Key result
Multimodal CNN-LSTM fusion of ECG and autonomic signals detects stress with ~97% accuracy.
Why the study?
Stress is a major contributor to physical and mental health disorders, creating a need for reliable and interpretable physiological stress detection systems.
Does an interpretable multimodal deep learning framework integrating ECG, electrodermal activity, and respiration signals improve binary physiological stress detection compared to single-modality and classical machine learning models?
Does an interpretable multimodal deep learning framework integrating ECG, electrodermal activity, and respiration signals improve binary physiological stress detection compared to single-modality and classical machine learning models?
Combining ECG, electrodermal activity, and respiration in a mid-fusion CNN-LSTM architecture enables highly accurate (97.3% LOSO accuracy) and physiologically interpretable stress detection that focuses on clinically meaningful features such as ECG R-peaks and phasic EDA responses.
Multimodal fusion advances stress detection models; hypothesis-generating and should not yet change clinical practice.
Stress is a major contributor to physical and mental health disorders, underscoring the need for reliable and interpretable detection systems. We present an interpretable multimodal deep learning framework that integrates electrocardiogram (ECG), electrodermal activity (EDA), and respiration signals using modality-specific neural architectures complemented by feature-based classical models. To ensure robustness and generalizability, the framework was evaluated under both subject-independent and subject-dependent protocols. Results show that multimodal fusion consistently outperformed single modalities, achieving 94.8% accuracy, 93.9% F1-score, and 0.97 AUC under subject-independent validation, compared with 86–91% accuracy for individual signals. In subject-dependent experiments, performance reached 97.2% accuracy and 96.8% F1-score. Comparative analysis with recent state-of-the-art methods demonstrated the competitiveness of our approach, while ablation studies confirmed the contribution of each modality and fusion strategy. Statistical significance testing with ( p < 0.05) further validated the reliability of these gains. Importantly, interpretability analyses using Grad-CAM and attention visualization provided physiological insights, revealing that the model emphasizes clinically meaningful regions such as ECG R-peaks and phasic EDA fluctuations. By uniting high performance, robustness, and interpretability, this work advances stress detection research and offers a physiologically grounded framework for continuous mental health monitoring.
No takes yet. Share an insight, caveat, or question.
Tibermacine et al. (2026) studied Physiological stress (n=15). Multimodal CNN-LSTM with mid-fusion integrating ECG, EDA, and respiration signals vs. Single-modality baselines (ECG-only, EDA-only, Respiration-only) and alternative fusion/classical machine learning models was evaluated on Binary stress versus non-stress classification accuracy under Leave-One-Subject-Out (LOSO) subject-independent cross-validation (p=<0.05). Multimodal CNN-LSTM mid-fusion of ECG, EDA, and respiration achieved 97.3% accuracy and 0.992 AUC in subject-independent stress detection, outperforming single modalities (p<0.05).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: