Predictive Modeling for Patient Hospitalization: Enhancing Healthcare Outcomes

The main motivator for this project is to make improvements to the healthcare system and resource management so that it is easier for hospitals and emergency responders to transport patients based on need and to allow emergency room patients to be treated based on need as well. This information will allow municipalities nationwide to reassess their hospital and related care capacities to better suit trends in their regions. Binary classification was utilized to achieve this, employing various models and concluding with an ensemble classifier that uses KNN, RF, AB, GNB, and XGB as the initial estimators, and GB as the final estimator. The accuracy hovers around 0.96, AUC around 0.90, precision macro average around.95, recall macro average around 0.91, and f1-score macro average around $\mathbf{0.93}$.