PheBC: bias correction methods for EHR derived phenotype

PheBC：EHR 衍生表型的偏差校正方法

• 批准号: 10839649
• 负责人: Yong Chen
• 金额: $ 24.77万
• 依托单位: UNIVERSITY OF PENNSYLVANIA
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-09-01 至 2025-05-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10839649
• 关键词: Address; Administrative Supplement; Adoption; Algorithms; Architecture; Artificial Intelligence; Clinical; Clinical Research; Cloud Service; Code; Communities; Complex; Computer software; Consensus; Consumption; Data; Development; Disease; Drug Exposure; Ecosystem; Electronic Health Record; Ensure; Goals; Individual; Information Retrieval; Intention; Internet; Knowledge; Knowledge Discovery; Learning; Life Cycle Stages; Literature; Logic; Manuals; Methods; Observational Study; Patients; Pharmaceutical Preparations; Phenotype; Process; Protocols documentation; Publishing; Pythons; Readability; Records; Reproducibility; Research; Research Personnel; Services; Software Engineering; Software Tools; System; Technology; Time; Visualization; cloud based; cohort; community engaged research; computable phenotypes; computing resources; concept mapping; deep learning algorithm; design; eHealth; feature extraction; flexibility; graphical user interface; improved; machine learning model; method development; open source; open source library; parent project; research and development; software development; tool; treatment response; usability; web based interface; web services

PROJECT SUMMARY Phenotyping drug observational to records advanced required refers to the process of identifying specific phenotypic patient statuses, such as disease status, exposure, and treatment response. I is one of the most critical data extraction tasks in real-world studies based on patient data. Traditionally, phenotyping has heavily relied on expert consensus create phenotype definitions for individual diseases. However, with the widespread usage of electronic health (EHRs) in clinical research and the development of artificial intelligence (AI) technologies, more phenotyping efforts have been devoted to automated feature extraction, reducing the manual effort to create precise phenotypes. t There on phenotyping for phenotyping Our phenotype Specifically, adding We research are significant challenges in acquiring and reusing existing phenotyping information and algorithms based electronic health records (EHRs) in a computable manner. Furthermore, in our current project, the information extraction system and ias correction tool were originally designed as stand-alone tools research purposes, we propose to develop a set of open services that f acilitate the sharing and reuse of information extraction tools and bias correction tools in research communities. overarching goa of t his Administrative Supplement is to disseminate machine-readable and computable definitions and algorithms to reduce duplication of effort and improve reproducibility in clinical studies. the two specific aims are: (1) Enhance the reusability of phenotyping information extraction tools by APIs and services. (2) Engage research communities to promote the adoption of bias correction tools. plan to refactor our software architecture and user interfaces to enhance the adoption of our tools among communities. b l

项目摘要 表型 药物 观察性 到 记录 先进 需 是指鉴定特定表型患者状态，如疾病状态， 暴露和治疗反应。I是现实世界中最关键的数据提取任务之一 基于患者数据的研究。传统上，表型分析在很大程度上依赖于专家共识 为个别疾病创建表型定义。然而，随着电子健康的广泛使用， （EHR）在临床研究和人工智能（AI）技术的发展，更多 表型分析的努力已经致力于自动化特征提取，减少了人工工作 来创造精确的表型 不 那里 对 表型 为 表型 我们 表型 具体地说， 添加 我们 研究 在获取和重用现有的表型信息和算法的基础上， 电子健康记录（EHR），以可计算的方式。此外，在我们目前的项目中， 信息提取系统和国际会计准则修正工具最初设计为独立工具 研究目的，我们建议开发一套开放的服务，促进共享和重用 研究界的信息提取工具和偏见纠正工具。 《行政补编》首要果阿是传播机读和可计算的 定义和算法，以减少重复工作，提高临床研究的再现性。 两个具体目标是：（1）通过以下方式增强表型信息提取工具的可重用性 API和服务。(2)让研究界参与进来，以促进偏见纠正工具的采用。 计划重构我们的软件架构和用户界面，以提高我们的工具在 社区. B L