Research Article
Explainable Artificial Intelligence (XAI) for Interpretable Heritage Building Maintenance Prediction
Authors: Nur Shahirah Jailani, Mohd Murtadha Mohamad, MD. Enjat Munajat, Ira Irawati, Heru Nurasa, Mohd Shahizan Bin Othman, Apri Junaidi
Affiliations: Universiti Teknologi Malaysia, Universitas Padjadjaran
Published: November 14, 2025
Executive Impact at a Glance
This research introduces a paradigm shift in heritage building maintenance, moving from reactive to proactive and evidence-based preventive conservation. The novel Decision-Support Framework (DSF) integrates predictive analytics with explainable AI to offer transparent, actionable insights for optimal resource allocation.
0.00
XGBoost F1-Score
0
Total Citations
0
Total Downloads
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
The Challenge in Heritage Maintenance
Traditional maintenance prioritization relies heavily on expert judgment and periodic inspections, which are often subjective, reactive, and lack integration with quantitative environmental data, leading to inconsistent assessments and delayed responses to emerging risks.
Introducing the Decision-Support Framework (DSF)
This paper introduces a novel Decision-Support Framework (DSF) that bridges the gap between predictive analytics and practical conservation workflows. The framework integrates an eXtreme Gradient Boosting (XGBoost) machine learning model, which predicts maintenance priority ratings based on historical microclimate data, with an interactive dashboard built on Microsoft Power BI. A key innovation is its emphasis on explainability, using SHapley Additive exPlanations (SHAP) to interpret model outputs, and visual analytics to transform complex predictions into actionable insights.
Our Core Contributions
- Robust predictive model (XGBoost) for maintenance priority.
- Integration of XAI (SHAP) for interpreting predictions.
- Interactive visual analytics dashboard (Power BI).
- Comprehensive evaluation with case study and expert feedback.
XGBoost Model Performance
The XGBoost model demonstrated superior performance with an F1-score of 0.88, effectively identifying true high-priority cases while minimizing false alarms. This highlights the model's robustness in predicting maintenance needs based on historical microclimate data.
Key Explainable AI (XAI) Insights
SHAP analysis revealed that humidity variation (32%) and mean temperature (27%) are the most dominant environmental stressors contributing to material degradation in heritage buildings. This transparency fosters trust and understanding among conservation professionals.
Strategic Impact & Future Outlook
This study enables a paradigm shift from reactive to proactive and evidence-based preventive conservation. The XAI-enhanced model transforms complex AI predictions into actionable conservation intelligence, empowering heritage managers with transparent, data-driven decision-making, and optimizing resource allocation for building preservation.
0.88
XGBoost F1-Score Achieved
Enterprise Process Flow
Data Acquisition
→
Data Cleaning & Normalization
→
Feature Engineering
→
Model Training (XGBoost)
→
SHAP XAI Module
→
Power BI Dashboard
→
Decision Support
| Traditional Approach | Our XAI-Driven DSF |
|---|---|
|
|
Case Study: Johor Bahru Railway Station
This historic Malaysian building, constructed in the 1930s, served as a crucial site for validating our DSF. Its mixed material composition and exposure to severe tropical climate conditions (high humidity, temperature) provided a rich dataset for modeling diverse degradation behaviors. The system successfully predicted maintenance priorities, showing that humidity variation (32%) and mean temperature (27%) were key environmental stressors accelerating material decay.
Calculate Your Potential AI Impact
Estimate the efficiency gains and cost savings your enterprise could realize by implementing AI-driven solutions.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Your AI Implementation Roadmap
A typical enterprise AI journey, broken down into actionable phases for clear understanding and successful deployment.
Phase 1: Discovery & Strategy
Initial assessment of current operations, identification of AI opportunities, and definition of clear objectives and KPIs. This phase involves stakeholder interviews, data audits, and feasibility studies to align AI initiatives with business goals.
Phase 2: Data Foundation & Engineering
Establishment of robust data pipelines, cleaning, and preparation of datasets for AI model training. This includes data governance, security protocols, and integration with existing enterprise systems to ensure data quality and accessibility.
Phase 3: Model Development & Training
Design, development, and training of custom AI models tailored to specific use cases. Iterative model refinement, hyperparameter tuning, and rigorous validation ensure optimal performance and accuracy.
Phase 4: Deployment & Integration
Seamless integration of trained AI models into existing operational workflows and IT infrastructure. This phase focuses on API development, system compatibility, and minimal disruption to ongoing business processes.
Phase 5: Monitoring, Optimization & Scaling
Continuous monitoring of AI model performance, post-deployment optimization, and scaling of solutions across the enterprise. Regular updates, retraining, and performance analytics ensure sustained value and adaptability to evolving business needs.
Ready to Transform Your Enterprise with AI?
Harness the power of AI to drive efficiency, innovation, and competitive advantage. Schedule a complimentary strategy session with our experts.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat