| Sumario: | Ensuring safety in next-generation intelligent cockpits demands accurate assessment of pilot states, particularly distinguishing between eustress and distress. Traditional stress monitoring lacks this nuance and struggles across varying flight tasks. This study proposes a multimodal neuro-cardiac framework combining functional near-infrared spectroscopy (fNIRS) and electrocardiography (ECG) to differentiate eustress and distress across tasks. Physiological data were collected from 35 participants under simulated flight missions inducing both stress types. Eleven features showing significant differentiation were identified and used to train classification models with machine learning algorithms. The model achieved 83.04% accuracy across tasks, and up to 90.83% within single tasks. These findings demonstrate the robustness of fNIRS-ECG-based monitoring in pilot stress classification. The proposed method offers objective biomarkers critical for adaptive intelligent cockpit systems, contributing directly to flight safety and human-machine interaction optimisation. PRACTITIONER SUMMARY: This study is the first to distinguish eustress and distress in pilots using multimodal physiological signals (fNIRS and ECG). Machine learning models achieved up to 90.83% accuracy, offering objective indicators of pilot stress states, supporting the development of adaptive intelligent cockpit systems, promoting optimised human-machine interaction and stress-aware decision-making. HIGHLIGHTS: Differentiates distress and eustress, not just stress, aligning with aviation safety. Uses low-interference methods, proving practical feasibility for real-world use. Applies a cross-task approach, enhancing scenario generalisation. Combines ergonomic application with insights into stress physiology.
|
|---|