STCWTNet: Spatio-temporal channel-wise transformer network for interpretable cross-subject mild cognitive impairment detection using EEG signals

Key Points

• Mild cognitive impairment detection achieved high accuracy with EEG signals—indicating potential for early diagnosis.
• Accuracy metrics reached over 90% in identifying cognitive impairment from EEG signals in the tested population.
• Analysis employed a novel spatio-temporal transformer approach with channel-wise attention mechanisms for better performance.
• Supports the use of EEG data as a reliable tool for diagnosing mild cognitive impairment, though broader testing may be needed.