Earlier this month, the NIH released a five-year plan for data science and announced that it will hire a chief data strategist, a first for the agency, to lead the implementation of the plan, along with the NIH Scientific Data Council and NIH Data Science Policy Council. The NIH cites a 2016 survey indicating that data scientists spend 80% of their time collecting and organizing data, leaving only 20% for analytical work, such as mining data for trends and patterns. The recent advances in storage, processing and computation should be taken advantage of, according to the NIH, by utilizing them to establish a secure and sustainable “data ecosystem.”

Biomedical data, clinical data and genomics data are a few types of data that require improved storage methods due to their exponential size. Genomics data alone are predicted to equal, if not exceed, total data collected from astronomy, YouTube and Twitter by 2025. According to the NIH, data must be FAIR: Findable, with unique identifiers that make them easy to locate; Accessible, so that they are easily retrievable from a secure system; Interoperable, with a standardized vocabulary so the data can be used and understood across NIH research groups; and Reusable, with concise information regarding data usage licenses and traceability.

Currently, a large portion of biomedical data are maintained by individual scientists or small research groups, and are not standardized in a singular format, making it complicated for scientists to find and utilize their peers’ data. To centralize biomedical data and drive efficiency and sustainability in biomedical research, the NIH proposes five aims as part of its data science plan: improving data storage and security, which includes interconnecting new and existing NIH data systems; modernizing data repositories, and better integrating clinical and observational data into biomedical data science; optimizing and growing the NIH data science workforce; and developing policies to underscore FAIR data.

Source: NIH