A Stability Index for Cross-Domain Degradation Detection

We report an empirical pattern: a stability index Φ, formed from normalized identity (baseline preservation), temporal coherence, and Shannon entropy, separates stable from failing regimes across diverse engineered systems using a common critical region near 0.25 under a fixed protocol. The protocol and critical region were fixed from bearing studies (October 2024) and then applied unchanged across six engineered domains in 31 evaluated cases: ten run-to-failure bearings, ten run-to-failure turbofan engines, two power-grid frequency recordings (including the Great Britain blackout of 9 Aug 2019), three earthquake-precursor cases plus a quiet baseline, two neural-network training cases, and 445 superconducting qubits across three IBM Quantum backends. Across non-quantum domains, evaluated failure/critical intervals yielded Φ below the critical region while included stable controls yielded Φ above it under their domain protocols. We additionally report a biological extension on cardiac and neural signals. For cardiac rhythm windows, arrhythmia discrimination achieved AUC ≈ 0.90. We report the observation without claiming a mechanistic derivation and provide sufficient methodological detail for independent scrutiny.