MSc. Dissertation Project
Project Description
Conducted with Novartis Ireland, Dublin
Prediction of rehospitalization and mortality due to Atherosclerotic Cardiovascular Disease (ASCVD) events in ICU setting using machine learning techniques on MIMIC III dataset
Cardiovascular disease remains the leading cause of morbidity and mortality in the world and accounts for one third of deaths globally each year. The current study aims to leverage real world data of patients who are hospitalized due to ASCVD event, to predict rehospitalization and mortality based on the Electronic Health Data (MIMIC III data set). We developed a prediction model for 30-day readmission and mortality based on patient profile and identified top 5 features with high precision. Further, we developed another machine learning model to predict readmission and mortality till 180 days using chart event, diagnoses, comorbidities, and procedures data for first 48 hours along with demographics data and length of stay. These results can be very helpful to predict the mortality and readmission of high-risk patients who are readmitted to ICU after first ASCVD event which might address the huge unmet need of aiding healthcare resource planning, better patient care and prevention of hospitalization and death by intensifying treatment interventions in those patients. Further work is needed to identify the cumulative factors for prediction and create a risk assessment and action plan
MIMIC III data set
Planned publications (In progress):
- Abstract 1: Prediction of 30-day mortality in high-risk ASCVD patients using electronic health record data from MIMIC III database (planned to be submitted to American Heart Association 2021)
- Abstract 2: Prediction of 30-day readmission in high-risk ASCVD patients using electronic health record data from MIMIC III database (planned to be submitted to Current Applications and Future of Artificial Intelligence in Cardiology)
- Manuscript: Prediction of rehospitalization and mortality due to Atherosclerotic Cardiovascular Disease (ASCVD) events in ICU setting using machine learning techniques on MIMIC III dataset (Manuscript Under preparation