Prediction of Cognitive Flexibility Based on Attention Control, Working Memory, and Problem-Solving Skills in Adolescents Using CatBoost
Objective: The present study aimed to predict cognitive flexibility based on attention control, working memory, and problem-solving skills in adolescents using the CatBoost machine learning algorithm.
Methods and Materials: This study was conducted using a quantitative, descriptive-correlational, and predictive design. The statistical population included adolescents studying in secondary schools in Tehran during the 2025–2026 academic year. The final sample consisted of 384 adolescents selected through multistage cluster sampling. Data were collected using standardized self-report instruments measuring cognitive flexibility, attention control, working memory, and problem-solving skills. After preliminary data screening, descriptive indices, reliability coefficients, Pearson correlation coefficients, and machine learning-based predictive analyses were conducted. The CatBoost algorithm was used to predict cognitive flexibility, and the dataset was divided into training and test subsets. Model performance was evaluated using mean absolute error, mean squared error, root mean squared error, coefficient of determination, and five-fold cross-validation. Feature importance analysis was also performed to determine the relative contribution of each predictor.
Findings: The results showed that cognitive flexibility had significant positive correlations with problem-solving skills (r = 0.61, p < 0.01), attention control (r = 0.55, p < 0.01), and working memory (r = 0.48, p < 0.01). The CatBoost model demonstrated acceptable predictive accuracy. In the training set, the model explained 76.5% of the variance in cognitive flexibility, while in the test set, it explained 65.2% of the variance. The cross-validation results also supported the stability of the model, with an average R² of 0.636. Feature importance analysis indicated that problem-solving skills had the highest predictive contribution, followed by attention control and working memory.
Conclusion: The findings indicate that cognitive flexibility in adolescents can be meaningfully predicted by problem-solving skills, attention control, and working memory. The CatBoost algorithm provided an effective predictive model and showed that problem-solving skills were the most important predictor.
