Exploration and Visualization of the Hidden Information from the Congestive Heart Failure Patients Data in MIMIC-III Database

In recent times, the systematic collection of clinical data (diagnosis, demographics, scans, hospital emergency admission, blood sample test results, etc.) using electronic devices called as electronic medical record (EMR) is increasing exponentially. EMR contains an enormous amount of information that is hidden from actual users and could be extracted only by doing proper data exploration and analysis. Irrespective of its unrestricted availability, the medical professionals who have lesser computer proficiency are unable to explore the hidden information in it. One such database is MIMIC-III which is a very big, freely available relational database, single centered and consists of details of patients admitted in the intensive care unit at Beth Israel Deaconess Medical Center with 15,682 unique diagnoses. The objective of this study is to show the importance of data exploration and visualization and thereby discover the hidden information in any dataset. For this study, an in-depth exploratory data analysis (EDA) is performed on congestive heart failure (CHF) patients in the benchmark MIMIC-III dataset. Finding the percentage of mortality in male and female patients and effective analysis on age attribute helps the physician in making critical decisions. Calculating the length of stay (LoS) in various units in the hospital helps in preplanned management of beds and drugs available in the hospital.