Toward Technostress Analysis in Menstrual Apps: An NLP Study in Indonesia
Year: 2026
Type: conference-paper
Abstract: Menstrual tracking applications are widely used digital tools for monitoring reproductive health and fertility windows. Despite their utility, inconsistencies in performance and usability may induce technostress, a psychological strain that arises from technological demands. To explore this, we propose an automated Natural Language Processing (NLP) pipeline to extract user complaints and negative sentiments as potential indicators of technostress. The analysis draws on 40,873 user reviews collected from 10 popular menstrual tracking apps on the Google Play Store up to 2026. Following standard text preprocessing, Latent Dirichlet Allocation (LDA) topic modeling was applied to extract latent themes from the corpus. The model identified five dominant aspects: prediction accuracy and calendar, health education and tracking, technical issues and account access, pregnancy planning and fertility, and user experience and language localization. An examination of the negative reviews indicates that inaccurate predictions, technical disruptions, and interface usability are the main sources of user dissatisfaction. These factors theoretically map onto specific technostress dimensions, namely techno-uncertainty and techno-complexity. For sentiment classification, a Support Vector Machine (SVM) model, evaluated with an 80:20 train-test split, achieved the highest performance relative to other baseline algorithms with an overall accuracy of 75.18%. These findings clarify common pain points in reproductive health applications, and the proposed pipeline offers a scalable methodological foundation for developers and researchers to investigate technostress triggers and support the design of more stress-aware and user-centered digital health tools.
Institution: Binus University
Language: English
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7216545166