Explainable Artificial Intelligence (XAI)-Enabled Probabilistic Fire Risk Prediction for High-Rise Residential Buildings: SHAP Attribution of Human and Organizational Risks

Fire safety in high-rise residential buildings depends on the reliable performance of active fire protection systems subject to technical, human, and organizational risks. Probabilistic risk assessment frameworks incorporating human and organizational errors (HOEs) show that HOEs raise expected risk-to-life by 20 to 37%, yet such models remain inaccessible to the building owners, facility managers, qualified persons, and regulators who must act on their outputs. This paper applies Explainable Artificial Intelligence, specifically SHAP (SHapley Additive exPlanations), to a Bayesian network probabilistic fire risk model integrated with Markov Chain Monte Carlo posterior uncertainty quantification, extending the validated T-H-O-Risk methodology across sixteen active fire safety system configurations and seven case studies in Singapore, Australia, Hong Kong, New Zealand, and the United Kingdom. Global SHAP analysis shows that maintenance-related HOEs (H8, insufficient safety check; H9, inadequate periodic inspection) account for 83.1% of total HOE attribution, outranking compliance and training variables and reframing the primary intervention from behavioural to structural. Validation against published results yields Pearson r = 0.927 across 112 building-design combinations. This is the first application of SHAP attribution to a Bayesian network fire risk model, giving regulators and qualified persons a transparent, uncertainty-aware tool for inspection-regime calibration and ALARP demonstration.