| Abstract: | Childhood depression is a critical public health issue that often goes undiagnosed due to subtle symptoms
and limited awareness. This research proposes a machine learning and deep learning based system to
predict childhood depression using supervised learning algorithms and neural network architectures such
as Multi-Layer Perceptron’s (MLP), Gated Recurrent Units (GRU), and Long Short-Term Memory
(LSTM) networks. The study addresses challenges such as the lack of organized datasets and the timeconsuming process of digitizing paper-based records. Feature selection techniques were utilized to identify
the most predictive attributes, while comparative analysis of models ensured the selection of the most
effective approach. Blockchain technology is suggested as an enhancement to improve data security and
transparency, enabling professionals and guardians to monitor mental health status seamlessly. The study
stresses the importance of incorporating real-time datasets to advance the model's accuracy and
responsiveness. The results show that, while promising accuracy was achieved, future research should
explore additional features and larger, more diverse datasets to further improve performance. This system
aims to assist mental health professionals in making timely, data-driven decisions and contribute to the
early identification and management of childhood depression. |
