cardiovascular base.
San Millán, I. · University of Colorado
Sleep architecture
Light timing, temperature and stage-specific sleep markers as recovery variables.
Walker, M. / Huberman, A.
Circadian and time-restricted eating
Eating windows and timing relative to light for metabolic health.
Panda, S.
Systematic measurement
High-frequency measurement and iterating a protocol on the person's own data.
Johnson, B. (Blueprint)