A Proof-of-Concept Deterministic Phase-Memory Operator for Respiratory Instability Detection Using Chest-Mounted Smartphone IMU Signals

Abstract

Wearable respiratory monitoring often relies on heuristic pipelines or opaque machine learning models, which can limit interpretability and auditability in safety-sensitive or clinical adjacent contexts. Here, we present a proof-of-concept deterministic phase-memory operator for respiratory instability detection using chest-mounted smartphone inertial measurement unit (IMU) signals. The proposed instability metric △Φ(t) quantifies deviations of instantaneous phase velocity from short-term phase memory, enabling transparent threshold-based decision logic without training dependence. A controlled validation protocol based on N = 5 publicly available Beth Israel Deaconess Medical Center (BIDMC) respiratory recordings with semi-synthetic perturbations was used to examine representative deviation regimes, including frequency drift, intermittent pauses, and burst irregularities. Performance was compared against low-overhead baseline methods based on root mean square (RMS)-envelope and fast Fourier transform (FFT)-peak tracking. Within this limited proof-of-concept setting, the framework showed reproducible responses to structured perturbations. The reference Python pipeline processed 60 s respiratory traces in 2.37 ms on a standard x86-64 system, while the intended mobile implementation remains compatible with streaming-capable, low-overhead on-device processing through a causal approximation of the analytic-signal stage. The present study should be interpreted as a methodological investigation of interpretable chest-based IMU respiratory instability sensing rather than as a clinical validation study. Further work is required to evaluate physiological specificity, robustness across heterogeneous cohorts, adaptive baseline strategies, and performance under broader real-world motion conditions.

Received: 27 February 2026 | Revised: 17 April 2026 | Accepted: 28 April 2026



Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.



Data Availability Statement

The reference implementation, validation scripts, baseline comparison modules, and reproducibility infrastructure described in this manuscript are publicly available at: https://github.com/dfeen87/Smartphone-Based-Chest-Monitoring
The repository includes the deterministic C++ core implementation, the Python validation pipeline, PhysioNet BIDMC integration scripts, controlled perturbation protocols, and computational profiling documentation.



Author Contribution Statement

Marcel Krüger: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization, Supervision, Project administration. Don Michael Feeney Jr.: Methodology, Software, Validation, Resources, Data curation, Writing – review & editing, Visualization.