Checking the power of the mathematical model with a computer program to determine the probability of developing congenital pneumonia in premature newborns with very low and extremely low birth weight

Abstract

Background. In the age of modern technologies, the relevance of preventive and personalized medicine is growing, therefore, the identification of preclinical markers of the disease is a priority.Objective. Checking the power of a mathematical model with a computer support program to determine the probability of developing congenital pneumonia in premature newborns with very low and extremely low birth weight.Design and methods. Based on the results of a comprehensive examination of 113 premature newborns with very low and extremely low birth weight, a mathematical model was developed (z = –4.59+2.54RFIII+2.40CVAF+1.83HCDH+ 1.81Asphyxia+1.66CCLP+0.86TP+0.74CRP+0.48CPIN)) with a computer support program to determine the likelihood of developing congenital pneumonia (http://pneu.bsmu.by ; https://www.bsmu.by in the Doctor menu/Pharmacist: “Diagnosis of congenital pneumonia”). The power test was carried out with an additional set of 58 premature newborns with very low and extremely low birth weight, not included in the resulting model.Results. The following data were obtained: sensitivity equal to 0.95, specificity — 0.90, prognostic significance of a positive test result — 0.95, prognostic significance of a negative test result — 0.90, likelihood ratio — 9.93.Conclusion. The results of external validation indicate the high power of the proposed model with a computer support program for determining the likelihood of developing congenital pneumonia in premature newborns with very low and extremely low birth weight.