Application of a Bayesian network to aid the interpretation of blood alcohol (ethanol) concentrations in air crashes

Abstract

In the investigation of a fatal air crash, it is important to determine if the pilot, at the time of death, was contravening the regulations in relation to 1) the permitted concentration of ethanol (alcohol) in the blood and 2) whether the pilot had consumed alcohol within a specified period before flying. It is also important to assess whether any alcohol detected in the toxicological samples was present either because of consumption or because of post-mortem alcohol formation. We have developed a Bayesian Network that models the relationships between analytical results, circumstantial evidence and the concentration of alcohol at the time of death in cases of air crash. The model provides a rational, coherent approach to forensic interpretation, moving interpretation from a largely subjective, generalist approach to a more objective, case-specific methodology utilising available relevant data and accommodating the inevitable uncertainties within a case.