A predictive analytics framework for undesired outcomes using vital signs data analysis

这项研究计划的长期目标是开发决策支持工具，为临床医生和患者提供预测性信息，以便个性化治疗和改善健康状况。这些工具将：1)从临床、个人和社区环境中的电子仓库收集和集成数据；2)将智能算法应用于这些数据以检测潜在的结果；3)将这些结果信息分发到电子健康记录或智能设备(供临床医生使用)和智能手机应用程序(供患者使用)。短期目标是开发和测试预测分析框架，分析和集成复杂的机器辅助环境中存在的大量数字信号(包括数据库和实时信号)，以：a)改善医疗设备之间的数据访问和交换，同时确保安全性/隐私/机密性；以及b)支持算法的实施和评估，以使用生命体征、医疗设备和患者数据预测不受欢迎的临床事件。*研究的重要性：通过促进数据交换，应用智能算法从复杂的生理和环境数据中预测结果，并使用有效的手段将结果可视化和交流，我们可以为实现医疗保健所需工具的现代化做出重要贡献。在不希望发生的临床事件(例如，ICU中的心脏骤停或手术后并发症)之前及早发现危重患者的病情恶化就是这样一种应用，因为它允许重新分配护理优先级，并代表着改善结果和降低成本的巨大机会。*预期结果：我们将与医疗保健组织、临床医生、数据科学家、研究人员和潜在的行业合作伙伴合作，开发预测分析框架的两个互补组件：数据交换组件，使用‘物联网’技术获取并整合实时和历史数据；以及结果预测建模和处理管道，它提取数据并将其转换为可操作的信息。*加拿大研究领域和利益：生物医学和计算机工程方面的创新是推进加拿大医疗保健不可或缺的一部分。这项研究计划将通过提供安全、可信和简单的医疗数据交换框架，减少机器对机器通信和此类数据的有意义使用方面的障碍。这一框架将广泛适用于研究和开发环境。它将支持包括预测分析模型在内的智能决策支持系统的发展，这些系统将支持减少虚假警报、闭环监测/辅助系统和早期预警系统，以检测和预防不良的健康后果。

The long-term goal of this research program is to develop decision support tools that provide predictive information to both clinicians and patients in order to personalize treatments and improve wellness. These tools will: 1) gather and integrate data from electronic warehouses in clinical, personal, and community settings; 2) apply smart algorithms to this data to detect potential outcomes; and 3) distribute this outcome information to electronic health records or smart devices (for clinicians) and smartphone applications (for patients). The short-term objective is to develop and test predictive analytics frameworks that analyze and integrate the multitude of digital signals (both data-banked and real-time) that exist in complex machine-assisted environments to: a) improve data access and exchange between medical devices, while ensuring security/privacy/confidentiality; and b) support the implementation and evaluation of algorithms to predict undesired clinical events using vital signs, medical device, and patient data. ******Importance of research: By facilitating data exchange, applying smart algorithms to predict outcomes from complex physiological and environmental data, and using effective means to visualize and communicate the results, we can make an important contribution to modernizing the tools we need for healthcare. The early detection of the deterioration of critically ill patients before undesired clinical events (e.g., cardiac arrest in the ICU or complications after surgery) is such an application, as it allows for re-assigning priority of care and represents an enormous opportunity to improve outcomes and reduce cost.******Anticipated Outcomes: In collaboration with healthcare organizations, clinicians, data scientists, researchers, and potential industry partners, we will develop two complementary components of the predictive analytics framework: a data exchange component, which obtains and integrates real-time and historic data, using ‘Internet of Things' technologies; and an outcome prediction modeling and processing pipeline, which extracts the data and transforms them into actionable information.******Benefit for Research Field & Canada: Innovation in biomedical and computer engineering is integral to advancing Canadian healthcare. This research program will mitigate barriers in machine-to-machine communication and the meaningful use of such data by providing a framework for safe, trusted, and simple medical data exchange. This framework will be widely applicable in research and development settings. It will underpin the development of smart decision support systems, including predictive analytics models, which will support the reduction of false alarms, closed-loop monitoring/assistive systems and early-warning systems to detect and prevent undesirable health outcomes.