Linking excellence in biomedical knowledge and computational intelligence research for personalized management of CVD within PHC

Funder

Organizational Unit

Publications

How really reliable is Real World Data? An implementation study in diabetes and cardiovascular risk

Publication . Almeida, Gonçalo Piriquito de; Ruano, M. Graça; Ribeiro, Rogério José Tavares

Real-world data (RWD) is, more and more often, being used in many medical areas, and data mining is becoming an important feature of any given study. The medical practice has been dominated by tools derived from controlled trials and focused studies, but now with the availability of RWD and information worldwide, new and better sets of tools are being created. With this in mind, a study was created to verify if the internal database of a private clinic is suitable for data-mining with the main objective of creating more efficient tools capable of being helpful for the clinicians in the near future. Working with a database that possesses records as late as 1999 poses many challenges. Besides requiring a constant update to include the new digital areas and recent clinical parameters, it is also necessary to evaluate the existence and impact of human errors and if other improvements are required to avoid lack or mistaken results on future database queries. After the establishment of relevant datasets and tables, a standardization of the data is required. A systematic analysis of the database management was undertaken regarding data mining. The present study focused only on the database structure and content related with cardiovascular diseases of diabetic patients. For this goal, a thorough and exhaustive mining and understanding of the data had to be made manually using only pure logic and SQL queries to capture all the information needed. Resultant analysis shows epidemiological results consistent with previous studies and concludes about the need of improvements of the database’s Front and Back offices. These improvements will facilitate future studies relating cardiovascular and diabetes pathologies. Besides the existence of a huge amount of valuable clinical information, usage of the database’s data for general data-mining becomes a difficult task due to its unfriendly querying structure.

2017-11-27Master thesis

Open Access