Improving Cardiovascular Drug Safety With Automated Bleeding Classification

项目总结 房颤(AF)的治疗通常包括口服抗凝剂(OAC)预防中风的药物治疗。 然而，出血是这些药物的常见并发症，影响多达四分之一的患者。《中心》 对于Medicare和Medicaid Services，最近将与OAC相关的药物安全列为关键质量指标。 然而，目前还没有方法可以准确地确定大量人群中的出血事件和严重程度。 现有的方法使用缺乏敏感度和临床细节的诊断代码，或手动检查图表，这 不能在大量人口中实施。拟议的研究旨在通过以下方式解决这一知识差距 应用基于自然语言处理(NLP)的方法识别出血事件并对严重程度进行分类 在一个真实的AF人群中。这些工具将在另一家机构接受治疗的患者中进行验证，以确保 不同提供商设置的重现性。此外，我们将应用出血分类工具来评估 出血严重程度与死亡率之间的关系。沙阿博士是一位崭露头角的年轻研究员，他的职业生涯 发展计划的重点是获得生物医学信息学技能，以准确识别和 减少对病人的伤害。她的培训计划侧重于学习自然语言处理的核心能力， 目标是将电子病历中的丰富数据转化为有用的知识。她会的 结合来自知名专家的指导和有针对性的课程工作，获得生物医学方面的技能 信息学、数据科学、高级分析方法和研究领导力。完成这些研究 培训目标将通过以下方式为未来的R01提案创建平台：(I)使安全重点放在比较 房颤的有效性研究(II)为确定其他心血管疾病的出血并发症奠定基础 以及(Iii)培养一套能够领导多学科研究团队的技能。通过这件事 职业发展计划，沙阿博士将以她以前在临床心脏病学和研究方面的培训为基础 并为高影响力的研究事业奠定了坚实的基础。

PROJECT SUMMARY Atrial fibrillation (AF) treatment often includes drug therapy with oral anticoagulants (OAC) to prevent stroke. Bleeding, however, is a common complication of these drugs, affecting up to one in four patients. The Center for Medicare and Medicaid Services recently prioritized OAC-related drug safety as a key quality measure. Currently, however, no method exists to accurately identify bleeding events and severity in large populations. Prior methods use diagnoses codes, which lack sensitivity and clinical detail, or manual chart review, which cannot be implemented in large populations. The proposed research aims address this knowledge gap by applying a natural language processing (NLP)-based approach to identify bleeding events and classify severity in a real-world AF population. The tools will be validated in patients treated at a different institution, to ensure reproducibility across provider settings. In addition, we will apply the bleeding classification tool to evaluate the association between bleeding severity and mortality. Dr. Shah is an emerging young investigator whose career development plan is focused on acquiring the biomedical informatics skills to needed to accurately identify and reduce patient harm. Her training plan focuses on learning core competencies in natural language processing, with the goal of turning the wealth of data in the electronic medical record into useable knowledge. She will combine mentorship from established experts and targeted coursework to acquire skills in biomedical informatics, data science, advanced analytic methods, and research leadership. Completion of these research and training aims will create a platform for future R01 proposals by: (i) enabling safety focused comparative effectiveness research in AF (ii) setting the stage to identify bleeding complications in other cardiovascular diseases and (iii) developing a skill set that allows leadership of a multidisciplinary research team. Through this career development plan, Dr. Shah will build upon her prior training in clinical cardiology and research methodology and lay a strong foundation for a high impact research career.