When Clean Signals Are Not Enough: Detecting Structural Ambiguity for Safe Wearable Stress Classification

Saba A. Farahani; Hung Cao; Amir M. Rahmani

cs.AIAug 19, 2026

When Clean Signals Are Not Enough: Detecting Structural Ambiguity for Safe Wearable Stress Classification

Authors: Saba A. Farahani, Hung Cao, Amir M. Rahmani

Organizations: University of California, Irvine Irvine, CA, USA

Abstract

Wearable stress classifiers can achieve strong average performance while failing completely for a particular individual. On WESAD, a Random Forest reaches 93.0% mean accuracy yet yields F1 = 0 for Subject 14, whose cross-signal coupling weakens near stress onset. We call this structural ambiguity: individually plausible physiological channels form an inter-signal pattern that is poorly supported by the person's non-stress reference. We introduce the Individual Conformal Coupling Monitor (ICCM), a lightweight and transparent pre-inference monitor that quantifies subject-specific coupling divergence and routes each window to classify, defer, or abstain without retraining the downstream classifier. Across WESAD (N = 15) and Stress-Predict (N = 35), full-cohort Pearson associations between ambiguity and accuracy are negative (r = -0.607, p = 0.016; r = -0.412, p = 0.014). Robustness analyses temper this finding: rank correlations are not significant, and the WESAD association disappears when Subject 14 is removed. ICCM changes false-positive counts from 29 to 27 and 94 to 92, although neither paired change is significant. It withholds 3 of Subject 14's 21 stress windows but does not repair the missed-stress failure. These results position ICCM as an interpretable signal of unsupported physiology and individual failure, rather than a stand-alone safety guarantee.

Figures & tables

Fig. 1: Aggregate accuracy conceals Subject 14’s missed-stress failure ( $F1=0$ ). EDA and BVP decouple near stress onset; residual artifacts cannot be excluded.

Fig. 2: ICCM system architecture. Wearable sensor streams (BVP, EDA, TEMP) are passed to the Coupling Divergence Engine within the Orchestrator. During offline calibration, the engine computes a subject-specific hybrid coupling baseline $\mathbf{v}^{0}$ and baseline distances $\{D_{1},\ldots,D_{k}\}$ from resting-state windows. At inference time, each 60-second window is evaluated using a hybrid coupling vector $\mathbf{v}(t)$ combining Pearson correlation, max-lag cross-correlation, and Granger-style directed coupling, converted to an empirical conformal-style rank score $p(t)$ , and routed through a 3-Zone Safety Gate: Zone 1 ( $p\geq\alpha$ ) passes to the classifier, Zone 2 ( $p\approx\alpha$ ) defers to the next window, and Zone 3 ( $p<\alpha/2$ ) triggers abstention. The term “safe” denotes the system objective, not a clinical guarantee.

	WESAD	Stress-Predict
Subjects	15	35
Mean accuracy / F1	0.930 / 0.799	0.739 / 0.154
Pearson $r$ ( $p$ )	$-0.607$ (.016)	$-0.412$ (.014)
Spearman $\rho$ ( $p$ )	0.016 (.955)	$-0.300$ (.080)
Pearson without S14	0.185 (.526)	–
FP: model / ICCM	29 / 27	94 / 92

TABLE I: Robustness and selective performance. Covered metrics condition on windows receiving a prediction.

Fig. 3: Full-cohort Pearson associations between structural ambiguity and LOSO accuracy. WESAD is high-leverage: excluding Subject 14 gives $r=0.185$ ( $p=0.526$ ), and Spearman $\rho=0.016$ ( $p=0.955$ ). Stress-Predict Spearman $\rho=-0.300$ ( $p=0.080$ ).

Explore similar work

ReliaGate: Reliability Routing for Low-Stakes Wearable Stress Prediction
Aug 16, 2026Jaden Moon, Yu Wu, Arvind Pillai +1Selective Prediction Wearable Health Monitoring
Dartmouth College Hanover, New Hampshire, USA
State-specific respiratory signatures for affective and stress recognition: Interpretable respiratory markers, autocorrelation lags, and compact CNN models
Jun 25, 2026Andrei Velichko, Mehmet Tahir HuyutWearable Sensing Convolutional Neural Networks
Institute of Physics and Technology, Petrozavodsk State University, Petrozavodsk, Russia · Department of Biostatistics and Medical Informatics, Faculty of Medicine, Erzincan Binali Yildirim University, Erzincan, Turkey
Stress Detection Using Wearable Physiological and Sociometric Sensors
Apr 14, 2026Oscar Martinez Mozos, Virginia Sandulescu, Sally Andrews +4Multimodal Sensor Fusion Wearable Sensing
Dept. of Electronics, Computer Technology and Projects, Polytechnic University of Cartagena, Plaza del Hospital, n1, 30202, Cartagena, Spain · Faculty of Automatic Control and Computer Science, Politehnica University of Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania · Division of Psychology, Nottingham Trent University, Burton Street, Nottingham, NG1 4BU, UK +2

Aug 16, 2026J/K move · Enter open · S save

ReliaGate: Reliability Routing for Low-Stakes Wearable Stress Prediction

Jaden Moon, Yu Wu, Arvind Pillai, Andrew Campbell

Dartmouth College Hanover, New Hampshire, USA

Selective Prediction Wearable Health Monitoring Wearable Sensing

Open paper AskPDF arXiv

We study when a wearable stress system should surface a prediction rather than change it. In low-stakes reflection and summary settings, aggregate accuracy is insufficient because withholding can reduce error while leaving some people with little or no information. We formulate fixed-label reliability routing: after a locked classifier emits a protocol-defined stress/non-stress label, a post-hoc gate surfaces that unchanged label or withholds it as unavailable. ReliaGate assembles established confidence, signal-quality/trust, agreement, train-standardized atypicality, and train-fitted geometry cues into a post-hoc correctness score. We evaluate four wearable datasets using subject-disjoint folds, validation-selected routing, paired held-out-subject intervals, and pooled and per-subject analyses. WESAD point estimates favored ReliaGate, UBFC-Phys primary coverage/risk intervals favored ReliaGate, and E4 checks were mixed. ReliaGate provides an operational framework for studying surfaced-label error, output availability, and accepted-output distribution across subjects, without revising labels or providing clinical or finite-sample risk guarantees.

Abstract

Explore similar work

ReliaGate: Reliability Routing for Low-Stakes Wearable Stress Prediction

State-specific respiratory signatures for affective and stress recognition: Interpretable respiratory markers, autocorrelation lags, and compact CNN models

Stress Detection Using Wearable Physiological and Sociometric Sensors