SBIR Phase I: Determination of complex outcome measures using narrative clinical data to enable observational trials

这个小型企业创新研究(SBIR)项目的更广泛影响/商业潜力是使基于电子健康记录(EHR)的准确研究能够支持精确医学。这个国家正在踏上使用真实世界证据(RWE)来调整现有护理标准的旅程。基于EHR的分组分析和比较有效性研究将越来越多地用于加强监管和报销审批。这些努力需要准确地识别现实世界情景中的临床结果。试图确定现实世界结果的研究，如疼痛和无病生存，在索赔和EHR离散数据中显示出低准确率。这项提议旨在使用先进的语义技术准确地从电子病历数据中发现具有挑战性的结果。其目标是使准确的RWE研究能够在临床实践中实现RWE的更安全和更有效的使用。这个SBIR第一阶段项目建议创建一个从真实世界数据中提取临床结果的应用程序，以实现基于EHR的实用临床试验(PCT)。该方法使用自然语言处理(NLP)来深入研究暴露、干预和结果数据的健康记录，这些数据在索赔和EHR离散数据中不存在或不准确。项目目标包括使用NLP和本体映射从临床数据中提取特征，开发一个反映观察和临床试验中常见结果的知识库，以及使用临床数据从提取的特征中推断结果。该项目将使用去识别的纵向临床数据来验证结果检测引擎，以评估特征提取和推断结果的准确性。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势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/commercial potential of this Small Business Innovation Research (SBIR) project is to enable accurate electronic health record (EHR)-based studies to support precision medicine. The country is embarking on a journey of using real world evidence (RWE) to adjust the existing standard of care. EHR-based subgroup analytics and comparative effectiveness studies will increasingly be used to augment regulatory and reimbursement approval. These efforts require accurately recognizing clinical outcomes in real world scenarios. Studies attempting to identify real world outcomes, such as pain and disease free survival, have shown low accuracy rates in claims and EHR discrete data. This proposal aims to accurately detect challenging outcomes from EHR data using advanced semantic technologies. The goal is to enable accurate RWE studies to achieve safer and more effective use of RWE in clinical practice.This SBIR Phase I project proposes to create an application to extract clinical outcomes from real world data to enable EHR-based pragmatic clinical trials (PCTs). The approach uses natural language processing (NLP) to dive deep into the health record for exposure, intervention, and outcome data that do not exist or are inaccurate in claims and EHR discrete data. Project objectives include extracting features from clinical data using NLP and ontologic mapping, developing a knowledge database that reflects common outcomes in observational and clinical trials, and inferring outcome from extracted features using clinical data. The project will validate the outcome detection engine using de-identified longitudinal clinical data to assess accuracy of feature extraction and inferred outcome.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.