A Deep Learning Prognosis Model Help Alert for COVID-19 Patients at High-Risk of Death: A Multi-Center Study

Lingwei Meng,Di Dong,Liang Li,Meng Niu,Yan Bai,Meiyun Wang,Xiaoming Qiu,Yunfei Zha,Jie Tian

DOI: https://doi.org/10.1109/jbhi.2020.3034296

IF: 7.7

2020-12-01

IEEE Journal of Biomedical and Health Informatics

Abstract:Since its outbreak in December 2019, the persistent coronavirus disease (COVID-19) became a global health emergency. It is imperative to develop a prognostic tool to identify high-risk patients and assist in the formulation of treatment plans. We retrospectively collected 366 severe or critical COVID-19 patients from four centers, including 70 patients who died within 14 days (labeled as high-risk patients) since their initial CT scan and 296 who survived more than 14 days or were cured (labeled as low-risk patients). We developed a 3D densely connected convolutional neural network (termed De-COVID19-Net) to predict the probability of COVID-19 patients belonging to the high-risk or low-risk group, combining CT and clinical information. The area under the curve (AUC) and other evaluation techniques were used to assess our model. The De-COVID19-Net yielded an AUC of 0.952 (95% confidence interval, 0.928-0.977) on the training set and 0.943 (0.904-0.981) on the test set. The stratified analyses indicated that our model's performance is independent of age, sex, and with/without chronic diseases. The Kaplan-Meier analysis revealed that our model could significantly categorize patients into high-risk and low-risk groups (p < 0.001). In conclusion, De-COVID19-Net can non-invasively predict whether a patient will die shortly based on the patient's initial CT scan with an impressive performance, which indicated that it could be used as a potential prognosis tool to alert high-risk patients and intervene in advance.

computer science, interdisciplinary applications,mathematical & computational biology,medical informatics, information systems

What problem does this paper attempt to address?

The paper aims to address the following issues: Since its outbreak in December 2019, the novel coronavirus disease (COVID-19) has become a global health emergency. Given its rapid transmission rate and high mortality rate, there is an urgent need for a tool that can predict the prognosis of high-risk patients to identify them in a timely manner and formulate treatment plans. Therefore, the research team developed a deep learning-based prognostic model (referred to as De-COVID19-Net) to predict whether COVID-19 patients face a high risk of death within 14 days after the initial CT scan by combining CT images and clinical information. Specifically, the researchers collected data from 366 severe or critically ill COVID-19 patients from four centers, including 70 high-risk patients who died within 14 days of the first CT scan and 296 low-risk patients who survived more than 14 days or were cured. The developed three-dimensional densely connected convolutional neural network (3D-CNN) model achieved an area under the curve (AUC) of 0.952 on the training set and 0.943 on the test set, demonstrating excellent classification performance. Moreover, stratified analysis indicated that the model's performance was not affected by age, gender, or the presence of chronic diseases, and Kaplan-Meier analysis showed that the model could significantly distinguish between high-risk and low-risk patient groups (p<0.001). Overall, De-COVID19-Net can predict the short-term mortality risk of patients based on their initial CT scan in a non-invasive manner, making it valuable as a potential early warning tool.

A Deep Learning Prognosis Model Help Alert for COVID-19 Patients at High-Risk of Death: A Multi-Center Study

Identification of Risk Factors for Mortality Associated with COVID-19

Development and Validation of a Deep Learning Model for Prediction of Severe Outcomes in Suspected COVID-19 Infection

Dynamic Prognosis Model for Predicting Survival in Severe and Critically Ill COVID-19 Patients Using Machine Learning

A Deep Learning Radiomics Model to Identify Poor Outcome in COVID-19 Patients With Underlying Health Conditions: A Multicenter Study

Development and Validation of a Deep Learning-Based Model Using Computed Tomography Imaging for Predicting Disease Severity of Coronavirus Disease 2019

Early Prediction of Mortality Risk among Patients with Severe COVID-19, Using Machine Learning

Deep learning for predicting COVID-19 malignant progression

Using Deep Learning-based Features Extracted from CT scans to Predict Outcomes in COVID-19 Patients

An Early Warning Tool for Predicting Mortality Risk of COVID-19 Patients Using Machine Learning

A deep learning model for predicting COVID-19 ARDS in critically ill patients

Deep‐learning artificial intelligence analysis of clinical variables predicts mortality in COVID‐19 patients

Machine learning based early warning system enables accurate mortality risk prediction for COVID-19

Identifying Key Clinical Indicators Associated with the Risk of Death in Hospitalized COVID-19 Patients

Feature-based deep neural network approach for predicting mortality risk in patients with COVID-19

An elastic net regression model for predicting the risk of ICU admission and death for hospitalized patients with COVID-19

Machine Learning Based Clinical Decision Support System for Early COVID-19 Mortality Prediction

ANDC: an early warning score to predict mortality risk for patients with Coronavirus Disease 2019

Machine learning-based derivation and external validation of a tool to predict death and development of organ failure in hospitalized patients with COVID-19

Predictive Models of Mortality for Hospitalized Patients With COVID-19: Retrospective Cohort Study

A Prediction Model for Clinical Outcomes of COVID-19 Hospitalized Patients: Construction and Accuracy Assessment