Bayesian Networks Applied to Risk Modeling of Depressurized Flight at High Altitude

Abstract

This study analyses the health risks of high-altitude parachute launch missions, such as hypoxia, decompression sickness, and otological barotrauma, using Bayesian probability networks to assess and mitigate these risks. All of these risk factors arise from the characteristics of the mission, which occurs in an environment hostile to human physiology in terms of pressure and temperature. Bayesian networks are effective tools for modelling uncertainties and diagnosing complex systems, aiding in event planning and decision-making. In the study, three Bayesian networks were constructed using Netica software, with steps for defining causes, effects, and causal relationships, in addition to determining probabilities. For the networks assembled, the results showed that hypoxia presents a risk of serious damage of 1 in 10,000 flights, while decompression sickness has a lesser impact, with a risk of moderate damage of 1 in 50,000 flights. Otological barotrauma presents a risk of moderate damage of 1 in 250 flights. In some of the analyses, the human factor proved to be an essential element in mitigating risks. Finally, the effectiveness of Bayesian networks in risk assessment is highlighted, suggesting the acquisition of more data to consolidate the probabilities
employed, and proposing studies that include logistical and operational aspects in the planning of missions of this profile.