This study, presented by Neufang et al., aims to identify gender-debiased classifiers for the diagnosis of attention deficit/hyperactivity disorder (ADHD) using machine-learning and bias mitigation algorithms. SVM and XGBoost models were trained using a data sample of 400 children and adolescents with and without ADHD from the Child Mind Institute. They were tested using two datasets consisting of the personal characteristic data, scores of the clinical questionnaire Child Behavior Checklist, and sMRI and fMRI data of 87 participants. Five bias mitigation algorithms from the AI Fairness 360 toolbox were implemented. Bias metrics decreased across all models, with reweighed XGBoost models having the highest accuracy of 88.3% in the fMRI dataset and 85.4% in the sMRI dataset. In the SVM models, accuracy was low with debiasing. The success of reweighed XGBoost models may be due to the way they handle binary variables such as sex as well as the simplicity and directness of reweighing. However, gender-specific features were found to be of low importance, with potential differences in the effects of comorbidities and prefrontal manifestations identified on an individual level but not reliably shown.