Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19
We validate a deep learning model predicting comorbidities from frontal chest radiographs (CXRs) in patients with coronavirus disease 2019 (COVID-19) and compare the model’s performance with hierarchical condition category (HCC) and mortality outcomes in COVID-19. The model was trained and tested on...
Main Authors: | , , , , , , , , , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Public Library of Science (PLoS)
2022-08-01
|
Series: | PLOS Digital Health |
Online Access: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9931278/?tool=EBI |
_version_ | 1797726945237008384 |
---|---|
author | Ayis Pyrros Jorge Rodriguez Fernandez Stephen M. Borstelmann Adam Flanders Daniel Wenzke Eric Hart Jeanne M. Horowitz Paul Nikolaidis Melinda Willis Andrew Chen Patrick Cole Nasir Siddiqui Momin Muzaffar Nadir Muzaffar Jennifer McVean Martha Menchaca Aggelos K. Katsaggelos Sanmi Koyejo William Galanter |
author_facet | Ayis Pyrros Jorge Rodriguez Fernandez Stephen M. Borstelmann Adam Flanders Daniel Wenzke Eric Hart Jeanne M. Horowitz Paul Nikolaidis Melinda Willis Andrew Chen Patrick Cole Nasir Siddiqui Momin Muzaffar Nadir Muzaffar Jennifer McVean Martha Menchaca Aggelos K. Katsaggelos Sanmi Koyejo William Galanter |
author_sort | Ayis Pyrros |
collection | DOAJ |
description | We validate a deep learning model predicting comorbidities from frontal chest radiographs (CXRs) in patients with coronavirus disease 2019 (COVID-19) and compare the model’s performance with hierarchical condition category (HCC) and mortality outcomes in COVID-19. The model was trained and tested on 14,121 ambulatory frontal CXRs from 2010 to 2019 at a single institution, modeling select comorbidities using the value-based Medicare Advantage HCC Risk Adjustment Model. Sex, age, HCC codes, and risk adjustment factor (RAF) score were used. The model was validated on frontal CXRs from 413 ambulatory patients with COVID-19 (internal cohort) and on initial frontal CXRs from 487 COVID-19 hospitalized patients (external cohort). The discriminatory ability of the model was assessed using receiver operating characteristic (ROC) curves compared to the HCC data from electronic health records, and predicted age and RAF score were compared using correlation coefficient and absolute mean error. The model predictions were used as covariables in logistic regression models to evaluate the prediction of mortality in the external cohort. Predicted comorbidities from frontal CXRs, including diabetes with chronic complications, obesity, congestive heart failure, arrhythmias, vascular disease, and chronic obstructive pulmonary disease, had a total area under ROC curve (AUC) of 0.85 (95% CI: 0.85–0.86). The ROC AUC of predicted mortality for the model was 0.84 (95% CI,0.79–0.88) for the combined cohorts. This model using only frontal CXRs predicted select comorbidities and RAF score in both internal ambulatory and external hospitalized COVID-19 cohorts and was discriminatory of mortality, supporting its potential use in clinical decision making. Author summary Artificial Intelligence algorithms in Radiology can be used not only on standard imaging data like chest radiographs to predict diagnoses but can also incorporate other data. We wanted to find out if we could combine administrative and demographic data with chest radiographs to predict common comorbidities and mortality. Our deep learning algorithm was able to predict diabetes with chronic complications, obesity, congestive heart failure, arrythmias, vascular disease, and chronic obstructive pulmonary disease. The deep learning algorithm was also able to predict an administrative metric (RAF score) used in value-based Medicare Advantage plans. We used these predictions as biomarkers to predict mortality with a second statistical model using logistic regression in COVID-19 patients both in and out of the hospital. The degree of discrimination both the deep learning algorithm and statistical model provide would be considered ‘good’ by most, and certainly much better than chance alone. It was measured at 0.85 (95% CI: 0.85–0.86) by the area under the ROC curve method for the artificial intelligence algorithm, and 0.84 (95% CI:0.79–0.88) by the same method for the statistical mortality prediction model. |
first_indexed | 2024-03-12T10:52:44Z |
format | Article |
id | doaj.art-a4e4f2efe4f2446d9fa2eee8e7a81ab4 |
institution | Directory Open Access Journal |
issn | 2767-3170 |
language | English |
last_indexed | 2024-03-12T10:52:44Z |
publishDate | 2022-08-01 |
publisher | Public Library of Science (PLoS) |
record_format | Article |
series | PLOS Digital Health |
