A deep learning model for detecting mental illness from user content on social media

Area of Science:

• Computational linguistics
• Psychiatry
• Machine learning

Background:

• Social media platforms are venues for users to express emotions and mental states.
• Identifying mental health conditions through user-generated content presents a significant challenge.

Purpose of the Study:

• To develop and evaluate a deep learning model for identifying mental health disorders from social media posts.
• To assess the model's accuracy in classifying specific conditions such as depression, anxiety, bipolar disorder, borderline personality disorder, schizophrenia, and autism.

Main Methods:

• Collected posts from mental health communities on Reddit.
• Developed a deep learning model to analyze user posting information.
• Trained the model to identify patterns indicative of specific mental disorders.

Main Results:

• The deep learning model accurately identified users' posts belonging to specific mental disorders.
• The model demonstrated proficiency in classifying conditions including depression, anxiety, bipolar disorder, borderline personality disorder, schizophrenia, and autism.

Conclusions:

• The proposed model can aid in identifying individuals potentially experiencing mental illness based on their social media activity.
• This model can serve as a supplementary tool for monitoring the mental health of frequent social media users.