# Data-driven Bayesian networks for risk scenario mapping of Falls from height accidents

## TL;DR

This paper uses Bayesian networks to analyze falls from height accidents in construction, identifying key risk factors and scenarios to improve safety.

## Contribution

The study introduces a data-driven Bayesian network model to map risk scenarios for falls from height accidents in construction.

## Key findings

- Five core dimensions influencing falls from height accidents were identified.
- Five high-risk variables within these dimensions were determined.
- Probabilities of accidents under different risk scenarios were analyzed.

## Abstract

Falls from height (FFH) represent the most frequent type of accident in the building industry, leading to substantial economic losses and posing serious threats to worker safety. While risk analysis plays a vital role in accident prevention, a more comprehensive understanding of risk can significantly contribute to reducing the occurrence of accidents. To better capture the complexity and uncertainty inherent in risk factors, this study adopted the concept of risk scenarios to investigate the underlying mechanisms and driving factors associated with FFH accidents. A total of 368 FFH accident reports from 2014 to 2024 were collected, and a Bayesian network model was developed based on the validated data extracted from these reports. Through this model, various risk scenarios of FFH accidents were systematically explored. The analysis identified five core dimensions influencing the occurrence of FFH accidents, along with five high-risk variables within these dimensions. Moreover, the study examined the probabilities of FFH accidents under different risk scenarios. This scenario-based approach offers new insights into construction safety management and provides valuable implications for enhancing FFH accident prevention strategies in construction projects.