Artificial Intelligence in Healthcare Construction Scheduling: Comparative Analysis of MLP, SVR, and Decision Trees

Abstract

Efficient scheduling of healthcare infrastructure projects plays a crucial role in enhancing public health services by ensuring timely delivery and optimal resource allocation. This study investigates the application of Artificial Intelligence (AI), particularly the Multi-Layer Perceptron (MLP) neural network, in predicting project completion times for healthcare construction projects. A dataset of 300 real-world healthcare projects in Iran was analyzed, incorporating variables such as budget, number of beds, geographic location, and contractor profile. The MLP model demonstrated superior predictive performance compared to Support Vector Regression (SVR) and Decision Tree Regression, achieving a Mean Absolute Error (MAE) of 12.6 days and a Root Mean Square Error (RMSE) of 16.8 days with an R² score of 0.94. These results highlight the potential of AI-driven models to support data-informed decision-making in healthcare project management, reduce scheduling uncertainties, and improve project success rates. The findings offer practical implications for policymakers and project managers seeking to modernize infrastructure planning through intelligent forecasting techniques.