Using Bayesian Network analysis to determine the main accident risk factors in Spain
Metadata
Show full item recordAuthor
Date
2015Subject/s
Unesco Subject/s
3305 Tecnología de la Construcción
5311.04 Organización de Recursos Humanos
6109.01 Prevención de Accidentes
Abstract
Occupational accidents cause considerable physical, psychological, and social harm to victims. The goal is to identify those risks that play a major role in reducing the likelihood of an accident. To do so, we propose probabilistic model, that involves nineteen occupational risk variables (Ri), the workplace accident variable (Accident) and the economic sector variable (Sector). The VII National Workplace Condition Survey provided the database used to fit a Bayesian network. The Bayesian network is a probabilistic graphical model showing existing dependency relationships between occupational risks, the economic sector and an accident. Sensitivity analyses are carried out to quantify the lowered accident probabilities by eliminating the two risk factors that together have the greatest potential to reduce the like- lihood of an accident. The Receive Operating Characteristic was used to evaluate or test the performance of Bayesian network models. The results show a noticeable difference by economic sector in both accident probability and in the overriding effect of certain risks. We conclude that the probabilistic model can be used to decide on preventive actions to lower workplace accident probability. © 2015 Taylor & Francis Group, London.
Occupational accidents cause considerable physical, psychological, and social harm to victims. The goal is to identify those risks that play a major role in reducing the likelihood of an accident. To do so, we propose probabilistic model, that involves nineteen occupational risk variables (Ri), the workplace accident variable (Accident) and the economic sector variable (Sector). The VII National Workplace Condition Survey provided the database used to fit a Bayesian network. The Bayesian network is a probabilistic graphical model showing existing dependency relationships between occupational risks, the economic sector and an accident. Sensitivity analyses are carried out to quantify the lowered accident probabilities by eliminating the two risk factors that together have the greatest potential to reduce the like- lihood of an accident. The Receive Operating Characteristic was used to evaluate or test the performance of Bayesian network models. The results show a noticeable difference by economic sector in both accident probability and in the overriding effect of certain risks. We conclude that the probabilistic model can be used to decide on preventive actions to lower workplace accident probability. © 2015 Taylor & Francis Group, London.
Collections
