11 statistical reports. 105,000+ biometric readings. One patient, one Oura Ring Gen 4, zero clinical infrastructure.
| Metric | Pre (71 days) | Post (8 days) | Change | Direction |
|---|---|---|---|---|
| Resting HR | 85.1 bpm | 81.8 bpm | −3.3 bpm | ↓ Improved |
| HRV (RMSSD) | 9.1 ms | 10.1 ms | +1.0 ms (+11%) | ↑ Improved |
| Temp deviation | +0.08 °C | −0.16 °C | −0.24 °C | ↓ Improved |
| Sleep quality | −0.14 SD | +0.37 SD | +0.51 SD | ↑ Improved |
| Sleep efficiency | ~73% | ~79% | +6 pp | ↑ Improved |
| SpO2 (BOS screen) | 96.1% | 96.0% | −0.1% | Low baseline · CT air trapping |
Mother (61F) enrolled with Oura Ring. ~50% shared genome. Same pipeline, no disease. Separates inherited baseline from disease signal.
30+ nights targeted. Between-subjects comparison will quantify how much of the autonomic signature is inherited vs. disease-driven.
Read roadmap →Heart rate variability deep dive — time-domain, frequency-domain, and nonlinear analysis
Patient-specific physiological model with treatment response simulation
Amazon Chronos time series foundation model for biometric prediction
Matrix profile and ensemble methods for clinically significant event detection
Three-dimensional visualization of multi-modal biometric trajectories
Open-source platform built by a single post-HSCT patient using Oura Ring Gen 4 data. MIT licensed through The Educational Equality Institute.
| Phase | Target | Description | Status |
|---|---|---|---|
| Now | Day 8 | Ruxolitinib response — current analysis window | Active |
| Day 28 | ~Apr 13 | Statistical significance threshold — if signals persist | Pending |
| Family control | Day 1 | Mother enrolled. ~50% shared genome, same pipeline | Enrolled |
| Multi-user | TBD | GvHD vs stroke vs healthy — between-subjects CausalImpact | Planned |
All analyses are reproducible Python scripts running against a single SQLite database. The full pipeline runs in under 10 minutes on commodity hardware.