TY - JOUR
AU - Logaras, Evangelos
AU - Billis, Antonis
AU - Kyparissidis Kokkinidis, Ilias
AU - Ketseridou, Smaranda Nafsika
AU - Fourlis, Alexios
AU - Tzotzis, Aristotelis
AU - Imprialos, Konstantinos
AU - Doumas, Michael
AU - Bamidis, Panagiotis
PY - 2022
DA - 2022/11/8
TI - Risk Assessment of COVID-19 Cases in Emergency Departments and Clinics With the Use of Real-World Data and Artificial Intelligence: Observational Study
JO - JMIR Form Res
SP - e36933
VL - 6
IS - 11
KW - COVID-19 pandemic
KW - risk assessment
KW - wearable device
KW - respiration evaluation
KW - emergency department
KW - artificial intelligence
KW - real-world data
AB - Background: The recent COVID-19 pandemic has highlighted the weaknesses of health care systems around the world. In the effort to improve the monitoring of cases admitted to emergency departments, it has become increasingly necessary to adopt new innovative technological solutions in clinical practice. Currently, the continuous monitoring of vital signs is only performed in patients admitted to the intensive care unit. Objective: The study aimed to develop a smart system that will dynamically prioritize patients through the continuous monitoring of vital signs using a wearable biosensor device and recording of meaningful clinical records and estimate the likelihood of deterioration of each case using artificial intelligence models. Methods: The data for the study were collected from the emergency department and COVID-19 inpatient unit of the Hippokration General Hospital of Thessaloniki. The study was carried out in the framework of the COVID-X H2020 project, which was funded by the European Union. For the training of the neural network, data collection was performed from COVID-19 cases hospitalized in the respective unit. A wearable biosensor device was placed on the wrist of each patient, which recorded the primary characteristics of the visual signal related to breathing assessment. Results: A total of 157 adult patients diagnosed with COVID-19 were recruited. Lasso penalty function was used for selecting 18 out of 48 predictors and 2 random forest–based models were implemented for comparison. The high overall performance was maintained, if not improved, by feature selection, with random forest achieving accuracies of 80.9% and 82.1% when trained using all predictors and a subset of them, respectively. Preliminary results, although affected by pandemic limitations and restrictions, were promising regarding breathing pattern recognition. Conclusions: This study represents a novel approach that involves the use of machine learning methods and Edge artificial intelligence to assist the prioritization and continuous monitoring procedures of patients with COVID-19 in health departments. Although initial results appear to be promising, further studies are required to examine its actual effectiveness.
SN - 2561-326X
UR - https://formative.jmir.org/2022/11/e36933
UR - https://doi.org/10.2196/36933
UR - http://www.ncbi.nlm.nih.gov/pubmed/36197836
DO - 10.2196/36933
ID - info:doi/10.2196/36933
ER -
pFad - Phonifier reborn
Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies:
Alternative Proxy
pFad Proxy
pFad v3 Proxy
pFad v4 Proxy