Activity-Recovery Coupling — Oura Digital Twin

Oura Ring Gen 4 sensor data — not clinical measurementsN=1 case study — not validated for clinical decisionsHEV diagnosed Mar 18; interpret findings cautiously in this Day 20 post-ruxolitinib window

HENRIK COUPLING r

Absent

-0.007

Steps vs HRV (lag 1, n=79)

MITCHELL COUPLING r

Absent

-0.092

Steps vs HRV (lag 1, n=518)

DIFFERENCE (FISHER z)

Not significant

0.488

p-value for coupling difference

HENRIK RECOVERY

Abnormal

Decoupled

negligible absent

MITCHELL RECOVERY

Abnormal

Decoupled

negligible absent

Executive Summary

This report examines whether physical activity on day N predicts sleep quality and autonomic recovery on day N+1 for both patients. Henrik (post-HSCT, ~2700 steps/day, HRV ~9ms) shows absent coupling (r=-0.007). Mitchell (post-stroke, ~10000 steps/day, HRV ~43ms) shows absent coupling (r=-0.092).

Both patients show absent activity-recovery coupling.

Activity vs Next-Day Recovery

Cross-Correlation Functions

Positive lags = activity leads recovery (causal direction). Negative lags = recovery leads activity (reverse check). * indicates p < 0.10.

Dose-Response: Activity Bins vs Next-Day HRV

Patient	N pairs	Bins	Kruskal H	p-value	Sig (p<0.10)
Henrik (post-HSCT)	79	4	0.21	p=0.976	No
Mitchell (post-Stroke)	518	4	9.03	p=0.029	Yes

Full Correlation Matrices (Lag 1)

Each cell shows Spearman r for activity[N] vs recovery[N+1]. * indicates p < 0.10. Colorscale: blue=positive, red=negative.

Steps vs Recovery Across Lags

Clinical Interpretation

Henrik (post-HSCT)

Coupling: absent (negligible) — r=-0.007, p=p=0.951, n=79
Recovery capacity: Decoupled
Activity and next-day recovery are effectively uncoupled. Other factors may dominate recovery dynamics.

Mitchell (post-Stroke)

Coupling: absent (negligible) — r=-0.092, p=p=0.036, n=518
Recovery capacity: Decoupled
Activity and next-day recovery are effectively uncoupled. Other factors may dominate recovery dynamics.

Cross-Patient Comparison

Both patients show absent activity-recovery coupling.

Fisher r-to-z test p-value: p=0.488

Caveats

Henrik has only ~79 usable lag-1 pairs. With small N, many true correlations will not reach statistical significance. Effect sizes (r values) are more informative than p-values.
Relaxed significance threshold (p < 0.10) used throughout to balance Type I and Type II error with limited data.
Observational design: correlations do not establish causation. Confounders (illness severity, medication, stress) are not controlled.
Henrik's extremely low activity level (~2700 steps, ~6 min active) provides limited variance to detect dose-response relationships.
Oura readiness 'resting_heart_rate' is a 0-100 score, NOT bpm. Heart rate values come from sleep_periods only.