AI-Driven Circular Construction: Integrating Social, Economic, and Environmental Dimensions through a Holistic Triple-Bottom-Line Model

The transformation of the construction industry towards sustainable models requires the integration of circular economy (CE) principles and the exploitation of advanced artificial intelligence (AI) techniques. This study presents an innovative comprehensive Triple Bottom Line (TBL) model that simultaneously quantifies environmental, economic, and social impacts in circular construction projects. To this end, a Deep Neural Network (DNN) architecture has been developed capable of capturing the complex non-linear relationships between key indicators such as water consumption, carbon footprint, energy use, maintenance costs, return on investment (ROI), skilled job generation and local participation. The model is trained and validated using synthetic data generated from realistic ranges observed in real projects, which allows simulating different scenarios and evaluating synergies and trade-offs between sustainability dimensions. In addition, the application of interpretability techniques (LIME and Permutation Feature Importance) reveals that indicators such as water use and carbon footprint are decisive in defining overall performance, as reflected in the TBL Score. The results show high accuracy in the prediction of this unified index and offer a practical tool for strategic decision making in the adoption of circular building practices. This approach not only lays the foundation for future empirical applications, but also opens the possibility of integrating new indicators that reinforce a holistic analysis of sustainability in the sector.