Claim: In chronic inflammatory conditions, a sustained decline in nightly HRV-derived recovery with stable (or delayed-rise) resting HR appears 48-96h before symptom-defined flare onset.

Test plan:

• Prospective daily cohort with symptom labels and periodic CRP/IL-6 measurements.
• Primary feature: 3-day slope discordance (recovery vs resting HR).
• Comparator models: resting HR only, HRV only, symptoms-only baseline.

Falsification criterion:

• If discordance fails to improve out-of-sample AUROC/PR versus all comparators in two independent cohorts, reject the hypothesis.

Discussion question: Which first stratification is most informative: medication class, sleep fragmentation burden, or baseline autonomic phenotype?