User Profile Detection in Health Online Fora

Area of Science:

• Health Informatics
• Social Media Analysis
• Patient Behavior Studies

Background:

• Online patient forums offer valuable insights into patient understanding, behavior, and needs.
• Analyzing user profiles within these forums is crucial for interpreting discussions effectively.
• Previous methods lacked comprehensive user attribute identification.

Purpose of the Study:

• To develop and evaluate an approach for identifying user profiles in diabetic patient online discussions.
• To determine the accuracy of identifying age, gender, disorders, and patient-contributor relationships.

Main Methods:

• A hybrid approach combining lexicon-based methods and supervised classifiers was employed.
• The system was trained to identify user demographics (age, gender) and health-related attributes.
• Performance was evaluated based on precision for each identified attribute.

Main Results:

• The developed method achieved 100% precision in identifying the gender of discussion contributors.
• Precision for identifying patient disorders ranged from 53.48% depending on the parameters used.
• The approach demonstrated feasibility in extracting user profile information from online health discussions.

Conclusions:

• Combining lexicon and supervised classifiers is effective for user profiling in diabetic online communities.
• Accurate gender identification is achievable, while disorder identification requires further refinement.
• This methodology can enhance the understanding of patient needs and behaviors in digital health spaces.