01
847 studies indexed
Count of indexed peer-reviewed studies referenced in the literature validation layer.
Exercise Physiology
Lactate dynamics, VO₂ kinetics, and metabolic flexibility under progressive endurance load.
DL-SCI · 2026 Live research
Physiology Engine · Research
A lab that turns wearable signals into a physiology engine
Driftline is not a chat bot. It turns HRV, pace, and load signals into physiological models, validates with literature, and updates the plan proactively — every step reproducible.
Telemetry · SummaryUTC+3 · 2026
0,0M+
Athlete data points
Anonymized HRV, pace, sleep, and load records — used for individual physiological normalization.
0
Reference papers
Peer-reviewed journal studies cited directly in the engine decision frame — not Driftline publications.
0
Case studies
Interventions validated on real athlete profiles in taper, pace-drift, and ACWR scenarios.
Explore
Physiology Engine
Not an AI coach — a physiology engine.
Driftline is not a chat interface. It is a decision engine that turns wearable signals into physiological models, validates drift with literature, and updates the plan proactively. WhatsApp is only the output channel.
Not a chat bot
The LLM layer only explains. Decisions come from physiological models and the rule engine.
Not a fixed plan
Signals are recomputed daily. The plan is not a static PDF — a living physiological surface.
Evidence chain
Every intervention: signal → hypothesis → literature → simulation → output. Reproducible and auditable.
Motor · v3.2 Live
Girdi
HRVTempoUykuYük
Motor
NormalizasyonSapmaSimülasyon
Doğrulama
LiteratürGüven skoru
Çıktı
PlanGerekçeWhatsApp
Karar gecikmesi p99 · 84ms · her müdahale denetlenebilir iz bırakır
Scientific Methodology
A six-stage, reproducible pipeline.
Every intervention follows the same protocol: signal collection, individual normalization, drift detection, literature validation, simulation, and transparent output. Not chat — experimental design.
011.2k/s
Signal collection
Continuous streams from Garmin, Strava, WHOOP, and Oura.
HRV, sleep architecture, pace, power, load, and readiness scores at 15-minute granularity. Multi-source fusion for missing data.
02σ individual
Physiological normalization
Personal baseline and z-score calculation per athlete.
Seasonal, menstrual, and training-phase effects are separated. RMSSD, CTL/ATL, and pace drift are normalized to the individual distribution — no fixed thresholds.
0394% precision
Drift detection
Anomaly engine turns physiological drift into a hypothesis.
Multi-signal correlation classifies overload, recovery debt, or adaptation plateau. Every alert carries a confidence score.
04847+ sources
Literature validation
Recommendations are cross-checked against indexed peer-reviewed work.
847+ peer-reviewed studies on HRV-guided training, load monitoring, and periodization. The engine produces an evidence chain, not chat.
0584ms latency
Intervention simulation
Models load and performance impact of a plan change.
Taper, deload, and intensity-shift scenarios tested with CTL/ATL projection. Lowest-risk intervention is selected.
0624/7
Proactive output
Updated plan + physiological rationale reach the athlete.
Engine decision via WhatsApp or app: which signal, which threshold, which literature, which plan change — transparent and reproducible.
Why ACWR alone is not enough
Classic ACWR (7-day mean ÷ 28-day mean) creates artificial day-to-day noise: one hard session inflates acute load while chronic lags; a rest day drops the ratio — calendar artifact, not true fatigue. Impellizzeri et al. (2020) showed statistical inconsistencies and limited injury-risk prediction of rolling-average ratios; Bourdon et al. (2017) recommend exponentially weighted moving averages (EWMA) instead.
Impellizzeri F.M., Tenan M.S., Kempton T., Novak A.R., Coutts A.J. (2020) — International Journal of Sports Physiology and Performance. doi:10.1123/ijspp.2019-0820
Driftline never uses load ratio (ACWR / CTL·ATL derivative) as the sole decision maker. Three parallel signals are watched: HRV recovery, easy pace drift, and acute–chronic load balance. Drift detection (step 03) fires not only when a load threshold is crossed — when signals co-anomaly; every alert carries individual z-score and confidence. The load layer targets EWMA-based modeling (Bourdon frame); volume is not pulled back on load alone if HRV and pace drift do not confirm the same intervention.
HRV recoveryPace driftLoad ratio
Case Studies
Real signals, real interventions.
Anonymized athlete profiles — how the engine catches drift, the physiological rationale for plan changes, and outcomes.
Elite triathlete · age 34
Ironman 70.3
8-week build → taper
HRV trend
RMSSD −18% (4 days)
Final build week, CTL 92, ATL 78
01
Tespit
Engine detected pre-taper autonomic suppression. CTL/ATL rose to 1.18; deep sleep −22% overnight.
02
Müdahale
Load cut 32%, Z2 ceiling applied. Thursday VO₂ set cancelled; 40 min technique swim instead. Taper protocol advanced 48 hours.
