A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters
Autor(a) principal: | |
---|---|
Data de Publicação: | 2024 |
Outros Autores: | , , , , , , , , , , , , , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
Texto Completo: | http://hdl.handle.net/10773/39320 |
Resumo: | Assessing the health status of critically ill patients with COVID-19 and predicting their outcome are highly challenging problems and one of the reasons for poor management of ICU resources worldwide. A better pathophysiological understanding of patients’ state evolution in the ICU can enhance effective medical interventions. Therefore, there is a need to monitor and analyze the pulmonary function of a ICU patient with COVID-19 and its impact on cardiovascular and other systems. To achieve this, chest X-rays (CXRs), respiratory sounds and all the routinely monitored parameters, scores and metrics in the COVID-19 ICU were recorded from 171 ICU patients with COVID-19 from June 2020 until December 2021. Features were extracted from respiratory sounds, deep learning analysis was conducted on CXRs, and logistic regression analysis was performed on routine ICU clinical variables. Deep learning pipelines were established to classify patients’ outcomes (survival or death) at two time points (ICU mortality or 90-day mortality) using three input configurations: (a) CXRs, (b) a fusion of CXRs and respiratory sounds features, or (c) a fusion of CXRs, respiratory sounds features, and principal features of the ICU clinical measurements. The performance of the latter approach was promising, achieving, for ICU mortality, an accuracy of 0.761 and an AUC of 0.759, and for 90-day mortality, an accuracy of 0.743 and an AUC of 0.752, while the performance of approaches (a) and (b) was worse. Therefore, using multi-source data and longitudinal COVID-19 ICU data offers a better prediction of the outcome in the ICU, thereby optimizing medical decisions and interventions. Furthermore, we show that adding the adventitious respiratory sounds features significantly increased AUC and accuracy for mortality prediction of ICU patients with COVID-19. |
id |
RCAP_6ce6f4a987054e32c0506eed1a7a2c9f |
---|---|
oai_identifier_str |
oai:ria.ua.pt:10773/39320 |
network_acronym_str |
RCAP |
network_name_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository_id_str |
7160 |
spelling |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parametersCOVID-19Deep learning fusionRespiratory soundsClinical variablesChest X-raysICU mortalityAssessing the health status of critically ill patients with COVID-19 and predicting their outcome are highly challenging problems and one of the reasons for poor management of ICU resources worldwide. A better pathophysiological understanding of patients’ state evolution in the ICU can enhance effective medical interventions. Therefore, there is a need to monitor and analyze the pulmonary function of a ICU patient with COVID-19 and its impact on cardiovascular and other systems. To achieve this, chest X-rays (CXRs), respiratory sounds and all the routinely monitored parameters, scores and metrics in the COVID-19 ICU were recorded from 171 ICU patients with COVID-19 from June 2020 until December 2021. Features were extracted from respiratory sounds, deep learning analysis was conducted on CXRs, and logistic regression analysis was performed on routine ICU clinical variables. Deep learning pipelines were established to classify patients’ outcomes (survival or death) at two time points (ICU mortality or 90-day mortality) using three input configurations: (a) CXRs, (b) a fusion of CXRs and respiratory sounds features, or (c) a fusion of CXRs, respiratory sounds features, and principal features of the ICU clinical measurements. The performance of the latter approach was promising, achieving, for ICU mortality, an accuracy of 0.761 and an AUC of 0.759, and for 90-day mortality, an accuracy of 0.743 and an AUC of 0.752, while the performance of approaches (a) and (b) was worse. Therefore, using multi-source data and longitudinal COVID-19 ICU data offers a better prediction of the outcome in the ICU, thereby optimizing medical decisions and interventions. Furthermore, we show that adding the adventitious respiratory sounds features significantly increased AUC and accuracy for mortality prediction of ICU patients with COVID-19.Elsevier2024-012024-01-01T00:00:00Z2026-01-31T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/articleapplication/pdfhttp://hdl.handle.net/10773/39320eng0957-417410.1016/j.eswa.2023.121089Wu, YunanRocha, Bruno MachadoKaimakamis, EvangelosCheimariotis, Grigorios-ArisPetmezas, GeorgiosChatzis, EvangelosKilintzis, VassilisStefanopoulos, LeandrosPessoa, DiogoMarques, AldaCarvalho, PauloPaiva, Rui PedroKotoulas, SerafeimBitzani, MilitsaKatsaggelos, Aggelos K.Maglaveras, Nicosinfo:eu-repo/semantics/embargoedAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2024-02-22T12:16:40Zoai:ria.ua.pt:10773/39320Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-20T03:09:29.038946Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse |
