Identifying Medications that Patients Stopped Taking in Online Health Forums

Patients may stop taking medications after a certain point for various reasons, such as severe side effects, prohibitive costs, or ineffective treatments. Being able to analyze the reason patients stop taking medications is very important to medical practitioners, for example, who can come up with new treatment plans, prescribe different medication if there are side effects. In this paper, we focus on online health forums and define the problem as a binary classification task (i.e., if a patient has stopped taking a medication or not). We chose to focus on health forums here since these are the platforms usually patients go to ask for support online. We propose linguistics features of various complexity and present an in-depth analysis of the results which give us new insights into the task at hand.