I-Corps: Postoperative Risk Prediction for Heart Failure Patients

I-Corps项目更广泛的影响/商业潜力是开发一个嵌入新型机器学习算法的风险评估平台，以预测心脏治疗和手术后心力衰竭患者的死亡率和不良事件。该技术旨在：1)解决医疗数据中普遍存在的不确定性和数据不平衡问题；2)提取与术后风险相关的重要特征和生物标志物；3)开发用于超声图像分析的新工具，以预测不良事件的概率。目前，心脏护理领域的医疗保健专业人员还没有成熟的技术来分析多个数据源，以进行心脏治疗前后的风险评估。提出的技术可以帮助临床医生利用先进的机器学习工具进行治疗结果评估和临床决策。该技术还可以促进数据科学方法和计算机模型在其他临床应用中的应用。该项目旨在实现准确的风险预测，增强对术后治疗的信心，提高心脏护理质量，最终使心脏病患者受益。I-Corps项目的基础是开发一种软件应用程序，用于预测心脏移植或左心室辅助装置（LVAD）植入等心脏手术后终末期心力衰竭患者的术后风险。该软件应用程序有一个后端服务器，该服务器使用以前的心脏移植和左心室辅助器接受者的电子健康记录（EHR）训练的新型机器学习算法。使用定制的机器学习算法和统计风险模型进行风险预测。新的聚类技术用于识别具有相似特征的患者组，重要的临床变量，以及数据中存在的噪声和异常值。该软件还提供了一个易于使用的界面，为客户提供临床有用的信息和与个体患者风险预测相关的风险因素的可视化。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。

The broader impact/commercial potential of this I-Corps project is the development of a risk assessment platform embedded with novel machine learning algorithms to predict mortality and adverse events in heart failure patients after cardiac treatments and surgeries. The technology seeks to: 1) address uncertainty and data imbalance issues prevalent in healthcare data, 2) extract important features and biomarkers relevant to post-operative risks, and 3) develop new tools for ultrasound image analysis to predict the probability of adverse events. There is currently no established technology for healthcare professionals in the cardiac care field to analyze multiple data sources for risk assessment pre- and post-cardiac treatment. The proposed technology may help clinicians leverage advanced machine learning tools for treatment outcomes assessment and clinical decision-making. The technology may also promote the use of data science approaches and computer models in other clinical applications. The project seeks to enable accurate risk prediction and enhance confidence in post-operative treatment, improving the quality of cardiac care, and ultimately benefitting cardiac patients.This I-Corps project is based on the development of a software application to predict post-operative risk for end-stage heart failure patients after cardiac surgeries such as heart transplantation or left ventricular assist device (LVAD) implantation. The software application has a back-end server with novel machine learning algorithms trained using electronic health records (EHR) from previous heart transplant and LVAD recipients. Risk prediction is performed using customized machine learning algorithms and statistical risk models. Novel clustering techniques are used to identify groups of patients with similar characteristics, clinical variables of importance, as well as noise and outliers present in the data. The software also offers an easy-to-use interface that provides clients with clinically useful information and a visualization of risk factors linked to the risk prediction for an individual patient.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.