dc.title.none.fl_str_mv |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
title |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
spellingShingle |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters Wu, Yunan COVID-19 Deep learning fusion Respiratory sounds Clinical variables Chest X-rays ICU mortality |
title_short |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
title_full |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
title_fullStr |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
title_full_unstemmed |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
title_sort |
A deep learning method for predicting the COVID-19 ICU patient outcome fusing X-rays, respiratory sounds, and ICU parameters |
author |
Wu, Yunan |
author_facet |
Wu, Yunan Rocha, Bruno Machado Kaimakamis, Evangelos Cheimariotis, Grigorios-Aris Petmezas, Georgios Chatzis, Evangelos Kilintzis, Vassilis Stefanopoulos, Leandros Pessoa, Diogo Marques, Alda Carvalho, Paulo Paiva, Rui Pedro Kotoulas, Serafeim Bitzani, Militsa Katsaggelos, Aggelos K. Maglaveras, Nicos |
author_role |
author |
author2 |
Rocha, Bruno Machado Kaimakamis, Evangelos Cheimariotis, Grigorios-Aris Petmezas, Georgios Chatzis, Evangelos Kilintzis, Vassilis Stefanopoulos, Leandros Pessoa, Diogo Marques, Alda Carvalho, Paulo Paiva, Rui Pedro Kotoulas, Serafeim Bitzani, Militsa Katsaggelos, Aggelos K. Maglaveras, Nicos |
author2_role |
author author author author author author author author author author author author author author author |
dc.contributor.author.fl_str_mv |
Wu, Yunan Rocha, Bruno Machado Kaimakamis, Evangelos Cheimariotis, Grigorios-Aris Petmezas, Georgios Chatzis, Evangelos Kilintzis, Vassilis Stefanopoulos, Leandros Pessoa, Diogo Marques, Alda Carvalho, Paulo Paiva, Rui Pedro Kotoulas, Serafeim Bitzani, Militsa Katsaggelos, Aggelos K. Maglaveras, Nicos |
dc.subject.por.fl_str_mv |
COVID-19 Deep learning fusion Respiratory sounds Clinical variables Chest X-rays ICU mortality |
topic |
COVID-19 Deep learning fusion Respiratory sounds Clinical variables Chest X-rays ICU mortality |
description |
Assessing the health status of critically ill patients with COVID-19 and predicting their outcome are highly challenging problems and one of the reasons for poor management of ICU resources worldwide. A better pathophysiological understanding of patients’ state evolution in the ICU can enhance effective medical interventions. Therefore, there is a need to monitor and analyze the pulmonary function of a ICU patient with COVID-19 and its impact on cardiovascular and other systems. To achieve this, chest X-rays (CXRs), respiratory sounds and all the routinely monitored parameters, scores and metrics in the COVID-19 ICU were recorded from 171 ICU patients with COVID-19 from June 2020 until December 2021. Features were extracted from respiratory sounds, deep learning analysis was conducted on CXRs, and logistic regression analysis was performed on routine ICU clinical variables. Deep learning pipelines were established to classify patients’ outcomes (survival or death) at two time points (ICU mortality or 90-day mortality) using three input configurations: (a) CXRs, (b) a fusion of CXRs and respiratory sounds features, or (c) a fusion of CXRs, respiratory sounds features, and principal features of the ICU clinical measurements. The performance of the latter approach was promising, achieving, for ICU mortality, an accuracy of 0.761 and an AUC of 0.759, and for 90-day mortality, an accuracy of 0.743 and an AUC of 0.752, while the performance of approaches (a) and (b) was worse. Therefore, using multi-source data and longitudinal COVID-19 ICU data offers a better prediction of the outcome in the ICU, thereby optimizing medical decisions and interventions. Furthermore, we show that adding the adventitious respiratory sounds features significantly increased AUC and accuracy for mortality prediction of ICU patients with COVID-19. |
publishDate |
2024 |
dc.date.none.fl_str_mv |
2024-01 2024-01-01T00:00:00Z 2026-01-31T00:00:00Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/article |
format |
article |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://hdl.handle.net/10773/39320 |
url |
http://hdl.handle.net/10773/39320 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
0957-4174 10.1016/j.eswa.2023.121089 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/embargoedAccess |
eu_rights_str_mv |
embargoedAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.publisher.none.fl_str_mv |
Elsevier |
publisher.none.fl_str_mv |
Elsevier |
dc.source.none.fl_str_mv |
reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação instacron:RCAAP |
instname_str |
Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
instacron_str |
RCAAP |
institution |
RCAAP |
reponame_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
collection |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository.name.fl_str_mv |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
repository.mail.fl_str_mv |
|
_version_ |
1799137745478615040 |