Risk assessment of progression to severe conditions for patients with COVID-19 pneumonia: a single-center retrospective study

Lijiao Zeng; Jialu Li; Mingfeng Liao; Rui Hua; Pilai Huang; Mingxia Zhang; Youlong Zhang; Qinlang Shi; Zhaohua Xia; Xinzhong Ning; Dandan Liu; Jiu Mo; Ziyuan Zhou; Zigang Li; Yu Fu; Yuhui Liao; Jing Yuan; Lifei Wang; Qing He; Lei Liu; Kun Qiao

We are still left with a few possible extensions. First, the laboratory testing data had not been integrated into the CT image-based model because blood sample was not collected along with imaging in the center. A more complicated statistical model is required to account for the error caused by time difference inherent in input data. Second, the occurrence of severity may depend on other factors such as treatment, viral load or genetic factors. Our model can include such additional covariates. However, the availability of well-processed data on these factors is scarce, and we have shown that the current model can have reasonable prediction performance even without considering them. Third, because of limited sample size of patients progressed to severity, we only evaluated model prediction performance by cross-validation. Further validation needs to be conducted on external datasets. Fourth, we restricted our analysis of the severe group up to the severity onset. Factors related to the recovery from severe conditions have not been evaluated.