Objective: To analyze community health forums such as MedHelp, Reddit, Facebook, etc., to obtain numerous questions and answers about opioid use disorder. In addition, categorize 180 posts from various internet-based community forums and compare with categorizations from clustering algorithm.
Methods:
Methods:
- Manually evaluate each of the 180 posts and categorize them into themes.
- Analyze how many posts are in each final category, outline what the new categories are, and compare those results with the software output.
Results:
- Our data set includes 180 original, unedited posts, with an average of 156 words per post and 821 characters per post. Six clusters were given by the software package. Each cluster includes 30 posts.
Conclusion: In conclusion, the 180 posts can be categorized into seven broad categories: no question, asking for advice, can anyone relate, withdrawal, clarifying question, recommendation, and other, with four sub-categories stemming from the advice and withdrawal sections. These are advice - withdrawal, advice - medications, withdrawal - symptoms, and withdrawal - remedies. The clustering algorithm seemed to perform well, for the most part. Some categories performed exceptionally well and were placed in the same cluster and other categories had sporadic results.