AI-powered cardiovascular ICU predictor: Intensive care unit admission prediction for cardiovascular patients

In 2019, cardiovascular diseases accounted for 17.9 million deaths globally, constituting 32% of all fatalities. As the primary cause of global mortality, conditions like Congestive Heart Failure, Acute Myocardial Infarction, Atrial Fibrillation, and Cardiogenic Shock require a paradigm shift in ICU admission practices This research integrates machine learning into cardiovascular healthcare, utilizing a dataset of 34,327 patients from the Medical Information Mart for Intensive Care-IV (2008-2019). The Random Forest model, with an accuracy of 93.24% for ICU admission prediction and 90.08% for length of ICU stay, proves promising. External model validation with Telehealth Intensive Care Unit Collaborative Research dataset emphasized real-world applicability, though acknowledging room for refinement due to missing variables (accuracy 0.78, ROC-AUC 0.85). The study contributes valuable insights into ICU admissions for cardiovascular patients, showcasing the potential of machine learning in healthcare decision-making.