Age and Gender Prediction on Health Forum Data

Age and Gender Prediction on Health Forum Data Prasha Shrestha1, Steven Bethard2, Ted Pedersen3, Nicolas Rey-Villamizar1, Farig Sadeque2, and Thamar Solorio1 1University of Houston, 2University of Alabama at Birmingham, 3University of Minnesota, Duluth pshrestha3@uh.edu, bethard@uab.edu, tpederse@d.umn.edu, nrey@uh.edu, farigys@uab.edu, solorio@cs.uh.edu Abstract Health support forums have become a rich source of data that can be used to improve health care outcomes. A user profile, including information such as age and gender, can support targeted analysis of forum data. But users might not always disclose their age and gender. It is desirable then to be able to automatically extract this information from users’ content. However, to the best of our knowledge there is no such resource for author profiling of health forum data. Here we present a large corpus, with close to 85,000 users, for profiling and also outline our approach and benchmark results to automatically detect a user’s age and gender from their forum posts. We use a mix of features from a user’s text as well as forum specific features to obtain accuracy well above the baseline, thus showing that both our dataset and our method are useful and valid. Keywords: medical forums, author profiling, gender, age 1. Introduction Users who are actively engaged on health support forums volunteer rich personal details including disease symptoms, treatment history, medications, family history, and traces of their emotional status. Some active users might even have their entire medical history contained among these posts and replies. Since these forums have a certain level of anonymity, people might be more willing to talk candidly about their conditions. People going through similar situations as the original poster may also open up. Some people might come to the forums directly before going to their doctors, while others might create a post after a visit to their doctor. A survey of users of the DailyStrength health forum found that 160 out of 274 users went to DailyStrength after visiting their doctor (Bell et al., 2011). Moreover, out of the people who gave a perfect trust score to their doctor, more than half turned to the Internet for additional information and support. Although the suggestions provided in these forums as a means of treatment might not be appropriate (Culver et al., 1997), there is no doubt that there is valuable information in them. For example, patient’s side effects or symptoms during the course of a treatment could be extracted from the posts and therefore complement the information available for physicians. Another relevant task is to identify patients for clinical trials since these communities are so widespread across different age groups, ethnicities, socio-economic statuses and regions of the world. To identify potential patients, basic inclusion- ary criteria have to be met, such as age and gender. But not all users will provide this information publicly (in our data 35% did not include age or gender). User profiling in this data can then solve this problem and also enable aggre- gated studies of user diseases and symptoms, as well as their online …

Content truncated to 500 words