Mobile Application for Managing Anxiety using Machine Learning Model

Alexandra A.M.M.; Gabriel G.D.C.; Elio N.V. · 2025 3rd Cognitive Models and Artificial Intelligence Conference, AICCONF 2025 - Proceedings (2025) · DOI: 10.1109/aicconf64766.2025.11064189

Resumen

Anxiety before and after cardiac surgery is a critical factor that can negatively affect patients' recovery and clinical outcomes. This study presents Kora, a mobile application that combines mobile health (mHealth) strategies with a machine learning model to recommend relaxation techniques tailored to each patient's anxiety level. The proposed system uses physiological data - such as heart rate, blood oxygen saturation, and sleep duration - along with State-Trait Anxiety Inventory assessments (STAI), to classify users into anxiety categories and recommend appropriate interventions (music therapy, guided meditation, or breathing exercises). Three machine learning models - Random Forest, Support Vector Machine, and Gradient Boosted Trees - were trained and evaluated using the LifeSnaps dataset (7,411 records), with Gradient Boosted Trees achieving the highest performance, reaching 80.76% accuracy and 82.33% F1-score. Additionally, validation with clinical experts confirmed the app's usability and relevance in healthcare settings. These results suggest that Kora is a promising digital tool to support the emotional well-being of cardiac surgery patients through personalized relaxation techniques.

Instituciones: Universidad Peruana de Ciencias Aplicadas