The impact of AI-based fitness applications on perceived stress in university students

Abstract

Introduction. Perceived stress among university students is influenced by lifestyle behaviors and the increasing use of digital health technologies. AI-based fitness applications are widely used to promote healthy habits; however, their effectiveness in reducing stress remains unclear.

Objective. This study aimed to examine the relationship between AI-based fitness application use, lifestyle factors, and perceived stress among university students, and to compare the predictive performance of statistical and machine learning models.

Methods. A cross-sectional study was conducted with 84 students. Data were collected using an online questionnaire assessing perceived stress (PSS-10), physical activity (IPAQ-SF), sleep duration, screen time, nutrition, and AI application usage. Descriptive statistics, Pearson correlations, and multiple regression analyses were performed. Machine learning models (Random Forest, Support Vector Machine, and K-Nearest Neighbors) were also applied.

Results. Participants reported moderate levels of perceived stress. Correlations between stress and lifestyle variables were weak. The regression model was not statistically significant (R² = 0.09), while machine learning models showed slightly improved performance, with Random Forest explaining 21% of the variance.

Conclusions. Perceived stress appears to be weakly associated with lifestyle behaviors and AI application use. The findings suggest that stress is multifactorial and requires more comprehensive and personalized approaches beyond behavioral factors alone.