CLASSIFICATION OF ADHD USING TRADITIONAL MACHINE LEARNING ALGORITHMS USING BEHAVIOURAL DATA OF PATIENTS
Accurate diagnosis of attention deficit hyperactivity disorder (ADHD) can be expensive and involves comprehensive assessments like interviews, observations, and evaluations of potential coexisting conditions. With the growing availability of data, there's potential to create machine-learning algorithms that can provide precise diagnostic predictions using cost-effective measures to assist human decision-making. We present the outcomes of employing different classification methods to forecast an ADHD diagnosis that clinicians agree upon. In our proposed study, we evaluate the classification performance of two machine-learning algorithms, Naive Bayes, and K-Nearest Neighbors (KNN), both with and without Principal Component Analysis (PCA). These algorithms are applied to a dataset of 95 samples gathered from open sources. The KNN classifier demonstrates a notable accuracy of 66%, which is significantly higher than the Naïve Bayes accuracy of 55%. Our findings suggest that the KNN classifier performs better in predicting ADHD with improved accuracy.
ADHD, clinicians, K-Nearest Neighbors, machine-learning algorithms, Naive Bayes.