Exploring the Ethical Concerns in User Reviews of Mental Health Apps using Topic Modeling and Sentiment Analysis

A new research paper titled 'Exploring the Ethical Concerns in User Reviews of Mental Health Apps using Topic Modeling and Sentiment Analysis' reveals significant gaps in how AI-driven mental health applications address ethical considerations. Researchers Mohammad Masudur Rahman and Beenish Moalla Chaudhry developed an NLP-based framework that analyzed user reviews from both Google Play Store and Apple App Store to systematically evaluate ethical aspects.

The study employed topic modeling to identify latent ethical themes and mapped them against established ethical principles from existing frameworks. Crucially, the researchers also applied a bottom-up approach using a transformer-based zero-shot classification model to detect new and emergent ethical concerns not covered by traditional frameworks. Sentiment analysis was then used to gauge user feelings about each identified ethical aspect.

The 22-page study, submitted to arXiv in February 2026, found that well-known ethical considerations are insufficient for modern AI-based technologies. The results demonstrate how current mental health apps either uphold or overlook key moral values, revealing missing emerging ethical challenges. This work contributes directly to developing ongoing evaluation systems that can enhance fairness, transparency, and trustworthiness in AI-powered mental health chatbots.

Practically, this research provides a methodology for continuous ethical monitoring of mental health applications, addressing growing concerns about user trust in AI-driven therapeutic tools. The framework enables developers and regulators to identify ethical blind spots and improve accountability in a rapidly expanding market where user wellbeing is directly impacted by algorithmic decisions.