Fitri Marisa, Giva Andriana Mutiara, Deshinta Arrova Dewi, Sharifah Sakinah Syed Ahmad, Agustinus Noertjahyana, Anastasia L Mauka
Mobile-based mental health applications are increasingly adopted; however, their real-world effectiveness remains constrained by low user retention, inconsistent gamification experiences, and variability in service quality, including paywalls, bugs, and system performance issues. This study aims to systematically examine how user experience aspects and gamification elements are associated with review sentiment, while assessing whether these patterns generalize across applications. This study proposes a review-driven analytical framework that integrates sentiment classification with multi-label annotation for experiential aspects, gamification elements, and social cues, with a specific analytical focus on social friction/pressure in the reported moderation results. The dataset comprises 10,040 user reviews collected from 16 mental health applications. To obtain a more realistic evaluation and mitigate app-specific leakage, a 5-fold GroupKFold strategy is implemented within a leakage-safe pipeline. In addition, a confidence-aware explainability protocol is introduced to extract representative review examples as auditable interpretation traces. The main results indicate that the TF-IDF + Logistic Regression baseline provides adequate cross-application generalization for large-scale analysis under an exact gold-only by-application evaluation setting, while additional linear-baseline comparisons confirm that the overall task-difficulty pattern remains stable across alternative conventional models. In contrast, social cues remain the most challenging task due to their contextual nature and relative sparseness. Substantively, the observed aspect × element × sentiment patterns suggest that negative sentiment attributed to gamification often emerges when gamification elements intersect with monetization mechanisms and usage friction, rather than from the gamification elements themselves. These findings support preliminary design implications, including reducing coercive reward mechanisms, adopting more flexible streak designs, limiting explicit competition through opt-out options, and prioritizing personalization and progress feedback. The main contribution of this study is to provide a more defensible and auditable cross-application evaluation framework for linking gamification, user experience, and social context in mental health applications. This article is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. License details: https://creativecommons.org/licenses/by-sa/4.0/
Informatics Engineering Department, Universitas Widya Gama Malang, Indonesia; Department of Applied Science, Telkom University, Indonesia; Center for Data Science and Sustainable Technologies, INTI International University, Malaysia; Faculty of Artificial Intelligence and Cyber Security, Universiti Teknikal Malaysia Melaka, Malaysia; Informatics Department, Petra Christian University, Indonesia; Industrial Engineering Department, President University, Indonesia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock