Unsupervised Machine Learning to Identify Positive and Negative Themes in Jordanian mHealth Apps

Abstract

User opinions are crucial in the development of mobile health (mHealth) applications (apps). This study aimed to investigate and qualitatively assess consumer attitudes toward mHealth apps and the main aspects of their design. The methodology was divided into four steps: (1) data collection, (2) preprocessing, (3) sentiment analysis by valence-aware dictionary and sentiment reasoner (VADER), and (4) thematic analysis by the latent Dirichlet allocation (LDA) algorithm. These steps were implemented in 836 reviews of eight mHealth apps on app stores in Jordan. The current study offers healthcare stakeholders insight into the positive and negative aspects of mHealth apps by identifying user-preferred features and recommending improvements. The findings indicate several aspects of design that mHealth app developers may use to improve overall efficacy, including user experience, client services, usability, and adherence.