Hospital Readmission Prediction Using Machine Learning An Interactive Healthcare Analytics Dashboard Based on the UCI Diabetes 130-US Hospitals dataset

Hospital readmission is an important healthcare quality and resource-utilization outcome. This study presents an educational and reproducible machine-learning workflow for binary hospital-readmission prediction using the UCI Diabetes 130-US Hospitals dataset. The project implements data cleaning, missing-value handling, identifier removal, categorical encoding, exploratory data analysis, Random Forest classification, model evaluation, feature-importance analysis, and an interactive Streamlit dashboard.Using the implemented preprocessing and model configuration, the Random Forest classifier achieved 63.50% accuracy on the held-out 20% test set. For the positive readmission class, precision was 62.61%, recall was 52.08%, and F1-score was 56.87%.The dashboard integrates dataset exploration, visualization, model evaluation, feature importance, and an interactive prediction interface. The project is presented as an educational and portfolio prototype rather than a clinically validated decision-support system. Limitations include potential preprocessing leakage, use of a single train-test split, simplified target construction, and the absence of external validation and clinical calibration.