Advanced Sensor Research Прикладные компьютерные науки

An Embedded Ambiguity‐Aware Interpretation System for Multimodal Urinary VOC and Glucose Sensing

Mario Soto Martinez, Yurii Tsyban, Aazir Ashikeen, Samanta Castro Rodriguez, Lavita Nuraviana Rizalputri, Khaled Nabil Salama
Wiley · 2026-11-01 · DOI 10.1002/adsr.70223

Аннотация

ABSTRACT Single‐modality metabolic measurements can be analytically valid yet physiologically ambiguous because the same readout may support different interpretations. Here, we present an ambiguity‐aware multimodal sensing system that acquires urine headspace volatile organic compounds (VOCs), urinary glucose, and systemic glucose within a coordinated measurement cycle and interprets their agreement or disagreement on‐device. The platform combines a six‐channel metal‐oxide gas sensor array with embedded electrochemical glucose sensing and a transparent rule‐based consistency framework that preserves each modality as an independent biological readout. The VOC subsystem discriminated engineered healthy‐like and diabetic‐like headspace conditions, while the electrochemical subsystem showed strong agreement with a laboratory reference in urine (R 2 = 0.9605, r = 0.9801) and serum (R 2 = 0.9918, r = 0.9959). Across five controlled metabolic‐state models, the framework reinforced concordant patterns, resolved a renal‐glucosuria‐like discordance in which urinary glucose was elevated while systemic glucose and the VOC readout remained normal, and retained ambiguity when the measured combination did not satisfy a predefined interpretation rule. These results demonstrate a device‐level strategy for transparent, ambiguity‐aware multimodal interpretation under controlled physiological models.

Читать у издателя →