Adaptive feature selection and dependency aware multi head attention transformer for psychiatric disorder classification

Key Points

• Improved classification accuracy for psychiatric disorders using adaptive feature selection, enhancing method robustness.
• Key evidence shows a notable increase in performance metrics, with multi-head attention providing significant advantages.
• This analysis employed a dependency aware multi-head attention transformer to identify relevant features systematically.
• The findings may enable more accurate diagnoses, although further validation in varied clinical settings is necessary.