publication date: Jun. 7, 2019
Big Data for outcomes and clinical research: major advance or improvement needed
By Roy B. Jones, Amar Chahal, Dianne Reeves, Gregory H. Jones, Charles S. Martinez
Creation of Big Data repositories is now emphasized at virtually all research institutions and the NIH, but the number of publications describing patient outcomes from these sources appears modest.1 Why is this so; what factors limit what should be a hugely productive resource, and how can we improve the impact of this use of Big Data? Why does this issue require greater physician engagement and understanding to solve? The integration of clinical, laboratory, and financial data is required to describe disease and treatment outcomes as well as treatment value.
Key to understanding these issues are the concepts of information, structured data, and data standards. Computers must use a uniform data structure to analyze any data set, and making this structure compatible across multiple institutions is required to merge their data into a single database (Figure 1). In contrast, most medical practice places major emphasis on collection of information which routinely lacks structure; for example, clinical notes.