Rare disease patients often experience diagnosis delays or misdiagnosis. The collection and public sharing of accurate, up-to-date, standardized data on rare diseases are key priorities in shortening the diagnostic odyssey. These data would inform health care providers about the up-to-date information/findings on the rare diseases they are treating, but also would educate patients and their families with critical information pertinent to their condition. Here, we introduce the NCATS Rare Disease Alert System (RDAS) as a public, comprehensive rare disease resource that supports searching, informing, educating, and alerting users on the latest information and findings pertinent to rare diseases. RDAS is built upon data integrated from multiple sources, including publications, clinical trials, and NIH grant funding.
RDAS comprises a frontend user interface (UI) and a backend data repository. The UI allows users to search, browse, and subscribe to RDAS, in order to receive the latest information and findings about their rare disease(s) of interest. The UI is built using the Angular front-end library, with a Node/Express server layer to connect to the underlying Neo4j database that stores the backend Knowledge Graphs (KGs). The backend data repository includes three KGs built by integrating information from PubMed articles, clinical trials, and NIH grant funding related to rare diseases. The KGs are: 1) Scientific Annotation Knowledge Graph (RDAS_SAKG) contains annotations generated from article titles and abstracts by PubTator, and epidemiology information extracted from article abstracts using our previously developed deep learning models. 2) Clinical Trial Knowledge Graph (RDAS_CTKG) semantically represents data extracted from clinicaltrial.gov. 3) NIH Grant Funding Knowledge Graph (RDAS_GFKG) contains annotations generated from project titles and abstracts of NIH funded projects related to rare diseases.
