Predicting the prognosis of patients with mild COVID-19 by chest CT based on machine learning

Autor: Ji, Bing, Kong, Li, Wang, Jian, Liu, Chen, Yuan, Kaiyi, Zhu, Liqiang, Liang, Hongqin
Zdroj: Chinese Journal of Academic Radiology; 20240101, Issue: Preprints p1-7, 7p
Abstrakt: Background: Chest computed tomography (CT) is of great significance for the preliminary diagnosis and disease evolution assessment of coronavirus disease 2019 (COVID-19). The limited ability of humans to process complex information may hinder the early recognition of patient deterioration. Objective: To develop an early-warning CT feature model for monitoring the follow-up outcome of COVID-19 patients by machine learning. Methods: CT images and clinical data of 140 patients with COVID-19 were retrospectively collected and divided into two groups (alleviation group and exacerbation group) by clinical symptoms and CT features. CT features data (distribution, morphology, etc.) were used to establish the prediction model by Fisher’s linear discriminant method and Unconditional logistic regression algorithm. The model was validated with 40 exception data. Results: The model filtered out three variables of CT featrues including distal air bronchogram, fiber bands, and reversed halo sign. Notably, the distal air bronchograms was less common in alleviation group, while the fiber bands and reversed halo sign were more common. The sensitivity, specificity and Youden index of unconditional logistic regression were 86.1%, 92.6% and 78.7%, respectively. For the analysis of Fisher’s linear discriminant, the sensitivity, specificity and Youden index were 83.3%, 94.1% and 77.4%, respectively. The generalization ability of both models was consistent with sensitivity of 95.89%, specificity of 100%, and Youden index of 83.33%. Conclusions: The CT imaging features-based machine learning model had a high sensitivity for predicting outcomes in COVID-19 patients.
Databáze: Supplemental Index
Externí odkaz: