<p>This study proposes an artificial intelligence-based predictive modeling framework to assess construction project performance using the project performance index (PPI), a composite indicator that integrates cost, schedule, and quality metrics. The PPI is derived from the cost performance index (CPI), schedule performance index (SPI), and quality performance index (QPI), offering a holistic measure of project efficiency. A dataset of 200 completed construction projects across India, obtained from industry partners, was used for training and evaluating six machine learning models: multiple linear regression (MLR), artificial neural network (ANN), decision tree regression (DTR), random forest regression (RFR), support vector regression (SVR), and extreme gradient boosting (XGBoost). The models were optimized using cross-validation and assessed using multiple metrics including R², RMSE, MAE, MAPE, IOA, VAF, and a20. Among these, XGBoost achieved the best performance with an R² of 0.876, RMSE of 0.113, and MAE of 0.087. Sensitivity and feature importance analyses revealed that project size, schedule delays, and labor cost were the most influential variables. The findings confirm that advanced machine learning techniques, particularly ensemble-based models, provide accurate and robust predictions, supporting data-driven decision-making and risk mitigation in construction project management. The major limitation of this study is its reliance on region-specific data; future research should explore broader datasets and incorporate real-time project dynamics.</p>