ABSTRACT:
Background: Healthcare systems increasingly rely on AI-driven analytics to improve patient experience, optimize resources, and strengthen operational decision-making. Modern hospitals function as complex networks where clinical, administrative, and financial elements interact, requiring robust data-driven tools to predict service demand and guide workflow management. Objectives: The present study aimed to evaluate key hospital management indicators, develop predictive models for patient length of stay (LOS), compare model accuracies, and generate visual analytics that support evidence-based decision-making. Methods: A large-scale hospital dataset (AV Healthcare Analytics II; >300,000 records) was obtained from Kaggle and analyzed using Python-based machine-learning workflows. Ten operational indicators—bed occupancy, LOS, readmission rate, patient satisfaction, operational cost index, staff efficiency, resource utilization, prediction confidence, risk score, and service delivery time—were computed as dynamic KPIs. Monthly incidence trends and department-wise treatment success rates were visualized using Matplotlib. Logistic Regression and Random Forest algorithms were trained on processed data, and their predictive accuracies were compared. Results: Operational KPIs revealed high bed occupancy (89%), moderate LOS (8 days), and strong staff efficiency (89%). Monthly incidence exhibited a consistent upward trend, indicating rising service demand. Department-wise outcomes varied, with Cardiology performing best (91% success) and Surgery lowest (79%). Predictive modeling showed Random Forest outperforming Logistic Regression, achieving accuracies of 0.70–0.75 versus 0.62–0.65. Conclusion: The findings demonstrate that integrating machine-learning models with operational metrics enables hospitals to improve forecasting, resource allocation, and patient experience. AI-driven analytics provide a robust foundation for proactive, data-informed healthcare management.