Described is a system for personalizing a human-machine interface (HMI) device based on a mental and physical state of a user. During performance of a task in a simulation environment, the system extracts biometric features from data collected from body sensors, and extracts brain entropy features from electroencephalogram (EEG) signals. The brain entropy features are correlated with the biometric features to generate a mental-state model. The mental-state model is deployed in a HMI device during performance of the task in an operational environment for continuous adaptation of the HMI device to its user's mental and physical states.