Search

Novel Screening Tool System for Depressive Disorders using Social Media & Artificial Neural Networks

Depression is one of the most common mental health disorders with over 264 million people suffering from it. Improvement in screening depressive disorders can lead to earlier treatment. However, some of the depression screening tools today have constraints and could be difficult to administer due to a lack of cooperation between patient and professional. It is indicated in some studies that there is a correlation between frequent use of social media and increased depression. With this finding, the authors aimed to develop a novel screening tool system incorporated with an artificial neural network that analyzes the patient’s tweets. A design of the screening tool software application was proposed, and an ANN model was developed using a dataset curated from Kaggle. The dataset was cleaned, and features were extracted using the TF-IDF approach. PCA was also used to lessen the number of features for faster training and testing time. Four algorithms were used in training - SVM, Logistic Regression, Perceptron, and KNN. Even though PCA lessens the time for training and testing, it didn’t greatly affect the performance of each model. The SVM model achieved the best performance followed by the Perceptron model. Both the SVM model and the Perceptron model achieved the highest accuracy (98%), but the SVM model achieved better results on the other parameters. However, the SVM model is very slow which prompted the authors to choose the Perceptron model that has a faster speed.


In the Receptiviti LIWC Research Series, we highlight important research conducted using our platform and science that has implications for our customers' and partners' businesses and for society at large:


Read the article


Citation:

Baes, A. M. M., Adoptante, A. J. M., Catilo, J. C. A., Lucero, P. K. L., Peralta, J. F. P. and de Ocampo, A. L. P. (2022) “A Novel Screening Tool System for Depressive Disorders using Social Media and Artificial Neural Network”, International Journal of Intelligent Systems and Applications in Engineering, 10(1), pp. 116–121. doi: 10.18201/ijisae.2022.274.