Development and Validation of a Machine Learning Approach for Automated Severity Assessment of COVID-19 Based on Clinical and Imaging Data: Retrospective Study

Abstract

Background COVID-19 has overwhelmed health systems worldwide. It is important to identify severe cases as early as possible, such that resources can be mobilized and treatment can be escalated. Objective This study aims to develop a machine learning approach for automated severity assessment of COVID-19 based on clinical and imaging data. Methods Clinical data—including demographics, signs, symptoms, comorbidities, and blood test results—and chest computed tomography scans of 346 patients from 2 hospitals in the Hubei Province, China, were used to develop machine learning models for automated severity assessment in diagnosed COVID-19 cases. We compared the predictive power of the clinical and imaging data from multiple machine learning models and further explored the use of four oversampling methods to address the imbalanced classification issue. Features with the highest predictive power were identified using the Shapley Additive Explanations framework. Results Imaging features had the strongest impact on the model output, while a combination of clinical and imaging features yielded the best performance overall. The identified predictive features were consistent with those reported previously. Although oversampling yielded mixed results, it achieved the best model performance in our study. Logistic regression models differentiating between mild and severe cases achieved the best performance for clinical features (area under the curve [AUC] 0.848; sensitivity 0.455; specificity 0.906), imaging features (AUC 0.926; sensitivity 0.818; specificity 0.901), and a combination of clinical and imaging features (AUC 0.950; sensitivity 0.764; specificity 0.919). The synthetic minority oversampling method further improved the performance of the model using combined features (AUC 0.960; sensitivity 0.845; specificity 0.929). Conclusions Clinical and imaging features can be used for automated severity assessment of COVID-19 and can potentially help triage patients with COVID-19 and prioritize care delivery to those at a higher risk of severe disease.

Insights Into Co-Morbidity and Other Risk Factors Related to COVID-19 Within Ontario, Canada

“… Fisman et al (2020) used logistic regression models to predict mortality risk of COVID-19 patients; their logistic regression model quantifies the weight of each input variable to the final prediction, making it straightforward to determine how the model is calculating the overall COVID-19 mortality risk. A similar study by Quiroz et al (2021) developed a logistic regression model using clinical and imaging data from two hospitals in Hubei, China, for automated severity assessment of COVID-19 for individual patients, obtaining an AUC of 0.950 using a combination of clinical and imaging features. They interpreted the importance of features using SHAP values and found patients in severe conditions had co-morbidities which included cardiovascular disease, diabetes, hypertension and cancer which is similar to findings obtained from previous studies (see e.g., Petrilli et al, 2020 ; Richardson et al, 2020 ; Shi et al, 2020 ; Siordia, 2020 ).…”

Predicting Antituberculosis Drug–Induced Liver Injury Using an Interpretable Machine Learning Method: Model Development and Validation Study

“…Various models are used to identify drugs associated with the risk of DILI at the preclinical stage [28]. Machine learning models have demonstrated strong predictive power and retained a simple form for communication with researchers [29][30][31][32][33][34][35][36][37][38][39]. XGBoost is a boosting ensemble machine learning algorithm that integrates a few classification and regression trees models to form a strong classifier [40,41].…”

Bayesian Model Infers Drug Repurposing Candidates for Treatment of COVID-19

“…Multiple informatic approaches have been applied to the integration of population level, molecular and imaging data associated with SARS-CoV-2 infection. These include conventional interactome networks, associations with abundance trajectories with viral proteins, neural networks, deep and self-supervised learning, as well as machine learning of digital data [8,10,[16][17][18][19][20]. Each type of informatic approach has its advantages and disadvantages based on a priori knowledge, power and structure of the dataset, as well as the ability to develop de novo knowledge.…”