Scholarlink

<p>In the context of widespread adoption of healthcare applications, it is important to investigate and understand user experience to improve app performance. This study aims to (i) understand user experience and emotions towards fitness applications; (ii) explore common problems that users encounter; (iii) propose actionable recommendations to enhance app performance. We collected online reviews of Samsung Health and Fitbit on Google Play Store using Google-Play-Scraper, resulting in 196,646 reviews from the COVID-19 period and 60,574 reviews from the post-COVID-19 period. Latent Dirichlet Allocation (LDA) topic modeling with Gensim library in Python was employed to extract hidden topics during and after COVID-19. Moreover, Robustly optimized BERT approach (RoBERTa) model fine-tuned on GoEmotions dataset was performed to detect user emotions associated with each topic. As a result, we identified four topics for each period which highlighted significant technical issues such as inaccurate tracking mechanism, unstable data synchronization and server errors. During COVID-19, users emphasized not only their compliments but also privacy concerns about access permissions for phone data. After COVID-19, topics regarded users’ complaints about unfriendly interface and removal of fitness challenges. Additionally, emotions of disappointment and disapproval were mainly expressed in both the COVID-19 and post-COVID-19 periods, indicating loss of trust and motivation for app usage because of technical errors. Therefore, our findings suggest app owners and development teams should prioritize solving recurring errors that users encounter, ensuring enhancement in app performance. This research demonstrates the application of topic modeling and emotion recognition to identify actionable strategies for healthcare apps improvement.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Applying topic modeling and emotion detection to understand user experience with healthcare applications: a comparative analysis during and after COVID-19

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Thi Thao Van Ho,
  • Mi Jin Noh

摘要

In the context of widespread adoption of healthcare applications, it is important to investigate and understand user experience to improve app performance. This study aims to (i) understand user experience and emotions towards fitness applications; (ii) explore common problems that users encounter; (iii) propose actionable recommendations to enhance app performance. We collected online reviews of Samsung Health and Fitbit on Google Play Store using Google-Play-Scraper, resulting in 196,646 reviews from the COVID-19 period and 60,574 reviews from the post-COVID-19 period. Latent Dirichlet Allocation (LDA) topic modeling with Gensim library in Python was employed to extract hidden topics during and after COVID-19. Moreover, Robustly optimized BERT approach (RoBERTa) model fine-tuned on GoEmotions dataset was performed to detect user emotions associated with each topic. As a result, we identified four topics for each period which highlighted significant technical issues such as inaccurate tracking mechanism, unstable data synchronization and server errors. During COVID-19, users emphasized not only their compliments but also privacy concerns about access permissions for phone data. After COVID-19, topics regarded users’ complaints about unfriendly interface and removal of fitness challenges. Additionally, emotions of disappointment and disapproval were mainly expressed in both the COVID-19 and post-COVID-19 periods, indicating loss of trust and motivation for app usage because of technical errors. Therefore, our findings suggest app owners and development teams should prioritize solving recurring errors that users encounter, ensuring enhancement in app performance. This research demonstrates the application of topic modeling and emotion recognition to identify actionable strategies for healthcare apps improvement.