Machine Learning for Early Detection of Child Depression: A Data-Driven Approach

Children face trauma due to numerous factors such as parental fights, bullied in school, being below average student, racist comments, physical abuse etc. This trauma deteriorates the child's mental health and may lead to depression as well. Medical field is immensely dedicated to combat the depression among children and one of such approaches adapted includes the use Machine Learning models to predict child depression. This research work is focusing on seven different machine learning algorithms, including SVM Poly, K-Nearest Neighbor, SVM RBF, SVM Linear, Decision Tree, Gradient Boosting, Random Forest and Logistic Regression. This research examines the performance of these algorithms in detecting depressive symptoms in a child and adolescent population. In accordance with the outcomes of the research, SVM Linear algorithm emerged as the most accurate method, with an excellent accuracy rate of 95.54%. This algorithm demonstrated higher sensitivity in detecting depression symptoms in children and teens opening the possibility to prompt intervention and assistance.