Extracting Medication Nonadherence Reasons with Sentiment-Enriched Deep Learning

Jiaheng Xie, Xiao Liu, Daniel Zeng, Xiao Fang

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Medication nonadherence (MNA) refers to the behavior when patients do not take medications as prescribed. Adverse health outcomes of MNA cost the U.S. healthcare systems $290 billion annually. Understanding MNA and preventing harmful outcomes are an urgent goal for researchers, practitioners, and the pharmaceutical industry. Past years have witnessed rising patient engagement in social media, making it a cost-efficient and heterogeneous data source that can complement and deepen the understanding of MNA. Yet, such dataset is untapped in existing MNA studies. We present the first study to identify MNA reasons from health social media. Health social media analytics studies face technical challenges such as varied patient vocabulary and little relevant information. We develop the Sentiment-Enriched DEep Learning (SEDEL) to address these challenges. We evaluate SEDEL on 53,180 reviews about 180 drugs and achieve an F1 score of 90.18%. SEDEL significantly outperforms state-of-the-art baseline models. This study contributes to IS research in two aspects. First, we formally define the MNA reason mining problem and devise a novel deep-learning-based approach; second, our results provide direct implications for healthcare practitioners to understand patient behaviors and design interventions.

Original languageEnglish (US)
Title of host publicationSmart Health - International Conference, ICSH 2019, Proceedings
EditorsHsinchun Chen, Daniel Zeng, Xiangbin Yan, Chunxiao Xing
Number of pages8
ISBN (Print)9783030344818
StatePublished - 2019
Externally publishedYes
Event7th International Conference for Smart Health, ICSH 2019 - Shenzhen, China
Duration: Jul 1 2019Jul 2 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11924 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference7th International Conference for Smart Health, ICSH 2019


  • Deep learning
  • Health analytics
  • Medication nonadherence
  • Text mining

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)


Dive into the research topics of 'Extracting Medication Nonadherence Reasons with Sentiment-Enriched Deep Learning'. Together they form a unique fingerprint.

Cite this