Mathematical Analysis of the Accuracy of the Evaluation Criterion X in Patients with COVID-19 Infection View PDF

^*Vladimir T. Ivashkin
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Evgeniya Y. Medvedeva
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Vladimir M. Nechaev
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Irina R. Popova
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Manana R. Shirtladze
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Yuliya O. Shulpekova
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation
Aleksandr A. Simushev
Medicine, Department Of Mathematical And Computer Modelling, National Research University, Moscow, Russian Federation
Sakhavet M. Zarbaliev
Medicine, Department Of Mathematical And Computer Modelling, National Research University, Moscow, Russian Federation

^*Corresponding Author:

Vladimir T. Ivashkin
Medicine, Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russian Federation

Published on: 2025-11-25

Abstract

Prognostic criterion X is calculated using an equation that includes 7 indexed parameters (age and body temperature of the patient, oxygen saturation (SatO?), number of blood neutrophils, values of creatinine, C-reactive protein and blood creatinine). Criterion X is measured in points, the number of which is directly proportional to the severity of the disease. The accuracy of criterion X was estimated by the method of ordinary least squares regression (OLS regression) and correlation analysis according to the criteria of r Pearson and rs Spearman in 305 patients with COVID-19 infection and pneumonia. The data obtained were compared by assessing the severity of the disease according to the general clinical parameters Xcl (state of consciousness, body T °C, pulse and respiratory rate per minute) and the NEWS-2 criterion. The results of the OLS regression demonstrate that there is a strong linear relationship between Xcl and X (Xev) (linear regression coefficient β1 = 0.8057, the standard deviation (std) of the coefficient β1 is s1 = 0.022). The coefficient of determination of this regression is R2 = 0.821. This means that the assessment of the severity of the disease and its prognosis for Xcl and X coincide with each other in 82% of cases. A comparison of the criteria X (Xev) and XNEWS-2 indicates the absence of a linear relationship between them (the value of the linear regression coefficient β1= 0.0553. The std of the coefficient β1 is s1 = 0.002). The comparison of the Xcl and XNEWS-2 criteria was not carried out due to the lack of data in the public domain. The evaluation criterion X meets the necessary statistical requirements and can be used to objectively assess the severity of inflammatory diseases, as well as to develop treatment regimens.

Keywords

COVID-19 infection, SARS-CoV-2-pneumonia, Criterion X, NEWS-2 scale, Early warning scores, Indexed laboratory parameters

Introduction

The most important condition for the effective treatment of any disease is an accurate assessment of its severity, which makes it possible not only to predict possible complications in a timely manner and achieve complete recovery (clinical remission), but also minimize side effects. All modern therapeutic regimens have been created with this requirement in mind. The existing evaluation criteria (APACHE III, SAPS II, DTEWS, MEWS, NEWS-2) are quite sensitive, but highly specialized and difficult to reproduce [1-5]. In addition, they contain parameters that are often poorly correlated with each other, namely: respirations per minute, systolic blood pressure, and level of consciousness. This circumstance reduces their evaluative value.

The gated recurrent unit prognostic model is also used. It is based on a recurrent neural network architecture and takes into account gender, height, weight, comorbidity according to the Charlson scale, body temperature, blood pressure, SatO_?, the presence of chronic kidney disease and diabetes mellitus, and blood test data (a total of 48 evaluation parameters) [6]. This model requires a powerful laboratory base and is designed primarily to predict mortality, rather than the severity of the patient’s condition at the initial stage of treatment. We have developed a new evaluation criterion X, which takes into account not only clinical parameters but also indexed laboratory parameters that systematically characterize the condition of the body [7]. These parameters are interrelated. This approach enhances the value of criterion X, but it requires mathematical validation. Criterion X is intended for selecting effective treatment and assessing prognosis. Mathematical analysis of the accuracy of the evaluation criterion X (X_ev = X_evaluation).