03
Sonuç
Personal best · TSS −28% last 10 days. No injury or illness reported. HRV normalized by taper day 5.
Race performance
Personal best · TSS −28% last 10 days
Javaloyes 2019Gabbett 2014
Research Areas
Six domains. One integrated lab.
Interdisciplinary research spanning physiology, data science, and applied coaching — designed for reproducibility and real-world impact.
02
12.4M data points
Biometric records used to analyze performance adaptation across training conditions.
HRV & Recovery
Autonomic nervous system markers, sleep architecture, and readiness prediction models.
03
23 labs linked
Research labs linked for multi-center validation and biomechanical calibration.
Biomechanics
Running economy, cadence drift, and power-phase analysis in multi-sport athletes.
04
v3.2 engine
Production physiology-engine version; every decision leaves an auditable trail.
Physiology Engine
Decision engine that turns wearable signals into physiological models and cross-checks interventions with literature.
05
156 protocols
Nutrition–hydration protocol records tested in race and hard training blocks.
Performance Nutrition
Race fueling protocols, hydration thermodynamics, and substrate utilization mapping.
06
94% accuracy
Classification accuracy on the cohort validation set for ACWR-based load-ratio prediction.
Periodization
Block periodization, taper optimization, and load-ratio prediction for peak performance.
Publications
Reference publications
Peer-reviewed papers cited directly in the engine decision frame — Sports Medicine, IJSPP, npj Digital Medicine, Frontiers in Physiology, and Sports.
Methodology note847+ indexed sources5 direct reference papers
The 847+ source literature pool is an editorially curated index of HRV-guided training, load monitoring, periodization, and wearable-sensor literature. Records are held with structured PubMed and DOI metadata; every engine intervention is cross-checked via semantic match against this pool. This is not a live automatic web crawl — a pre-validated, updated reference library. The 5 publications below are direct decision anchors; the rest of the index is secondary evidence for the same decision chain. The research assistant UI is currently demo mode; production literature validation runs via rule engine + index matching.
Decision anchors — 5 reference papers
Each row shows which physiology-engine rule the paper anchors. Full text and methodology details follow in the cards below.
- 012019International Journal of Sports Physiology and PerformanceHRV & Recovery
Heart Rate Variability–Guided Training Prescription in Cycling
Decision rule · When daily HRV falls for 3+ consecutive days, intensity redistribution and pre-taper load reduction decisions rest on this study.
- 022020Frontiers in PhysiologyExercise Physiology
HRV Fractal Correlation Properties for Intensity Distribution in Endurance Exercise: A New Biomarker?
Decision rule · Pace drift and intensity-distribution drift detection in easy zones — volume adjustment via fractal HRV features rests on this work.
- 032019npj Digital MedicinePhysiology Engine
Wearable Sensors for Monitoring the Internal and External Workload of Athletes
Decision rule · Internal–external load fusion and signal reliability from multi-wearable sources (Garmin, WHOOP, Oura) rest on this synthesis.
- 042020SportsBiomechanics
Validity of the Stryd Power Meter for Measuring Running Parameters at Submaximal Speeds
Decision rule · Power-meter and pace monitoring validity at submaximal speeds — easy-run pace-drift thresholds rest on this study.
- 052014Sports MedicinePeriodization
The Training–Injury Prevention Paradox: Should Athletes Be Training Smarter and Harder?
Decision rule · ACWR-based overload detection, deload triggers, and taper timing rest on this framework.
Tam metin kartları
Araştırma amacı, kullanılan veri, yöntem ve sonuç — her referans için genişletilmiş özet.
2019International Journal of Sports Physiology and PerformanceHRV & Recovery
Heart Rate Variability–Guided Training Prescription in Cycling
Karar kuralı · When daily HRV falls for 3+ consecutive days, intensity redistribution and pre-taper load reduction decisions rest on this study.
2020Frontiers in PhysiologyExercise Physiology
HRV Fractal Correlation Properties for Intensity Distribution in Endurance Exercise: A New Biomarker?
Karar kuralı · Pace drift and intensity-distribution drift detection in easy zones — volume adjustment via fractal HRV features rests on this work.
2019npj Digital MedicinePhysiology Engine
Wearable Sensors for Monitoring the Internal and External Workload of Athletes
Karar kuralı · Internal–external load fusion and signal reliability from multi-wearable sources (Garmin, WHOOP, Oura) rest on this synthesis.
2020SportsBiomechanics
Validity of the Stryd Power Meter for Measuring Running Parameters at Submaximal Speeds
Karar kuralı · Power-meter and pace monitoring validity at submaximal speeds — easy-run pace-drift thresholds rest on this study.
2014Sports MedicinePeriodization
The Training–Injury Prevention Paradox: Should Athletes Be Training Smarter and Harder?
Karar kuralı · ACWR-based overload detection, deload triggers, and taper timing rest on this framework.
1847atıf
doi:10.1007/s40279-014-0253-z