Early detection of silent hypoxia in COVID-19 pneumonia using deep learning and IoT

Unlike normal pneumonia, in COVID-19 pneumonia, the shortfall of oxygen occurs without any noticeable breathing difficulties leads to multiple organ failure and death. The early detection of silent hypoxia in COVID-19 pneumonia is the key to save many lives from this deadly disease. This paper has p...

Full description

Saved in:
Bibliographic Details
Published in:Multimedia tools and applications Vol. 83; no. 8; pp. 24527 - 24539
Main Authors: Bandopadhaya, Shuvabrata, Roy, Amarjit
Format: Journal Article
Language:English
Published: New York Springer US 01.03.2024
Springer Nature B.V
Subjects:
Computer Communication Networks
Computer Science
COVID-19
Data collection
Data Structures and Information Theory
Deep learning
Hypoxia
Internet of Things
Mathematical models
Multimedia Information Systems
Oxygen
Oxygen content
Pneumonia
Special Purpose and Application-Based Systems
Track 2: Medical Applications of Multimedia
Deep learning
Internet of things
Silent hypoxia
COVID-19 pneumonia
LSTM encoder-decoder model
Time-series forecasting
ISSN:1573-7721, 1380-7501, 1573-7721
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Unlike normal pneumonia, in COVID-19 pneumonia, the shortfall of oxygen occurs without any noticeable breathing difficulties leads to multiple organ failure and death. The early detection of silent hypoxia in COVID-19 pneumonia is the key to save many lives from this deadly disease. This paper has proposed an e-health solution for early detection of the hypoxia condition of COVID-19 patients using internet of things (IoT) and deep learning techniques. The proposed solution has implemented an IoT framework to collect the percentage of oxygen saturation level in the blood (SpO 2 ) of the patient on real-time basis. It has proposed a time-series forecasting model with deep learning, that being trained with the collected data, forecasts two upcoming SpO 2 readingsefficiently. The model is validated with the SpO 2 level data of 261 hospitalized COVID-19 infected patients with varying level of criticality updated in each 2 hours, and the absolute percentage of errors (APE) in the prediction process has been observed around ~1.56%. The proposed methodology has great potential to control fatality rate in COVID-19 as the early detection of hypoxia helps to initiate the necessary course of action at appropriate time.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-7721
1380-7501
1573-7721
DOI:10.1007/s11042-023-16473-9