Assessing performance of a Hepatitis C Emergency Department (HepC-END) Screening Tool

评估丙型肝炎急诊科 (HepC-END) 筛查工具的性能

• 批准号: 10754614
• 负责人: Haesuk Park
• 金额: $ 70.24万
• 依托单位: UNIVERSITY OF FLORIDA
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-15 至 2028-07-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10754614
• 关键词: Accident and Emergency department; Adult; Algorithms; Antiviral Agents; Antiviral Therapy; Artificial Intelligence; Caring; Chronic Hepatitis C; Clinical; Clinical Data; Clinical Research; Continuity of Patient Care; Data; Development; Diagnosis; Diagnostic; Disease Progression; Educational workshop; Effectiveness; Electronic Health Record; Emergency department screening; Exploration, Preparation, Implementation, and Sustainment; Feedback; Florida; Funding; Goals; HCV screening; HIV; Health; Healthcare Systems; Hepatitis; Hepatitis C; Hepatitis C Therapy; Hepatitis C Transmission; Hepatitis C virus; Homelessness; Individual; Infection; Intervention; Interview; Link; Liver; Machine Learning; Methods; Morbidity - disease rate; National Institute of Drug Abuse; Natural Language Processing; Notification; Nurses; Patient Self-Report; Patients; Performance; Persons; Physicians; Play; Positioning Attribute; Proctor framework; Provider; Public Health; Questionnaires; Recommendation; Research; Risk; Risk Assessment; Risk Behaviors; Risk Factors; Role; Route; Screening procedure; Social Behavior; Substance Use Disorder; Testing; Time; Translating; United States; Universities; Venous blood sampling; Visit; Work; clinical practice; co-infection; cost; cost effective; cost effectiveness; design; economic outcome; electronic health record system; high risk; high risk behavior; implementation framework; implementation outcomes; improved; injection drug use; innovation; machine learning algorithm; machine learning prediction algorithm; mortality; opioid epidemic; prediction algorithm; predictive modeling; prototype; recurrent neural network; risk stratification; screening; screening program; structured data; tool; transmission process; usability; viral transmission

PROJECT SUMMARY/ABSTRACT Hepatitis C virus (HCV) infection has markedly increased in the United States, primarily resulting from injection drug use (IDU) associated with the ongoing opioid epidemic. Furthermore, >50% of 3.2 million individuals with chronic HCV remain undiagnosed, leading to significant morbidity and mortality despite the availability of effective direct-acting antiviral therapy. Due to shared routes of transmission, HCV infection occurs in 15%-40% of persons infected with human immunodeficiency virus (HIV) and may be used as a marker of HIV exposure. Emergency departments (EDs) play major roles in screening for HCV infection and HIV infection. Several ED- based HCV screening programs have been implemented and have identified previously unrecognized HCV infections, but many challenges remain. Because targeted screening programs use methods that often fail to detect high-risk behaviors (e.g., self-reported information on prescreening questionnaires or review of patient problem lists at time of visit), they do not effectively identify persons at high risk of HCV infection (e.g., IDU). Nontargeted HCV screening strategies require less assessment of risk behaviors. However, concerns such as high costs and unnecessary tests make nontargeted screening strategies difficult to implement and sustain. Therefore, an innovative, effective, and sustainable HCV screening strategy is urgently needed. We propose to develop, implement, and evaluate a tailored, effective, and sustainable, prediction algorithm- based screening tool called Hepatitis C Emergency Department (HepC-EnD) that can be used by health care systems to identify patients at high risk of HCV infection. We will achieve these goals through three specific aims. Aim 1 will develop and validate prediction algorithms using machine learning and natural language processing to identify patients at risk of HCV infection through Florida’s all-payer electronic health records (EHRs) accessed via the OneFlorida+ Clinical Research Consortium. In Aim 2, we will design a HepC-EnD prototype that incorporates the best prediction algorithms to provide automatic notification to ED providers of patients at high risk of HCV infection. Informed by implementation science frameworks, we will enhance the functionality and usability of HepC-EnD through a workshop and qualitative interviews. In Aim 3, we will integrate HepC-EnD into the University of Florida Health EHR system to deploy and test HepC-EnD in two EDs (Gainesville and Jacksonville) and compare the performance of HepC-EnD with nontargeted screening using a difference-in- differences approach. Performance will be assessed by evaluating the usability, acceptability, effectiveness, and cost-effectiveness of the tool. Our proposed research is highly significant in its integration of a cutting-edge machine-learning–based prediction and risk stratification tool into an e-platform that will better inform clinical practice for improving HCV/HIV screening and linking patients with care. Our findings will provide timely data and adaptable strategies that are key in attaining the national and global goals of eliminating HCV infection.

项目总结/摘要 丙型肝炎病毒（HCV）感染在美国显著增加，主要是由于注射 与持续的阿片类药物流行相关的药物使用（IDU）。此外，在320万人中， 慢性丙型肝炎仍然未被诊断，导致显着的发病率和死亡率，尽管可用 有效的直接作用的抗病毒治疗。由于共有的传播途径，HCV感染发生在15%-40% 感染人类免疫缺陷病毒（HIV）的人，并可用作HIV暴露的标志物。 急诊科（ED）在筛查HCV感染和HIV感染方面发挥着重要作用。几个艾德- 已经实施了基于HCV的筛查计划，并确定了以前未识别的HCV 感染，但仍然存在许多挑战。因为有针对性的筛查项目使用的方法往往不能 检测高风险行为（例如，关于预筛选问卷或患者审查的自我报告信息 访问时的问题列表），它们不能有效地识别HCV感染高风险的人（例如，IDU）。 非靶向HCV筛查策略需要较少的风险行为评估。然而，诸如 高成本和不必要的测试使得非靶向筛查策略难以实施和维持。 因此，迫切需要一种创新、有效和可持续的HCV筛查策略。 我们建议开发，实施和评估一个量身定制的，有效的，可持续的预测算法- 基于筛查工具，称为丙型肝炎急诊科（HepC-EnD），可用于医疗保健 系统来识别HCV感染的高风险患者。我们将通过三个具体目标实现这些目标。 Aim1将使用机器学习和自然语言处理来开发和验证预测算法 通过访问佛罗里达的所有付款人电子健康记录（EHR）， OneFlorida+临床研究联盟在目标2中，我们将设计一个HepC-EnD原型， 结合了最佳预测算法，以向艾德提供者提供高血压患者的自动通知。 HCV感染的风险。通过实施科学框架，我们将增强功能， HepC-EnD的可用性通过研讨会和定性访谈。在目标3中，我们将HepC-EnD整合到 佛罗里达大学健康EHR系统部署和测试HepC-EnD在两个ED（盖恩斯维尔和 杰克逊维尔），并比较HepC-EnD与非靶向筛选的性能， 差异的方法。将通过评价可用性、可接受性、有效性和 工具的成本效益。我们提出的研究是非常重要的，在其整合的尖端 将基于机器学习的预测和风险分层工具整合到一个电子平台中， 改善HCV/HIV筛查并将患者与护理联系起来的做法。我们的发现将提供及时的数据 和适应性强的战略，这是实现消除HCV感染的国家和全球目标的关键。