Enhancing pilot vigilance assessment: The role of flight data and continuous performance test in detecting random attention loss in short IFR flights

Situational awareness (SA) and fatigue management are crucial aspects of aviation safety, particularly during demanding flight phases. This study introduces an innovative approach employing flight data, machine learning, and Continuous Performance Test (CPT) metrics to predict pilot performance and SA during instrument approaches under Instrument Meteorological Conditions (IMC). Data were collected from over 10 pilots across more than 40 flights in a high-fidelity Cessna 172 analog flight simulator.

Significant correlations were observed between dynamic cognitive performance parameters and the exceedance shape factor, a novel measure of pilot sustained attention introduced in this research. Key variables identified through correlation analysis included variability, interstimulus change, and reaction time standard deviation.

Importantly, commission scores and reaction time standard deviation emerged as key predictors in the machine learning model, specifically the Optimizable Gaussian Process Regression (GPR) model with a radial basis function kernel. The model achieved a validation R-squared of 0.90 and a test R-squared of 0.70. These systems could incorporate additional data sources, such as eye-tracking and scan pattern analysis, for a better assessment of pilot SA and fatigue levels. While post-flight measurements are inherently reactive, they are effective for monitoring the degradation of pilot CPT scores after each leg of high-frequency, short-duration flights.

Notable limitations include the need to understand individual cognitive differences among pilots, such as age, experience, and cognitive style. The predictive model also requires validation in actual flight conditions to determine its ecological validity. Future research should aim to address these limitations, generalize the findings, and integrate CPT data with other sensor inputs to provide a more comprehensive understanding of pilot performance.