Identifying Insomnia From Social Media Posts: Psycholinguistic Analyses of User Tweets

Area of Science:

• Computational linguistics
• Psychological informatics
• Social media analytics

Background:

• Insomnia is a prevalent sleep disorder affecting many individuals globally.
• Social media platforms offer a rich source of data for health condition diagnosis, including insomnia.
• Existing research has gaps in predicting insomnia from linguistic patterns and personality traits on social media.

Purpose of the Study:

• To develop an insomnia prediction model utilizing psycholinguistic patterns from social media.
• To analyze word usage, semantics, and Big 5 personality traits in relation to insomnia.

Main Methods:

• Extracted psycholinguistic profiles based on word choice and semantic relationships from tweets.
• Investigated correlations between users' Big 5 personality traits and insomnia.
• Developed a double-weighted ensemble classification model for insomnia prediction.

Main Results:

• The classification model demonstrated a strong prediction potential of 78.8% for insomnia.
• Insomniacs exhibited distinct word usage patterns, including increased use of negative, anxious, and sad language.
• High neuroticism and conscientiousness scores were significantly correlated with insomnia, while extraversion showed a negative correlation.

Conclusions:

• The developed model effectively predicts insomnia from social media interactions.
• Integration into software can aid in early detection of insomnia by family and healthcare professionals.
• This approach offers a novel method for identifying potential insomnia sufferers through their online communication.