Multi-Level Attention Graph Network of improved Depression Type Detection

Ambika Reghunatha Kurup, Unnikrishnakurup (2025) Multi-Level Attention Graph Network of improved Depression Type Detection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (738kB) | Preview
Preview	PDF (Configuration Manual) Download (546kB) | Preview

Abstract

Subtype classification of depression via social media is still problematic because of the overlapping of symptoms in such depression categories as Major Depression Disorder, Bipolar Depression and Psychotic Depression. In the present study, MAGNET-D (Multi-Level Graph Network of improved Depression Type Detection) is a new graph neural network named after the explicit relationship modeling between symptoms to better pinpoint subtypes, better multi-scale signal folding with Cross-Level Attention and the overall patient’s characterization with Multi-Task Learning. MAGNET-D performs better than the best baseline (Logistic Regression: 83.52% accuracy, 0.84 macro-F1) by 3 percentage points using Multi-Class Depression Detection Dataset (14,983 Twitter posts with the six classes). MAGNET-D also shows consistent performance on all the subtypes and the performance on the following classes is quite strong: Major Depressive (68% F1-score) and Psychotic depression (77% F1-score). Findings confirm the usefulness of the adaptive graph structure to model the relationships between symptoms and improve classification recall and offer more details to recognize mental health in social media text.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Staikopoulos, Athanasios UNSPECIFIED
Additional Information:	Depression subtypes; GNN; Attention; Multi-Task learning; social media texts
Subjects:	Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning R Medicine > RA Public aspects of medicine > RA790 Mental Health Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4150 Computer Network Resources > The Internet > World Wide Web > Websites > Online social networks T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > The Internet > World Wide Web > Websites > Online social networks
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	07 Sep 2026 08:43
Last Modified:	07 Sep 2026 08:43
URI:	https://norma.ncirl.ie/id/eprint/9844

Actions (login required)

View Item