Sensor Hardware and Intelligent Tools for Assessing the Health Effects of Heat Exposure

环境温度上升的趋势不断加剧，使农业和农村人口等边缘化群体 经常暴露于高温潮湿环境的工人，急性健康风险增加 热暴露的影响，特别是热相关疾病（HRI）和脱水。HRI和脱水是 特别阴险，因为他们可以迅速从中度不适进展到混乱和受损 因此，受影响的工人寻求必要的医疗照顾的能力。热暴露 引发了多种生理反应模式因此，多模态传感器和计算算法， 来自这些传感器集成数据流是必要的，以更好地了解热暴露如何导致急性 健康影响。我们建议开发一种包含多个纳米级传感器的无线可穿戴装置， 实时整合关键生理信号，并能够预测和生成有关不良反应的警告。 事件也是实时的。该项目围绕三个目标组织：1）开发一种柔软的、基于纳米膜的， 用于监测生理状况的可穿戴生物手表，包括皮肤温度、皮肤水合作用、心率， 呼吸频率、血氧饱和度、运动和心电图，沿着远程蓝牙 发射器; 2）开发用于新颖计算算法的多传感器多任务学习框架， 机器学习，实时整合传感器数据，提取特征，并考虑内部和内部 模态相关性为了开发与HRI症状相关的生物标志物的预测模型， 脱水，和脱水的生物标志物;和3）确定该技术的性能时，使用 野外作业条件下的户外作业人员。我们将在第1年和第2年对连续的原型进行实地测试， 户外工作者受雇于亚特兰大地铁穿生物手表和其他齿轮，我们已经在我们的 对职业性热暴露的生理反应的其他研究。我们将在以下条件下进一步测试生物贴片 更崎岖的条件下，招募佛罗里达农业工人为同一协议的评估，在第3年 和4.这项工作的重要性在于，我们将能够确定和整合 热暴露的生理反应，以识别HRI症状发作前的不良反应。创新 这项工作是将纳米工程和计算算法结合起来，以跟踪、监测和预测 工人对极端高温条件的反应。我们已经组建了一个跨学科的团队， 来自埃默里大学和格鲁吉亚理工学院的研究人员在纳米工程，计算机科学， 以及农业工人热生理学的实地研究。该项目的发现将导致未来 使用生物贴片将数据发送到工人、伙伴和/或机组人员持有的Android设备的干预研究 主任，以确定是否使用这种技术可以减少与职业热暴露有关的发病率。 HRI检测和警报系统的实际应用将依赖于来自 建议的研究，以生成警报，如果工人被预测有不利的结果。

Escalating trends of increasing environmental temperatures place marginalized populations such as agricultural workers, who have routine occupational exposure to hot, humid environments, at increased risk for acute health effects of heat exposure, in particular heat-related illness (HRI) and dehydration. HRI and dehydration are particularly insidious as they can quickly progress from moderate discomfort to confusion and impaired judgement, thereby diminishing the affected worker’s ability to seek necessary medical attention. Heat exposure invokes multiple modes of physiologic response. Thus, multi-modal sensors and computational algorithms to integrate data streams from these sensors are necessary to better understand how heat exposure leads to acute health effects. We propose to develop a wireless wearable unit containing multiple nanoscale sensors that can integrate key physiological signals in real-time and with the ability to predict and generate warning about adverse events also in real-time. This project is organized around three aims: 1) develop a soft, nanomembrane-based, wearable biopatch for monitoring physiological conditions, including skin temperature, skin hydration, heart rate, respiratory rate, blood oxygen saturation, motion, and electrocardiogram, along with a long-range Bluetooth transmitter; 2) develop a multi-sensor multi-task learning framework for novel computational algorithms using machine learning to integrate sensor data in real-time, extracting features and accounting for inter- and intra- modality correlations in order to develop predictive models for biomarkers associated with HRI symptoms, dehydration, and biomarkers of dehydration; and 3) determine performance of this technology when used by outdoor workers in field conditions. We will field test successive prototypes throughout years 1 and 2, recruiting outdoor workers employed in metro Atlanta to wear both a biopatch and other gear that we have used in our other studies of physiological responses to occupational heat exposure. We will further test the biopatch under more rugged conditions, recruiting Florida agricultural workers for the same protocol for evaluation in years 3 and 4. This work is significant in that we will be able to determine and integrate the multiple modalities of physiologic response to heat exposure to recognize adverse effects before HRI symptom onset. Innovation of the work lies in the integration of nanoengineering and computational algorithms to track, monitor, and predict worker response to extreme heat conditions. We have assembled a stellar interdisciplinary team of experienced investigators from Emory University and Georgia Tech with expertise in nanoengineering, computer science, and field studies of heat physiology in agricultural workers. Findings from this project will lead to a future intervention study using the biopatch to send data to Android devices held by the worker, a buddy, and/or crew chief to determine if using this technology can reduce morbidity associated with occupational heat exposure. Real-world application of the HRI detection and alert system will rely on machine learning algorithms from the proposed study to generate alerts if the worker is predicted to have an adverse outcome.