spelling | doaj.art-a4e4f2efe4f2446d9fa2eee8e7a81ab42023-09-02T06:49:14ZengPublic Library of Science (PLoS)PLOS Digital Health2767-31702022-08-0118Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19Ayis PyrrosJorge Rodriguez FernandezStephen M. BorstelmannAdam FlandersDaniel WenzkeEric HartJeanne M. HorowitzPaul NikolaidisMelinda WillisAndrew ChenPatrick ColeNasir SiddiquiMomin MuzaffarNadir MuzaffarJennifer McVeanMartha MenchacaAggelos K. KatsaggelosSanmi KoyejoWilliam GalanterWe validate a deep learning model predicting comorbidities from frontal chest radiographs (CXRs) in patients with coronavirus disease 2019 (COVID-19) and compare the model’s performance with hierarchical condition category (HCC) and mortality outcomes in COVID-19. The model was trained and tested on 14,121 ambulatory frontal CXRs from 2010 to 2019 at a single institution, modeling select comorbidities using the value-based Medicare Advantage HCC Risk Adjustment Model. Sex, age, HCC codes, and risk adjustment factor (RAF) score were used. The model was validated on frontal CXRs from 413 ambulatory patients with COVID-19 (internal cohort) and on initial frontal CXRs from 487 COVID-19 hospitalized patients (external cohort). The discriminatory ability of the model was assessed using receiver operating characteristic (ROC) curves compared to the HCC data from electronic health records, and predicted age and RAF score were compared using correlation coefficient and absolute mean error. The model predictions were used as covariables in logistic regression models to evaluate the prediction of mortality in the external cohort. Predicted comorbidities from frontal CXRs, including diabetes with chronic complications, obesity, congestive heart failure, arrhythmias, vascular disease, and chronic obstructive pulmonary disease, had a total area under ROC curve (AUC) of 0.85 (95% CI: 0.85–0.86). The ROC AUC of predicted mortality for the model was 0.84 (95% CI,0.79–0.88) for the combined cohorts. This model using only frontal CXRs predicted select comorbidities and RAF score in both internal ambulatory and external hospitalized COVID-19 cohorts and was discriminatory of mortality, supporting its potential use in clinical decision making. Author summary Artificial Intelligence algorithms in Radiology can be used not only on standard imaging data like chest radiographs to predict diagnoses but can also incorporate other data. We wanted to find out if we could combine administrative and demographic data with chest radiographs to predict common comorbidities and mortality. Our deep learning algorithm was able to predict diabetes with chronic complications, obesity, congestive heart failure, arrythmias, vascular disease, and chronic obstructive pulmonary disease. The deep learning algorithm was also able to predict an administrative metric (RAF score) used in value-based Medicare Advantage plans. We used these predictions as biomarkers to predict mortality with a second statistical model using logistic regression in COVID-19 patients both in and out of the hospital. The degree of discrimination both the deep learning algorithm and statistical model provide would be considered ‘good’ by most, and certainly much better than chance alone. It was measured at 0.85 (95% CI: 0.85–0.86) by the area under the ROC curve method for the artificial intelligence algorithm, and 0.84 (95% CI:0.79–0.88) by the same method for the statistical mortality prediction model.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9931278/?tool=EBI |
spellingShingle | Ayis Pyrros Jorge Rodriguez Fernandez Stephen M. Borstelmann Adam Flanders Daniel Wenzke Eric Hart Jeanne M. Horowitz Paul Nikolaidis Melinda Willis Andrew Chen Patrick Cole Nasir Siddiqui Momin Muzaffar Nadir Muzaffar Jennifer McVean Martha Menchaca Aggelos K. Katsaggelos Sanmi Koyejo William Galanter Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 PLOS Digital Health |
title | Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 |
title_full | Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 |
title_fullStr | Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 |
title_full_unstemmed | Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 |
title_short | Validation of a deep learning, value-based care model to predict mortality and comorbidities from chest radiographs in COVID-19 |
title_sort | validation of a deep learning value based care model to predict mortality and comorbidities from chest radiographs in covid 19 |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9931278/?tool=EBI |
work_keys_str_mv | AT ayispyrros validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT jorgerodriguezfernandez validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT stephenmborstelmann validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT adamflanders validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT danielwenzke validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT erichart validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT jeannemhorowitz validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT paulnikolaidis validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT melindawillis validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT andrewchen validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT patrickcole validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT nasirsiddiqui validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT mominmuzaffar validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT nadirmuzaffar validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT jennifermcvean validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT marthamenchaca validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT aggeloskkatsaggelos validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT sanmikoyejo validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 AT williamgalanter validationofadeeplearningvaluebasedcaremodeltopredictmortalityandcomorbiditiesfromchestradiographsincovid19 |