Attention Deficit Hyperactivity Disorder (ADHD) arises from a blend of genetic, environmental, and neurological influences. Early diagnosis and tailored treatments can greatly enhance outcomes. However, advancements in computerized detection raise privacy concerns due to unsecured data transmission. While Machine Learning (ML) enhances productivity, it encounters practical challenges. To mitigate these issues, we suggest using Privacy-Preserving Machine Learning (PPML) algorithms to secure ADHD mental health data. This involves preprocessing raw data with methods like missing value imputation and min-max normalization. The Synthetic Minority Over-sampling Technique (SMOTE) tackles class imbalance, notably ADHD patients. Following this, a differential privacy algorithm is implemented along with ML classification algorithms such as K-Nearest Neighbor (KNN), Gradient Boosting (GB), Support Vector Machine (SVM), and Random Forest (RF). Gradient Boosting (GB) stands out for its performance, with a minor accuracy compromise acceptable for ensuring data privacy. This method aims to advance ADHD understanding and management while protecting sensitive information.