Biomedical Data Translator Technical Feasibility Assessment and Architecture Design

点击翻译按钮获取中文摘要

The Carolina Biomedical Data Translator (CBDT) will address the significant challenges in clinical and translational research by permitting clinicians, nurses, biomedical researchers, and clinical scientists to query, exploit, and contribute to a federated collection of complex and highly disparate data with medical importance. This data federation—a “quasi-comprehensive” medical federation—will enable interrogators to pose questions about human health and disease that could not heretofore have been imagined. We anticipate that these analyses will increase, in ways that are paradigm shifting, our understanding of new treatments and treatment targets, mechanisms of disease and disease transmission, environmental triggers, opportunities for drug repurposing, molecular candidates, and systems biology pathways. To realize this vision, we propose a trajectory that permits the realization of such a federation. We intend to start modestly and prototype the CBDT using the unique tools we have been building for other informatics projects, and then iteratively improve the CBDT in collaboration with NCATS. Ultimately, the CBDT federation will need to be socialized, permitted, and nurtured in order to realize its full potential. Indeed, we believe that the biggest single challenge will be establishing a governance model that permits the assembly of information pulled from across institutions and organizations. We discuss this issue in Section 2.2.1. The CBDT must be: • Dynamic: regularly updated to accommodate the addition of new data and data sources; • Agile: modifiable for uses beyond the original design and adaptable for other data analysis; • Usable: permitting differential access and levels of access (view, edit, etc.) to the data, and permitting use by both experts and new users with modest training; • Rational: the data will be clinically and scientifically relevant, as well as informed by policy and protocols, and the system will provide answers with contextual meta-data; and • Curated: the data will have associated provenance and quality measures, including statistical relevance where appropriate. We envision a dynamic system that supports data ingestion and subsequent curation, federation, and exploitation (see figure). We have assembled a team that has a track record of interdisciplinary collaboration and development of open source tools. We propose a demonstration project that is feasible in the 2-year time frame, but demonstrates all of the key functionality that the final system must possess.