What does this research mean for the field?

An integrated framework combining a Physics-Informed Temporal Graph Transformer and a Bayesian Markov-switching econometric model accurately predicts structural reinforcement needs and lifecycle repair urgencies, achieving 93.8% classification accuracy and a repair-urgency AUC of 0.956. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

Develop a construction-to-lifecycle risk intelligence framework to optimize sequential decisions for structural reinforcement during execution and repair prioritization during operations under incomplete data and operational constraints.

September 14, 2026International Journal of Construction Management

Construction-to-lifecycle risk intelligence: a noise-robust Physics-Informed Temporal Graph Transformer and Bayesian Markov-Switching sample-selection framework for reinforcement and repair decisions

View Full Paper

Ask AI

Bookmark

Authors

ASAli Shehadeh OAOdey Alshboul HNHamsa Nimer

Discussion

Member takes

Overview

Computational framework demonstrates accurate reinforcement classification and repair prioritization in infrastructure lifecycles, indicating robust decision support under operational uncertainty.

Key Points

Develop a construction-to-lifecycle risk intelligence framework to optimize sequential decisions for structural reinforcement during execution and repair prioritization during operations under incomplete data and operational constraints.
Integrated a Physics-Informed Temporal Graph Transformer (PITGT) with contrastive noise filtering to capture spatial, temporal, and network dependencies while decoupling true structural needs from financially distorted maintenance logs.
Combined uncertainty-aware ensemble learning with a hierarchical Bayesian Markov-switching sample-selection econometric model to estimate latent deterioration regimes, time-varying repair thresholds, and project heterogeneity.
Achieved a 93.8% classification accuracy for four-level reinforcement recommendations and a macro-F1 score of 0.921.
Demonstrated robust dynamic performance with a transition-event F1 score of 0.816 and an area under the curve (AUC) of 0.956 for probabilistic repair urgency.

Cite This Study

Shehadeh et al. (2026) studied this question.

synapsesocial.com/papers/6aa7b3bf0926e14a848b2e5d https://doi.org/10.1080/15623599.2026.2721502

View Full Paper

Ask AI

Bookmark

September 14, 2026International Journal of Construction Management

Construction-to-lifecycle risk intelligence: a noise-robust Physics-Informed Temporal Graph Transformer and Bayesian Markov-Switching sample-selection framework for reinforcement and repair decisions

View Full Paper

Ask AI

Bookmark

Authors

ASAli Shehadeh OAOdey Alshboul HNHamsa Nimer

Discussion

Member takes

Overview

Computational framework demonstrates accurate reinforcement classification and repair prioritization in infrastructure lifecycles, indicating robust decision support under operational uncertainty.

Key Points

Develop a construction-to-lifecycle risk intelligence framework to optimize sequential decisions for structural reinforcement during execution and repair prioritization during operations under incomplete data and operational constraints.
Integrated a Physics-Informed Temporal Graph Transformer (PITGT) with contrastive noise filtering to capture spatial, temporal, and network dependencies while decoupling true structural needs from financially distorted maintenance logs.
Combined uncertainty-aware ensemble learning with a hierarchical Bayesian Markov-switching sample-selection econometric model to estimate latent deterioration regimes, time-varying repair thresholds, and project heterogeneity.
Achieved a 93.8% classification accuracy for four-level reinforcement recommendations and a macro-F1 score of 0.921.
Demonstrated robust dynamic performance with a transition-event F1 score of 0.816 and an area under the curve (AUC) of 0.956 for probabilistic repair urgency.

Cite This Study

Shehadeh et al. (2026) studied this question.

synapsesocial.com/papers/6aa7b3bf0926e14a848b2e5d https://doi.org/10.1080/15623599.2026.2721502

View Full Paper

Ask AI

Bookmark