Multi-Aspect Deep Active Attention Network for Healthcare Explainable Adoption.

Area of Science:

• Computational linguistics
• Mental health informatics
• Artificial intelligence in healthcare

Background:

• Depression significantly impacts individuals' lives and well-being.
• Detecting depression often relies on subjective assessments.
• Emerging research explores using online user-generated data for mental health analysis.

Purpose of the Study:

• To develop an effective method for detecting depression from social media text.
• To overcome limitations in current text classification models for mental disorders.
• To improve the accuracy and interpretability of depression detection systems.

Main Methods:

• Utilized attention networks with self-attention mechanisms for text analysis.
• Extended emotion lexicons by incorporating hypernyms to capture nuanced emotional expressions.
• Applied the model to unlabeled forum data for symptom identification.

Main Results:

• Achieved an Area Under the Curve (ROC) of 0.87 for depression detection.
• Demonstrated superior performance compared to existing methods.
• Maintained interpretability and transparency in the model's decision-making process.

Conclusions:

• The proposed attention network model effectively identifies depression symptoms from online text.
• Combining enhanced emotion lexicons with attention networks offers a promising approach for mental health monitoring.
• The method facilitates the identification of depression from internet forum data with increased accuracy.