A Wearable NMES-LRR Device for Functional Screening of Lower-Limb Venous Outflow Restriction

Abstract

Deep vein thrombosis (DVT) is potentially life-threatening when embolization results in pulmonary embolism, yet first-line diagnosis using duplex ultrasound is equipment- and operator-dependent and is not well suited to frequent, at-home assessment. This study presents a wearable functional venous assessment system that integrates calf light-reflection rheography (LRR) with neuromuscular electrical stimulation (NMES). The device time-locks a repeatable calf-muscle-pump provocation (NMES) to dual-wavelength reflective optical acquisition (LRR) and streams event-marked data for analysis. LRR captures the associated venous emptying and refilling response. The system was evaluated in 32 healthy volunteers using a graded thigh-cuff protocol (0, 100, and 150 mmHg) as a controlled surrogate for increasing venous outflow restriction. For classification, recordings were labelled as Normal (0 mmHg) or DVT-like (100/150 mmHg cuff restriction). After quality control, 455 LRR recordings were retained and analyzed using a fixed subject-wise train/validation/test split (319/68/68) to prevent participant leakage. A feature-based support vector machine (SVM) was developed for high-sensitivity binary screening. Using a joined IR+RED feature representation and a validation-selected high-recall operating point, the SVM achieved a test accuracy of 0.9559, with sensitivity 0.9118 and specificity 1.0000. Because the reference labels are cuff-defined hemodynamic states, this study is positioned as feasibility evidence for functional screening/triage rather than clinical DVT diagnosis.