Leveraging linked registry and electronic health records to examine long-term patient outcomes after peripheral vascular intervention"

利用链接的登记和电子健康记录来检查患者的长期结局， 外周血管介入 项目概要/摘要 外周动脉疾病（PAD）影响全球超过2亿人。外周血管介入 (PVI)是用于管理PAD的最常见程序。现有随机对照 PVI后患者结局的试验（RCT）和观察性研究的随访时间均有限， 长期数据收集的困难。此外，支架置入的治疗效果异质性（HTE） vs.目前，单独的经皮腔内血管成形术（PTA）还没有得到很好的理解， 效果修饰剂评估方法。真实世界数据（RWD），特别是与电子数据库链接的登记系统 健康数据（EHR）对于研究血管手术后的长期结果是有用的。然而，方法 用于处理多个数据源和分析非结构化文本数据的技术仍在不断发展。拟议 研究旨在解决肺静脉隔离手术后长期患者结局的现有证据缺口。这将 通过创新性地应用和完善数据链接、自然语言处理（NLP）和效果， 修改器评估方法。具体而言，该项目将链接注册表和EHR数据，以1）检查长期 支架置入后与单纯PTA相比的主要肢体不良事件以及评估治疗异质性 受患者特征的影响; 2）使用机器学习方法开发NLP管道，以分析 非结构化文本数据，并检查支架置入后与单独PTA相比的长期有效性终点，以及; 3） 为自然语言处理工具的部署确定可行性，并更新要求，以实现肺静脉隔离的长期效果 评估其他机构。为了支持研究活动和向独立过渡， 候选人将在奖励期间从事以下职业发展活动：1）获得 深入理解NLP和机器学习方法; 2）完善数据科学专业知识， EHR纳入医疗器械流行病学研究; 3）加强对当前和新型血管疾病的认识 疾病治疗; 4）发展和提高赠款写作和学术领导技能; 5）培训 负责任地进行研究。候选人将由一个专家小组指导， 在手术和器械结局研究、自然语言处理和机器学习方面的优势，以及 血管疾病和外科手术。拟议的职业发展和研究活动将发展 候选人的技能和专业知识，并导致R01级应用程序。候选人的长期目标是 成为一个独立的研究人员，专注于先进的多学科的开发和应用 在血管疾病领域评价手术和器械结果的方法，支持临床， 患者和监管决策。

Leveraging linked registry and electronic health records to examine long-term patient outcomes after peripheral vascular intervention Project Summary/Abstract Peripheral arterial disease (PAD) affects over 200 million people worldwide. Peripheral vascular interventions (PVI) are the most common procedures that are performed to manage PAD. Existing randomized controlled trials (RCTs) and observational studies of patient outcomes after PVIs all had limited follow-up lengths due to difficulties in long-term data collections. In addition, heterogeneity of treatment effect (HTE) for stent placement vs. percutaneous transluminal angioplasty (PTA) alone has not been well understood with the current approach of effect modifier assessment. Real-world data (RWD), particularly registries linked with electronic health data (EHR), are useful for studying long-term outcomes after vascular procedures. However, methods for working with multiple data sources and analyzing unstructured text data are still evolving. The proposed research aims to address current evidence gaps in long-term patient outcomes after PVI procedures. This will be facilitated by innovatively apply and refine data linkage, natural language processing (NLP), and effect modifier assessment methods. Specifically, this project will link registry and EHR data to 1) examine long-term major adverse limb events after stent placement vs. PTA alone as well as assess heterogeneity of treatment effect by patient characteristics; 2) develop an NLP pipeline with machine learning methods to analyze unstructured text data and examine long-term efficacy endpoints after stent placement vs. PTA alone, and; 3) establish feasibility and updating requirements for the deployment of the NLP tool for long-term PVI outcome assessment to other institutions. To support the research activities and the transition toward independence, the candidate will undertake the following career development activities during the award period: 1) gaining an in- depth understanding of NLP and machine learning methods; 2) refining data science expertise to integrate EHR into medical device epidemiologic research; 3) strengthening knowledge in current and novel vascular disease treatment; 4) developing and improving skills in grant writing and academic leadership; 5) training in responsible conduct of research. The candidate will be mentored by a team of experts with complementary strengths in surgical and device outcomes research, natural language processing and machine learning, and vascular disease and surgery. The proposed career development and research activities will develop the candidate's skillset and expertise and lead to an R01 level application. The candidate's long-term goal is to become an independent researcher focusing on the development and application of advanced multidisciplinary methods in the evaluation of surgical and device outcomes in the vascular disease area, supporting clinical, patient, and regulatory decision-making.