MindWatch: An Open-Source End-to-End Wearable Framework for Real-Time Physiological Stress Monitoring and Prediction

MindWatch demonstrates that a fully open, transparent hardware and software stack can match the feature depth of clinical-grade wearable stress monitors in sensor fusion, algorithmic transparency, and system latency. By implementing zero-allocation DSP, tonic/phasic EDA decomposition at the microcontroller edge, and an auditable XGBoost inference pipeline, the system achieves 98.2% three-class classification accuracy with a mean ROC-AUC of 0.9991 on the WESAD benchmark. The enforcement of strict train-inference parity and confidence-gated fallbacks ensures robust, low-latency performance in real-world deployment.

1. W. Boucsein; Electrodermal Activity, 2nd ed. Springer, New York, (2012). 2. Y. T. Can; B. Arnrich; C. Ersoy; Stress detection in daily life scenarios using smartphones and wearable sensors: A survey, J. Biomedical Informatics, 92, (2019). 3. G. Douzas; F. Bacao; SMOTE: Synthetic minority oversampling technique for imbalanced data classification — a revisit with new variants, Information Sciences, 627, 81–109 (2023). 4. M. Gjoreski; H. Gjoreski; M. Luštrek; M. Gams; Accurate kNN-based real-time prediction of stress on wristband data, Proc. UbiComp (2017). 5. J. Healey; R. W. Picard; Detecting stress during real-world driving tasks using physiological sensors, IEEE Trans. Intelligent Transportation Systems, 6(2), 156–166 (2005). 6. M. Karayem; B. Said; O. El-Telbany; Towards real-time stress detection using smartwatches and deep learning: A hybrid CNN-LSTM approach, Sensors, 23(3), 1421 (2023). 7. X. Li; H. Zhao; Deep learning for mental health monitoring via wearable biosensors: A systematic review, IEEE Reviews in Biomedical Engineering (2024). 8. A. Mondal; A. Ghosh; S. Dey; S. Mandal; A. K. Das; S. Dutta; MindWatch: Next-generation wearable framework for stress monitoring and prediction, Proc. IEMECON (2025), DOI: 10.1109/IEMECON69302.2025.11365881. 9. P. Schmidt; A. Reiss; R. Duerichen; C. Marberger; K. Van Laerhoven; WESAD: A multimodal dataset for wearable stress and affect detection, ACM Trans. Sensor Networks, 19(2), 1–27 (2018). 10. C. Setz; B. Arnrich; J. Schumm; R. La Marca; G. Tröster; U. Ehlert; Discriminating stress from cognitive load using a wearable EDA device, IEEE Trans. Information Technology in Biomedicine, 14, 410–417 (2010).