Evaluating a Simulator-based Spatial Disorientation Training Program for Pilots: Key Predictors and Machine Learning Insights

Background: Spatial disorientation (SD) remains a critical safety concern in aviation, particularly for military fighter pilots operating in high-performance environments. While effective aeromedical training is essential to mitigate SD incidents, research quantifying the efficacy of simulator-based programs using advanced analytical methods remains limited within the Indian military context. Aims & Objectives: This study evaluated the effectiveness of a 1-day simulator-based SD training program for Indian military fighter pilots. The primary objectives were to assess improvements in illusion awareness, identify predictors of successful learning using statistical analysis and machine learning (Random Forest) techniques, and explore how these insights can optimize future training protocols. Materials & Methods: A mixed-methods, observational cohort design involved 303 active-duty fighter pilots (aged 20–50). Participants underwent a standardized, high-fidelity simulator-based training program replicating common in-flight illusions. Pre- and post-intervention Illusion Awareness Scores (IAS) were measured. Data analysis utilized Pearson’s Chi-square tests, ANOVA, ANCOVA, and Random Forest classification to assess training effectiveness and isolate predictive factors. Results: The intervention significantly enhanced illusion awareness across all participants. Baseline IAS was a critical determinant; “SD susceptible” pilots (IAS <30) demonstrated substantially higher mean IAS gain (54.39%) compared to “well-aware” pilots (16.24%; P < 0.001). Perceived simulator realism and prior experience with illusions were significantly associated with positive outcomes. The machine learning model achieved high predictive accuracy (correlation coefficient 0.9569), confirming baseline awareness, simulator fidelity, and prior exposure as the most influential predictors of success. Conclusion: A single-day simulator-based SD training program effectively improves illusion awareness, particularly among pilots with lower initial knowledge. The study confirms that training success depends heavily on baseline knowledge, simulator fidelity, and prior illusion experience. These findings underscore the necessity for realistic simulation environments and personalized strategies—enhanced by AI-driven insights—to improve flight safety and reduce dangerous SD-related accidents.