<p>Attention deficit hyperactivity disorder (ADHD) is a multifaceted illness that influences brain neurodevelopment. ADHD individuals demonstrate impulsivity, hyperactivity, and inattention. ADHD’s variability, comorbidities, leading to longer wait times, and a global dearth of diagnostic physicians all contribute to its delayed diagnosis. To mitigate this issue the proposed model is divided into three Cases such as supervised, unsupervised, and metaheuristic learning. For all these cases, the collected ADHD dataset is pre-processed using min-max normalization and missing value replacement. In unsupervised learning, the unlabelled dataset is clustered using the BIRCH (Balanced Iterative Reduction and Hierarchical Clustering) method for labeling the class as ADHD. Both the supervised and unsupervised learning datasets are extracted for features using the LDA technique. Then, the metaheuristic learning method uses the Golden Jackal Optimization(GJO) technique to select the optimal features. These features are classified based on the eight machine-learning techniques to evaluate the performance of the model. In this evaluation, the KNN(K-Nearest Neighbor) algorithm performs better for supervised learning and attained an accuracy of 92.7%. Then, the Decision Tree (DT) algorithm performs better for unsupervised and metaheuristic learning with the accuracy of 99.6% and 99.6%. As a result of comparing the values of the designed model using various learning algorithms, improved feature extraction methods and ML(Machine Learning)-based algorithms are developed for detecting ADHD in children.</p>