NSF Convergence Accelerator Track L: HEADLINE - HEAlth Diagnostic eLectronIc NosE

改善全球健康需要使用先进的诊断应用技术和健康监测系统，特别是以便携式设备的形式用于医疗点应用。随着最近的科学研究显示人体/呼吸释放的气体/挥发性有机化合物（VOCs）与健康状况之间的相关性，本项目旨在设计一种用于非侵入性诊断应用的混合和模块化电子鼻（e-Nose）系统。通过一项国际合作，来自美国两所大学、瑞典一所大学和一家瑞典公司的研究人员共同努力，将他们对气体传感和新型人工智能(AI)/机器学习（ML）模型相关的各种技术的研究融合在一起，设计出一种具有全球影响潜力的先进医疗工具。目前基于金属氧化物半导体的电子鼻技术在检测和浓度测量需要作为健康状态指标进行监测的关键化学品方面存在局限性。更重要的是，这些传感器的低选择性对测量的准确性产生了负面影响。这个初始项目将一系列传感器技术与新型传感材料结合在一起，形成一个混合电子鼻，可以选择性地检测大范围的挥发性有机化合物。这种混合传感平台需要不同的AI/ML模型，这些模型可以处理来自不同传感器的数据，以识别目标气体和挥发性有机化合物的组成，并具有医疗应用所需的足够高的准确性和精度。因此，最终目标是提供混合气体/VOC传感器平台和AI/ML模型，以解释主要用于医疗诊断应用的各种传感器的信号。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Improving global health demands using advanced technologies for diagnostic applications and health monitoring systems, particularly in the form of portable devices for point-of-care applications. With the recent scientific studies showing correlations between gases/volatile organic compounds (VOCs) released from a person’s body/breath and their health status, this project aims to design a hybrid and modular electronic nose (e-Nose) system for non-invasive diagnostic applications. Through an international collaboration, researchers from two universities in the US, one university in Sweden, and a Swedish company work together to converge their research on various technologies related to gas sensing and novel artificial intelligence (AI)/machine learning (ML) models to design an advanced medical tool with the potential of a global impact. The current e-Nose technology based on metal-oxide semiconductors has limitations in the detection and concentration measurements of critical chemicals that needed to be monitored as health status indicators. More importantly, the poor selectivity of those sensors negatively impacts the accuracy of the measurements. This initial project brings together a range of sensor technologies, alongside novel sensing materials, into a hybrid e-Nose that can selectively detect a large range of VOCs. Such a hybrid sensing platform requires different AI/ML models that can process the data from different sensors to identify the composition of the target gases and VOCs with a high enough accuracy and precision needed for medical applications. Hence, the eventual goal is to deliver a hybrid gas/VOC sensor platform and AI/ML models to interpret the signals from various sensors mainly for medical diagnostic applications.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.