I-Corps: Translation Potential of a Machine Learning Risk Stratification Tool for Venous Thromboembolism

I-Corps项目更广泛的影响是开发了一种风险分层软件工具，用于预测当静脉血栓形成时发生的情况，即静脉血栓栓塞（VTE）。静脉血栓栓塞是全球血管相关死亡的第三大原因。预防静脉血栓栓塞的措施已被证明对降低发病率和死亡率都是有效的。尽管有可能预防，但静脉血栓栓塞的诊断仍然是一项复杂的临床挑战，主要归因于涉及的多种危险因素。这种机器学习技术具有管理和解释预测模型中通常涉及的一系列变量的潜力，并可能影响静脉血栓栓塞预防策略和临床指南，从而促进更好的患者预后。这个I-Corps项目利用体验式学习和对行业生态系统的第一手调查来评估该技术的翻译潜力。该解决方案基于一种预测算法的开发，该算法使用机器学习技术，旨在识别有静脉血栓栓塞（VTE）风险的个体。该技术可用于识别通过传统统计方法不易明显的模式和相关性。该模型可以提供更准确的静脉血栓栓塞风险评估。通过对包含16,000多例患者记录的庞大数据集的分析，确定了几个关键的预测风险因素。这些因素包括患者的年龄、创伤和治疗之间的间隔以及吸烟状况。利用这些见解，我们设计了多个预测模型，并对其他数据集进行了严格的测试。最有效的模型显示出预测静脉血栓栓塞发生率的准确率为90%，显示出其通过促进早期干预和个性化护理策略显着改善患者预后的潜力。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 of this I-Corps project is the development of a risk stratification software tool to predict a condition that occurs when a blood clot forms in a vein called venous thromboembolism (VTE). Venous thromboembolism ranks as the third leading cause of vascular-related deaths worldwide. Actions taken to prevent VTE have proven effective in reducing both morbidity and mortality. Despite being potentially preventable, diagnosing VTE remains a complex clinical challenge, largely attributable to the multitude of risk factors involved. This machine learning technology has the potential to manage and interpret the array of variables typically involved in predictive models and may influence VTE prevention strategies and clinical guidelines, promoting better patient outcomes.This I-Corps project utilizes experiential learning coupled with a first-hand investigation of the industry ecosystem to assess the translation potential of the technology. The solution is based on the development of a prediction algorithm using machine learning techniques designed to identify individuals at risk of venous thromboembolisms (VTE). The technology may be used to identify patterns and correlations that are not readily apparent through conventional statistical methods. This model may provide more accurate risk assessments for VTE. Through the analysis of an expansive dataset encompassing over 16,000 patient records, several key predictive risk factors were identified. These include patient age, the interval between trauma and treatment, and smoking status. Leveraging these insights, multiple predictive models were engineered and subjected to rigorous testing from other datasets. The most effective model demonstrated an ability to forecast VTE occurrences with an accuracy rate of 90%, showcasing its potential to significantly improve patient outcomes by facilitating early intervention and personalized care strategies.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.