Feature selection model for diagnosis, electronic medical records and geographical data correlation

Electronic Medical Records (EMRs) collect and describe events and patient health history, related to his interaction with a healthcare facility or clinical trials. Raw data in EMRs are voluminous and heterogeneous. They need to be collected and stored to allow clinical management, treatment and to apply prevention protocols. Using informatics techniques (e.g., data mining models) allows to automatize the process of information extraction and to support health data management. We focus on biological data present in EMRs starting from blind data gathered from University Hospital of Catanzaro. In collaboration with Biochemical Laboratory of the University Hospital, we designed a workflow based system to analyze biological values. The system is able to relate biological data to diagnosis codes and with additional information integrated and correlated to EMRs data. Prediction models have been used and tested on 3 specific diagnosis, proving that system is able to: (i) identify blood test features that are important to detect a pathology and (ii) finding correlations among patients features.