Neurofibromatosis type 1 (NF1) is one of the most prevalent rare genetic diseases, confronting patients with numerous medical (neurological, cutaneous, osseous) and neuropsychological symptoms. To identify risk factors and allow for targeted neuropsychological intervention, potential associations between the heterogenous somatic and cognitive phenotype have been proposed.The present study harnesses the benefits of machine learning analysis to investigate potential non-/linear associations between the medical and neuropsychological profiles of n=236 pediatric NF1 patients. A total of 14 models was constructed to predict objective, behavioral, and parent-rated measures of 7 distinct attention domains based on medical characteristics (eg. CNS tumors, focal MRI signal intensity).While the young patients’ attention scores differed significantly from the expected, normative values, only behaviorally observed measures of sustained attention, hyperactivity and observed attention problems could be predicted from the chosen biological factors. Additional information on the location of neurological alterations did not significantly improve the predictions. The most important predictors were younger age and male sex, with significant age-sex interactions in sustained attention and overall observed attention deficits.These findings highlight the relevance of attention deficits in pediatric NF1 patients, especially in young boys and adolescent girls. However, the results indicate the absence rather than the non-linearity of associations between gross NF1-related neurological characteristics and attention skills. Therefore, it is recommended to prioritize age and sex over neurological characteristics as risk factors when planning neuropsychological interventions to prevent psychosocial consequences of attention deficits on the quality of life of young patients with NF1.