Abstract
BACKGROUND: Although the Screen for Child Anxiety Related Emotional Disorders (SCARED) is a widely used tool for assessing anxiety, its 41-item format makes it a time-intensive method for identifying children and adolescents at high risk of anxiety. This study aims to develop an optimized version of the SCARED for Chinese children and adolescents using a novel machine learning approach, Fast and Accurate Interpretable Risk Scores (FasterRisk), to improve the efficiency of prediction and intervention. METHOD: The full version of the SCARED scale and sociodemographic information were given to 8,315 children and adolescents aged 4-9 years in Henan Province, China. The FasterRisk model was utilized to select the optimal items for constructing the Chinese version of SCARED, and receiver operating characteristic (ROC) curves were employed to determine the optimal cutoff scores. RESULTS: The results showed that a 5-item Chinese version of the SCARED accurately reproduced full SCARED scores. By evaluating the performance of risk scoring models containing 1 to 8 items, the 5-item model showed the best performance in AUC (0.96), and other performance indicators, with high prediction accuracy (R²= 0.82). Under the condition of an equal number of items, the AUC value of the newly developed 5-item Chinese version of the SCARED (0.96) surpassed that of the existing SCARED-5 (0.92), with the optimal cutoff score determined to be 14. CONCLUSION: The 5-item Chinese version of the SCARED is a reliable self-report tool that aids users with limited time and resources in assessing anxiety among children and adolescents in China. TRIAL REGISTRATION: Ethical approval in this study was approved by the Ethics Committee for Social Development and Public Policy at Beijing Normal University (SSDPP-HSC20230014).