AI‐Assisted Multimodal Breath Sensing System with Semiconductive Polymers for Accurate Monitoring of Ammonia Biomarkers

Abstract

Breath ammonia is an essential biomarker for patients with many chronic illnesses, such as chronic kidney disease (CKD), chronic liver disease (CLD), urea cycle disorders (UCD), and hepatic encephalopathy. However, existing breath ammonia sensors fail to compensate for the impact of breath humidity and complex breathing motions associated with a human breath sample. Here, a multimodal breath sensing system is presented that integrates an ammonia sensor based on a thermally cleaved conjugated polymer, a humidity sensor based on reduced graphene oxide (rGO), and a breath dynamics sensor based on a 3D folded strain‐responsive mesostructure. The miniaturized construction and module‐based configuration offer flexible integration with a broad range of masks. Experimental results present the capabilities of the system in continuously detecting diagnostic ranges of breath ammonia under real, humid breath conditions with sufficient sensing accuracy and selectivity over 3 weeks. A machine‐learning algorithm based on K‐means clustering decodes multimodal signals collected from the breath sensor to differentiate between healthy and diseased breath concentrations of ammonia. The on‐body test highlights the operational simplicity and practicality of the system for noninvasively tracing ammonia biomarkers.