With the goal of increasing the safety of civilian air flight, the detection of a decrease in pilot attention is becoming an important need in civilian aeronautics. Multiple models used for the detection of hypovigilant states have been developed over the years in experimental conditions, but barriers still exist limiting current use. First, some of these models require the execution of behavioral tasks that can disrupt pilot workflow. Second, other models relying on the use of physiological monitoring devices are still too cumbersome.