HPRXF Model: An Ensemble Transfer Learning-based Fusion model for handling Pandemic-related Calls received by the Emergency Response Support System

The COVID-19 epidemic has surely disrupted daily life and resulted in unexpected shifts that have resulted in serious psychological reactions and mental health crises. In the early days of COVID-19, little was known about it, which understandably generates worry and distress, especially for those of...

Full description

Saved in:
Bibliographic Details
Published in:Journal of ambient intelligence and humanized computing 2024-03, Vol.15 (3), p.2035-2046
Main Authors: Nimmi, K., Janet, B., Kalai selvan, A., Sivakumaran, N.
Format: Article
Language:English
Subjects:
Artificial Intelligence
Computational Intelligence
Coronaviruses
COVID-19
Datasets
Emergency communications systems
Emergency response
Engineering
Natural language processing
Original Research
Pandemics
Public health
Robotics and Automation
Social networks
Support systems
User Interfaces and Human Computer Interaction
Viral diseases
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The COVID-19 epidemic has surely disrupted daily life and resulted in unexpected shifts that have resulted in serious psychological reactions and mental health crises. In the early days of COVID-19, little was known about it, which understandably generates worry and distress, especially for those of us who suffer from anxiety disorders. In the early phases of the pandemic, there were large volume of calls to emergency lines, thus it was critical to divert non-emergency inquiries regarding the disease, particularly those looking for information, and to offer help for emergency calls about “breathing difficulty" and “person collapsing". It became vital to distinguish between emergency and non-emergency calls. This study examines the call content of calls landed in the emergency response support system during the pandemic and help in categorising the calls as “emergency" or “non-emergency" based on call content related to pandemic. The proposed model is a Hadamard product-based pretrained ensemble model of RoBERTa and XLNet using the ERSS dataset. The performance of the HPRXF proposed model is evaluated using the Hadamard product of the final probability vectors of the RoBERTa and XLNet models. The HPRXF Model outperforms existing models with an accuracy: 81.16% and an F1-score: 80.66%.
ISSN:1868-5137
1868-5145
DOI:10.1007/s12652-023-04690-x