Application of Multi-Criteria AI-Based Modeling in Predicting and Assessing Fire Safety in Healthcare Facilities
Code: G-1115
Authors: Samaneh Salari ℗, Mehdi Ghasri , Ali Karimi *
Schedule: Not Scheduled!
Tag: Health Policy, Law & Management in AI
Download: Download Poster
Abstract:
Abstract
Background and aims: Fire safety in healthcare facilities is of paramount importance due to the limited evacuation capabilities of occupants in these centers. Consequently, inadequate fire safety precautions result in increased fatalities and financial losses. This research introduces an integrated model to predict and assess fire safety for occupants within these buildings. Method: To achieve this, 315 fire scenarios were simulated to determine the fire safety level through emergency evacuation modeling (using PathFinder software) and fire and smoke spread modeling (using PyroSim software). In each scenario, 13 features, including fire load density, occupancy, and building characteristics, were selected as model inputs to predict fire safety and evacuation risk. Results: Among the AI-based techniques examined, the MLP-PSO (500 iterations) method demonstrated the best performance for predicting the fire safety level of hospital occupants, with an RMSE (Total data) of 0.078, an R² (Total data) of 0.99, and an SD (Total data) of 1.46. When compared to the FRAME approach, the results of the MLP-PSO model exhibited a significant correlation in fire safety predictions with respect to the Fire Risk Assessment Method for Engineering (FRAME). Conclusion: Hospital administrators can enhance emergency preparedness and mitigate fire risks with the aid of this precise approach. The integration of simulation and machine learning provides a more comprehensive and accurate assessment and prediction of fire safety, ultimately contributing to safer healthcare buildings.
Keywords
Fire Safety, MLP-PSO, Hospital, Evacuation Risk.
