AI Health Models
Structured physiology. The tokenizer for heart rate.
Foundation models in health train on raw sensor traces—noisy, unlabeled, unstructured. TrueZone converts the same stream into labeled, physiologically-grounded features that actually teach the model something.
Raw Data Today
Noise without labels
HR traces, step counts, sleep stages. No physiological ground truth. Models memorize individuals, not physiology.
TrueZone Feature Layer
Labeled physiology
E, V, P + 60 derived metrics per session. Thresholds, substrate balance, fatigue state, recovery kinetics—all interpretable, all structured.
Unlock
Foundation-model ready
Train health models on features grounded in ODE physiology, not black-box neural embeddings. Generalizes across populations.
Compatible with any biometric dataset · preserves interpretability · generates raw label axes from existing HR archives.
The Problem
Raw heart rate data doesn't teach models physiology.
A foundation model trained on millions of HR traces learns statistical patterns—correlations between heart rate values and timestamps. It doesn't learn that a heart rate of 155 bpm means something completely different for a marathoner (below threshold, sustainable for hours) than for a sedentary person (above threshold, glycolytic, unsustainable).
Without physiological labels, models can't distinguish fitness from fatigue, aerobic from anaerobic, adaptation from overtraining. They fit curves without understanding what the curves mean.
What raw HR gives a model
- Timestamps and BPM values
- Statistical correlations (HR vs time of day, vs activity)
- Population-level patterns
- No individual physiology
- No threshold positions
- No metabolic context
What TrueZone features give a model
- Individual physiological parameters (E, Vmax, P)
- Threshold positions (LT1, LT2, Fatmax, VO₂max)
- Substrate balance (fat vs carb oxidation ratio)
- Fatigue state and recovery kinetics
- Metabolic fitness score (MFI) with clinical correlation
- 60+ derived metrics, all interpretable
The Feature Layer
60+ physiological features per session. All from heart rate.
TrueZone fits a physiological ODE to each individual's heart rate response, extracting three base parameters. From those, it derives a complete physiological profile—every feature grounded in the model's physics, not learned from data.
Base parameters
Endurance (E), Max Speed (Vmax), HRmax (P), confidence intervals, variance matrix
Thresholds
LT1, LT2, Fatmax, VO₂max speed, 6 speed thresholds (V0–V5), heart rate at each threshold
Energy & metabolism
RMR, session calories, fat/carb oxidation, fat max rate, carb footprint, substrate balance, TEF, TDEE
Zones & load
6 HR zones, 3 metabolic zones, 12 fiber zones, training load (oxidative/glycolytic split), glycolytic ratio
Recovery
HR recovery curve, tau constants, EPOC (fast/slow), muscular recovery hours, time to HR recovery
Cardiac drift
Drift magnitude, drift rate, metabolic debt, drift percentage
Body composition
Muscle %, fat %, muscle mass, fat mass, BMI-adjusted metrics
Performance
Race predictions (100m–marathon), PrimeScore, Cooper test, VO₂max, training paces
Why This Matters
Physics-grounded features generalize. Statistical embeddings don't.
A neural network trained on raw HR data from 100,000 runners learns patterns specific to that population. Change the demographic, the device, or the activity type and the model breaks. TrueZone's features are derived from a physiological ODE—they mean the same thing regardless of who collected the data or what device was used.
Interpretable
Every feature has a physiological meaning. E = 0.75 means something specific about aerobic capacity. A neural embedding dimension does not. Clinicians and researchers can audit what the model is learning.
Portable
Features transfer across devices, populations, and sports. A model trained on Apple Watch data with TrueZone features works on Garmin data. The physiology is the same; only the raw signal differs.
Retroactive
Any existing HR archive can be processed through TrueZone to generate physiological labels. Millions of sessions already collected become a labeled training set overnight.
Use Cases
Where structured physiology unlocks new models.
Metabolic health prediction
Train models to predict glycemic response, insulin sensitivity, or metabolic syndrome risk using MFI, RZI, and substrate balance as input features instead of raw HR. MFI already correlates R² = 0.99 with glycemic markers.
Personalized training AI
Recommendation engines that prescribe training based on individual endurance, threshold positions, and recovery state—not population averages. Features update with every session.
Clinical digital biomarkers
Physiological features as endpoints in clinical trials. Track metabolic flexibility, aerobic decline, or intervention response continuously from wearable data instead of periodic lab visits.
Population health surveillance
Process millions of existing HR records into structured physiological profiles. Identify metabolic risk trends at population scale. Every wearable user becomes a data point with physiological meaning.
Integration
SDK or API. HR in, physiology out.
Process sessions through the REST API or run the SDK on your own infrastructure. Input: heart rate + speed/power time series. Output: 60+ structured physiological features per session, Bayesian parameter history, and confidence intervals. No training data, no GPU, no population model.
Send HR data
Heart rate + speed/power time series. Any device, any format. Historical archives work too.
ODE fitting
Bayesian parameter extraction. E, Vmax, P converge per individual. Confidence intervals included.
Get features
60+ physiological features returned per session. Structured JSON. Ready for your ML pipeline, clinical dashboard, or recommendation engine.
Stop training on noise. Start training on physiology.
TrueZone is available as a C++ SDK, REST API, or NPM package. Patent-protected to 2045. Compatible with any HR data source.