Integrating a hybrid model for a short term Mental Health prognosis

Vijayan, Hariharan (2025) Integrating a hybrid model for a short term Mental Health prognosis. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (7MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Increased rates of stress, anxiety, and depression have resulted in the need to develop mental health screening instruments that would be available to more individuals in a shorter period of time. Conventional evaluation techniques rely on doing it manually and using stagnant questionnaires which require time to respond to and are hard to customize to particular requirements and frequent changes in the mental state which only occur over days or weeks may be overlooked. These issues suggest the usefulness of automated mechanisms that are capable of processing mental health data, and also helping people gain more interesting and personalized means of considering their wellbeing. The proposed study presents a prototype that involves two steps, which include existing machine learning methods and a reflective questioning system fueled by a large language model. By using a large questionnaire data set comprising of DASS questions and demographics, the system will compute the score of depression, anxiety, and stress and categorize each participant into three risk groups. To predict such levels of risks, several supervised learning models such as Random Forest, Gradient Boosting, AdaBoost and XGBoost were trained and tested. The models performed well with the weighted F1 scores of 0.90-0.94, indicating that all the approaches have strong classification. During the second stage, the system takes the level of predicted risk and relevant data features together to produce customized reflective questions and brief narrative understanding with one of the large language models. Although this prototype is not the one that will be used in clinical diagnosis and treatment, it demonstrates the effectiveness of predictive modeling and AI.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Rustam, Furqan UNSPECIFIED
Subjects:	P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning R Medicine > RA Public aspects of medicine > RA790 Mental Health
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	09 Sep 2026 11:06
Last Modified:	09 Sep 2026 11:06
URI:	https://norma.ncirl.ie/id/eprint/9926

Actions (login required)

View Item