Evaluating the ability of the NLHA2 and artificial neural network models to predict COVID-19 severity, and comparing them with the four existing scoring systems. (October 2022)

• Record Type: Journal Article
• Authors: Dong, Yue; Wang, Kai; Zou, Xu; Tan, Xiaoping; Zang, Yi; Li, Xinyu; Ren, Xiaoting; Xie, Desheng; Jie, Zhijun; Chen, Xiaohua; Zeng, Yingying; Shi, Jindong
• Is Part Of: Microbial pathogenesis. Volume 171(2022)
• Publication Date: 2022-10
• Subjects: PSI -- CURB-65 -- SMARTCOP -- MuLBSTA -- COVID-19 -- Machine learning; Pathogenic microorganisms -- Periodicals; Pathology, Molecular -- Periodicals; Communicable Diseases -- microbiology -- Periodicals; Communicable Diseases -- parasitology -- Periodicals; Micro-organismes pathogènes -- Périodiques; Pathologie moléculaire -- Périodiques; Electronic journals; 616.9041
• DOI: 10.1016/j.micpath.2022.105735
• Languages: English
• ISSNs: 0882-4010

Abstract

Abstract: To improve the identification and subsequent intervention of COVID-19 patients at risk for ICU admission, we constructed COVID-19 severity prediction models using logistic regression and artificial neural network (ANN) analysis and compared them with the four existing scoring systems (PSI, CURB-65, SMARTCOP, and MuLBSTA). In this prospective multi-center study, 296 patients with COVID-19 pneumonia were enrolled and split into the General-Ward-Care group (N = 238) and the ICU-Admission group (N = 58). The PSI model (AUC = 0.861) had the best results among the existing four scoring systems, followed by SMARTCOP (AUC = 0.770), motified-MuLBSTA (AUC = 0.761), and CURB-65 (AUC = 0.712). Data from 197 patients (training set) were analyzed for modeling. The beta coefficients from logistic regression were used to develop a severity prediction model and risk score calculator. The final model (NLHA2) included five covariates (consumes alcohol, neutrophil count, lymphocyte count, hemoglobin, and AKP). The NLHA2 model (training: AUC = 0.959; testing: AUC = 0.857) had similar results to the PSI model, but with fewer variable items. ANN analysis was used to build another complex model, which had higher accuracy (training: AUC = 1.000; testing: AUC = 0.907). Discrimination and calibration were further verified through bootstrapping (2000 replicates), Hosmer-Lemeshow goodness of fit testing, and Brier score calculation. In conclusion, the PSI model is the best existing system for predicting ICU admission among COVID-19 patients, while two newly-designed models (NLHA2 and ANN) performed better than PSI, and will provide a new approach for the development of prognostic evaluation system in a novel respiratory viral epidemic. Highlights: PSI can effectively predict the severity risk of adult COVID-19 patients. The NLHA2 model with only five variables had a similar predictive effect as PSI scoring. The prediction model constructed by artificial neural network algorithm performed better than models. Deep learning algorithms have great potential in clinical prediction.