Automated detection and monitoring of health risks, such as COVID-19

全球COVID-19疫情期间的一个新趋势是政府和企业采用新技术，以跟踪和减轻病毒的传播和未来的健康风险。世界各地的行业都被建议实施积极的筛选协议，以帮助保护他们的员工。新技术必须专注于这些筛查程序的自动化和标准化，以帮助保护一线工作人员并提高整体筛查的可靠性。该项目与加拿大公司Eigen Innovations合作，该公司专门将数字化热成像数据解决方案集成到先进的工业和商业应用中，使用一系列最先进的成像技术开发一种用于自动检测和监测健康风险的连接工具。该团队由Eigen的解决方案工程师（博士），软件架构师和产品开发负责人以及UNB的Dubay博士（PI）和Pickard博士（PDF）组成。人工智能研究将被用来分析收集的图像数据，以预测个人的某些特征（例如，年龄和性别），通过分配虚拟识别标签来识别唯一的个体，以及检测和处理图像中的遮挡对象（例如，面罩或眼镜）。来自各种类型传感器的标记数据的可用性，结合经过训练的学习模型，将通过检查多个可见标准（如体温升高、皮疹和呼吸困难）来提高筛查的可靠性。这些训练的模型，以及其他考虑因素（例如，传感器校准、主动黑体和外部数据源），将用于通过允许调整与个人相关的独特标记标准来不断提高筛查可靠性。此外，拟议的健康监测工具将导致建立一个基于加拿大的健康数据库，并通过利用Eigen现有的数据收集和分析平台，合作研究人员和政府将能够更容易地跟踪和研究健康风险的传播。

An emerging trend during the global COVID-19 pandemic is the adoption of new technologies by governments and businesses in order to track and mitigate the spread of the virus and future health risks. Industries throughout the world are being advised to implement active screening protocols to help protect their employees. New technologies must focus on automating and standardizing these screening procedures to help protect front-line workers and improve overall screening reliability. This project partners with Eigen Innovations, a Canadian company that specializes in integrating digitized thermographic data solutions into advanced industrial and commercial applications, to use an array of state-of-the-art imaging technologies to develop a connected tool for the automated detection and monitoring of health risks. The team comprises of a solutions engineer (PhD), a software architect and a product development leader from Eigen , and Dr. Dubay (PI) and Dr. Pickard (PDF) from UNB. Artificial intelligence research will be leveraged to analyze the collected image data to predict certain traits of an individual (e.g., age and gender), identify unique individuals by assigning virtual identification tags, and to detect and handle occluding objects in the images (e.g., a face mask or glasses). The availability of labelled data from various types of sensors, combined with trained learning models, will be utilized to improving screening reliability by checking multiple visible criteria, such as elevated body temperature, skin rash, and difficulty breathing. These trained models, as well as other considerations (e.g., sensor calibration, active black bodies, and external data sources), will be used to continuously improve the screening reliability by allowing adaptation of the unique flagging criteria associated with an individual. Furthermore, the proposed health monitoring tool will result in the creation of a Canada-based health database, and by utilizing Eigen's existing data collection and analytics platform, collaborating researchers and governments will be able to more easily track and study the spread of health risks.