Multimodal Stress Analysis Using Traditional and Deep Learning Models

Sureshbabu, Chethan (2023) Multimodal Stress Analysis Using Traditional and Deep Learning Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (883kB) | Preview

Abstract

The current field of stress analysis demands sophisticated diagnostic systems for accurate assessments, particularly in cardiovascular health. Traditional methods, like Echocardiography (ECG), are often considered time-consuming and require specialized knowledge. This project is a pioneering effort to transform stress analysis by integrating audio, text, and image datasets. The goal is to create a comprehensive framework that can help healthcare professionals, regardless of their specialization, in making swift and accurate stress-related diagnoses. Our approach is designed to simplify stress analysis by employing a multi-data approach. Robust data mining methodologies are developed for extensive medical datasets. A hybrid model, integrating feature selection and classification algorithms, is proposed to identify crucial stress-related features and categorize stress levels. Model performance is evaluated based on accuracy, precision, and F1-score. The integrated model, utilizing audio, image, and text data, effectively identified key stress-related features from different datasets. The Late Fusion model achieved a good classification accuracy of 80%, 80% weighted average F1-score and precision values of 86% and 73% for non-stress and stress classes. The combination of audio, image, and text data showcases the comprehensive nature of stress analysis. Comparative analysis reveals that the model's accuracy (80%) surpasses conventional diagnostic methods (ECG) and aligns with contemporary stress analysis frameworks. This research provides an efficient model for stress analysis, integrating data from different sources to enable healthcare professionals to make stress-related diagnoses more effectively. Future work will focus on fine-tuning the model for optimal performance across varied stress scenarios.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Nagahamulla, Harshani UNSPECIFIED
Uncontrolled Keywords:	Stress Analysis; Multi-Data Approach; Integrated Model; Late Fusion model
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science 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:	23 May 2025 14:26
Last Modified:	23 May 2025 14:26
URI:	https://norma.ncirl.ie/id/eprint/7625

Actions (login required)

View Item