🤖 AI Summary
This study addresses the challenge of effectively detecting pre-stroke neurovascular instability (NVI) using conventional unimodal wearable devices. To this end, we propose Melaguard, a lightweight multimodal Transformer model that establishes the first edge-computing-oriented physiological signal fusion framework. Melaguard integrates heart rate variability, peripheral perfusion index, blood oxygen saturation, and bilateral phase coherence to generate a composite NVI score. Built upon a 1.2M-parameter Transformer-lite architecture with four-head self-attention, the model enables real-time inference on resource-constrained edge devices such as Cortex-M4 microcontrollers, achieving a worst-case execution time of ≤4 ms. Evaluated on the PhysioNet CVES cohort, Melaguard attains an AUC of 0.755, significantly outperforming LSTM, SVM, and random forest baselines. Notably, PPG-derived metrics exhibit strong correlation with ECG gold standards (r > 0.69), and their morphological features achieve an AUC of 0.923 for cerebrovascular disease classification on an independent dataset, offering an efficient and viable solution for community-level stroke risk screening.
📝 Abstract
We propose Melaguard, a multimodal ML framework (Transformer-lite, 1.2M parameters, 4-head self-attention) for detecting neurovascular instability (NVI) from wearable-compatible physiological signals prior to structural stroke pathology. The model fuses heart rate variability (HRV), peripheral perfusion index, SpO2, and bilateral phase coherence into a composite NVI Score, designed for edge inference (WCET <=4 ms on Cortex-M4). NVI - the pre-structural dysregulation of cerebrovascular autoregulation preceding overt stroke - remains undetectable by existing single-modality wearables. With 12.2 million incident strokes annually, continuous multimodal physiological monitoring offers a practical path to community-scale screening. Three-stage independent validation: (1) synthetic benchmark (n=10,000), AUC=0.88 [0.83-0.92]; (2) clinical cohort PhysioNet CVES (n=172; 84 stroke, 88 control) - Transformer-lite achieves AUC=0.755 [0.630-0.778], outperforming LSTM (0.643), Random Forest (0.665), SVM (0.472); HRV-SDNN discriminates stroke (p=0.011); (3) PPG pipeline PhysioNet BIDMC (n=53) -- pulse rate r=0.748 and HRV surrogate r=0.690 vs. ECG ground truth. Cross-modality validation on PPG-BP (n=219) confirms PPG morphology classifies cerebrovascular disease at AUC=0.923 [0.869-0.968]. Multimodal fusion consistently outperforms single-modality baselines. Code: https://github.com/ClevixLab/Melaguard