Bionformatics Resource Core

点击翻译按钮获取中文摘要

PROJECT SUMMARY In this renewal, we build on our experience serving as the VERITY Bioinformatics Core, expanding Core services to include advancements in the field since the last application. Research using electronic medical record (EMR) data continues to grow with increasing availability of EMR data. Simultaneously, methods to utilize data for research have also advanced including natural language processing (NLP) and machine learning (ML) to extract crucial clinical data embedded in narrative notes, and to include these data in models of disease risk and outcomes. However, there remains a large gap between access to raw EMR data optimized for billing and patient care, and the ability to fully and appropriately utilize these data in clinical research. Through consultations, courses offered as part of the Bioinformatics Core, and current Core projects, we have identified 4 areas of high demand and/or unmet need for clinical investigators: (1) phenotyping using EMR data; (2) extraction of clinical data from narrative notes using NLP, including early applications of this technology to study social determinants of health; (3) use of EMR for studies of treatment effects and applications of causal inference methods; and (4) approaches for multi-institutional EMR studies without requiring direct sharing of data (termed federated learning). The mission of the Bioinformatics Core remains supporting investigators from the pediatric and adult rheumatic and musculoskeletal (MSK) research community to apply and integrate bioinformatics approaches to clinical research studies using EMR data. While our target audience remains trainees and junior faculty, in this renewal, our expanded services are also designed for established investigators interested in incorporating bioinformatics to their research programs. Aim 1. To provide methods for investigators to obtain robust and accurate phenotypes using information from EMRs and integrating these data for clinical studies. This requires applying supervised and unsupervised machine learning approaches for phenotyping with EMR data. As well, we will utilize causal inference methods applied to EMR data for studies of treatment effects. Aim 2. To provide NLP support enabling clinical research studies with EMR data. We will support and develop the use of NLP to incorporate social determinants of health (SDoH) in studies of health equity using EMR data. As well, we will support and educate investigators on the use of NLP based data and tools. Aim 3. To strengthen existing ties and build new partnerships between the rheumatic and MSK clinical research and bioinformatics communities through Core platforms and consulting services. The Bioinformatics Core team and a network of expert advisors will perform consultations, provide educational services, and deliver bioinformatics research services to the Research Community. Additionally, we will build on our foundation to enable cross-institutional studies with federated learning approaches for clinical studies on rheumatic and musculoskeletal conditions.