Understanding Specific Learning Disorders in youth: A machine learning approach to socio-emotional and cognitive aspects

Le résumé fourni par la source

Youth with Specific Learning Disorders (SLD) experience not only academic and cognitive difficulties, but also socio-emotional challenges. Whether socio-emotional factors contribute to SLD identification remains unclear. No studies have jointly examined cognitive and socio-emotional variables using supervised machine learning. This study included youth aged 8–16 with SLD ( N = 70) versus non-diagnosed peers ( N = 130), matched for age and sex, who completed a working memory task and questionnaires assessing test anxiety, social anxiety, cognitive emotion regulation, and loneliness. Support Vector Machine showed high specificity (88.5%) but low sensitivity (44.3%). Nonetheless, overall performance exceeded chance (Balanced Accuracy = 66.4%; AUC = 0.74), indicating reliable multivariate discrimination. Higher test and social anxiety, peer- and parent-related loneliness, and maladaptive emotion regulation strategies characterized SLD, whereas stronger working memory, adaptive regulation, and affinity for aloneness were associated with non-diagnosed youth. Findings underscore the value of integrating cognitive and socio-emotional factors in SLD assessment. Our study found that high levels of test anxiety - particularly intrusive thoughts - along with feelings of loneliness, social anxiety, and the use of specific emotion regulation strategies (e.g., acceptance, catastrophizing) are associated with Specific Learning Disorders (SLD). These findings suggest that socio-emotional and cognitive factors together play an important role in identifying and supporting students with SLD. On a practical level, the results highlight the need for educators to address both learning challenges and emotional well-being through integrated support programs. Helping students manage anxiety, build adaptive coping strategies and positive relationships may improve both their academic performance and their overall school experience. • Machine learning was applied to SLD using socio-emotional and cognitive factors. • Support Vector Machine seems to outperform traditional univariate methods. • Socio-emotional factors improved SLD prediction at the individual level. • Findings support multidimensional assessment and intervention for SLD.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.