Understanding Medication Nonadherence from Social Media: A Sentiment-Enriched Deep Learning Approach

Medication nonadherence (MNA)can leadto serioushealthramificationsandcosts U.S. healthcare systems$290 billion annually. Understanding the reasons underlying patients’ MNA is thus an urgent goal for researchers, practitioners, andthepharmaceuticalindustry inordertomitigatenegativehealthandeconomicconsequences. In recent years, patient engagementon socialmedia sites has soared, making it a cost-efficient andrich information source that can complement prior survey studies and deepen the understanding of MNA. Yet these data remain untapped in existing MNA studies because of technical challenges such as long texts, decision-making based on negativesentiment,variedpatientvocabulary,andthescarcityofrelevantinformation.Forthisstudy,wedeveloped asentiment-enricheddeeplearningmethod(SEDEL)toaddressthesechallengesandextractreasonsforMNA. We evaluated SEDEL using 53,180 reviews concerning 180 drugs and achieved a precision of 89.25%, a recall of 88.48%, and an F1 score of 88.86%. SEDEL significantly outperformed state-of-the-art baseline models. We identified nine categories of MNA reasons, which were verified by domain experts. This study contributes to IS researchbydevisinganoveldeep-learning-basedapproachforreasonminingandbyprovidingdirectimplications forthehealthindustryandforpractitionersregardingthedesignofinterventions.