This study examines user experiences in digital health applications through a hierarchical text-classification framework developed based on a healthcare- specific taxonomy. User reviews of the Sehhaty mobile application were collected from Google Play between 2019 and 2025 and analyzed bilingually in Arabic and English. After data cleaning and preprocessing, a corpus of more than 120,000 reviews was represented using the term frequency–inverse document frequency TF-IDF method.The hierarchical analytical framework organized user feedback into main themes including usability, performance, and support and corresponding subthemes reflecting detailed aspects of user interaction. Two supervised learning models were employed: a margin-based linear classifier for subtheme categorization and a probabilistic regression model for sentiment classification, both configured under few-shot learning conditions using limited labeled data. The thematic classifier achieved 99% accuracy, while the sentiment model obtained an F1- score of approximately 88%, demonstrating robust multilingual performance.Beyond textual features, emoji representations were visualized and analyzed as supplementary affective indicators within the perception layer of the analytical framework. Their inclusion enriched the interpretation of emotional patterns by revealing implicit user sentiments often overlooked in textual reviews. Temporal analysis revealed fluctuations in user satisfaction, with increased negative sentiment during the COVID-19 period due to greater reliance on remote healthcare. Overall, the findings highlight the significance of linguistically grounded computational modeling enriched with hierarchical and multimodal features for systematic monitoring and evidence-based improvement of national digital health infrastructures.