Community infrastructure management remains largely based on current-condition assessment and corrective or preventive interventions. This approach has a major limitation: it provides insufficient information about the likely deterioration trajectory of assets and the time at which their sustainability may cross a critical threshold. Research on the Global Sustainability Index (IGD) has introduced a multidimensional infrastructure assessment integrating physical, functional, environmental, economic, social and institutional dimensions. However, when calculated at a single point in time, the IGD remains essentially a state indicator.

This paper proposes a methodological protocol for developing and validating an Artificial Intelligence-based Predictive Global Durability Index (IGDₚ). The approach relates to a longitudinal database combining repeated IGD assessments with technical characteristics, usage intensity, environmental exposure, maintenance history, spatial information, and economic, social and institutional variables. Several model families are proposed for comparison, including statistical baselines, Random Forest, Gradient Boosting, XGBoost and, where justified by data availability, neural and temporal models.

The validation framework combines five dimensions: predictive performance, temporal validation, robustness and generalization, explainability, and decision usefulness. Proposed performance metrics include mean absolute error, root mean square error and coefficient of determination. Explainability is integrated through feature-contribution approaches, including SHAP. Procedures for uncertainty quantification, sensitivity analysis of IGD weights and external validation are also proposed.

The methodological architecture predicts the future values of individual durability dimensions before aggregating them into the IGD ₚ. This strategy preserves the multidimensional interpretability of the index while adding a prognostic capability. The proposed protocol therefore constitutes an intermediate step between static sustainability assessment and prescriptive community infrastructure management.