Lifestyle, psychological and demographic predictors of anxiety: insights from a large-scale survey and machine learning analysis.

Objective: Anxiety is influenced by a combination of lifestyle, psychological, and demographic factors. This study aimed to evaluate these associations and explore the potential of machine learning in predicting anxiety severity. Methods: Anxiety levels were evaluated using a large survey-based dataset of 11, 000 adults alongside demographic, physiological, and psychological measures. Descriptive statistics and inferential analyses were conducted in IBM SPSS to identify associations between key variables. Several machine learning regression algorithms, including linear, regularized, and ensemble models, were implemented in Python to predict anxiety levels. Model performance was evaluated using standard error metrics. Results: Our findings revealed significant associations of anxiety with stress and sleep duration, while demographic attributes such as family history of anxiety and occupation also influenced outcomes. Ensemble machine learning algorithms achieved superior performance compared to single and linear-model approaches. Feature importance analysis identified stress, sleep, and caffeine intake as top predictors of anxiety. Conclusions: The integration of statistical approaches with machine learning applications highlights the multifactorial nature of anxiety and demonstrates the potential of predictive modeling in mental health care. Future research should emphasize longitudinal designs and the incorporation of biological and digital markers to enhance clinical applicability and prediction.