<![CDATA[The anxiety levels of patients in the Emergency Department (ED) have a significant impact on medical management and care, making accurate early detection crucial. This study aims to develop and evaluate a machine learning-based classification model to predict patient anxiety levels into four categories: Normal (0), Mild (1), Moderate (2), and Severe (3), using the CatBoost Classifier algorithm. This approach utilizes physiological parameters such as systolic and diastolic blood pressure, heart rate, respiratory rate, as well as demographic variables and hypertension history. The data were trained and validated through appropriate dataset splitting, with comprehensive evaluation using accuracy, precision, recall, F1-score metrics, and learning curve analysis to assess model generalization. The evaluation results showed very high performance, with an overall accuracy reaching 99%. The model provides remarkable consistency across all classes: for the Normal class (0), precision, recall, and F1-score reached 1.00; the Mild class (1) achieved 0.98, 0.99, and 0.99; the Moderate class (2) each reached 0.98; and the Severe class (3) reached 1.00, 0.99, and 0.99. The learning curve indicates no overfitting and the model's ability to learn effectively as the amount of training data increases. Feature importance analysis confirms that both systolic and diastolic blood pressure are dominant predictors of anxiety levels, in line with observed patient physiological responses. Overall, the CatBoost model is proven to be highly reliable and holds great potential as a clinical decision support system that can assist healthcare professionals in the early detection and management of patient anxiety in a dynamic ER environment.]]>
