A Deep Learning Approach to Identify Chest Computed Tomography Features for Prediction of SARS-CoV-2 Infection Outcomes

Amirhossein Sahebkar, Mitra Abbasifard, Samira Chaibakhsh, Paul C. Guest, Mohamad Amin Pourhoseingholi, Amir Vahedian-Azimi, Prashant Kesharwani, Tannaz Jamialahmadi

Research output: Chapter in Book/Conference paperChapterpeer-review

Abstract

There is still an urgent need to develop effective treatments to help minimize the cases of severe COVID-19. A number of tools have now been developed and applied to address these issues, such as the use of non-contrast chest computed tomography (CT) for evaluation and grading of the associated lung damage. Here we used a deep learning approach for predicting the outcome of 1078 patients admitted into the Baqiyatallah Hospital in Tehran, Iran, suffering from COVID-19 infections in the first wave of the pandemic. These were classified into two groups of non-severe and severe cases according to features on their CT scans with accuracies of approximately 0.90. We suggest that incorporation of molecular and/or clinical features, such as multiplex immunoassay or laboratory findings, will increase accuracy and sensitivity of the model for COVID-19 -related predictions.

Original languageEnglish
Title of host publicationMethods in Molecular Biology
EditorsPaul Guest
PublisherHumana Press
Pages395-404
Number of pages10
DOIs
Publication statusPublished - 2022

Publication series

NameMethods in Molecular Biology
Volume2511
ISSN (Print)1064-3745
ISSN (Electronic)1940-6029

Fingerprint

Dive into the research topics of 'A Deep Learning Approach to Identify Chest Computed Tomography Features for Prediction of SARS-CoV-2 Infection Outcomes'. Together they form a unique fingerprint.

Cite this