Predictive Model for the Type of Active Retinopathy of Prematurity Based on Data from Angiological Studies

Abstract

The purpose — to develop a prognostic model of the type of active retinopathy of prematurity (ROP) based on informative prognostic parameters of optical coherence tomography of the retina in angio-mode (OCT-A) and color duplex scanning (CDS) for a reliable choice of patient management tactics.Patients and methods. 30 premature infants (30 eyes) with favorable and unfavorable types of 1–3 stages of active ROP were performed with optical coherence tomography of the retina in angio-mode. Children were born at a gestational age of 25–34 weeks. The birth weight of children was 680–2030 g. 55 premature infants (55 eyes) with active ROP were performed with CDS in modes of color Doppler mapping and pulsed Doppler sonography. Children were born at a gestational age of 25–32 weeks. The birth weight of children was 680–1760 g. ROC analysis was used to estimate the prognostic information content of the quantitative indicators of OCT-A and CDS in relation to the type of active ROP. The discriminant analysis was used to estimate the contribution of the studied quantitative indicators to the intergroup differences in the type of active ROP. A regression model was built to obtain a prediction of a favorable or unfavorable type of disease. The assessment of the acceptability of the model as a whole was carried out according to the index of determination and the level of significance of the resulting model (p < 0.05).Results. The most informative prognostic indicators of the type of active ROP were determined, according to OCT in angio-mode: the area of the foveal avascular zone — and, according to CDS: maximum systolic and end diastolic blood flow velocity in the ophthalmic artery, maximum systolic blood flow velocity in the posterior short ciliary arteries, minimum velocity blood flow in the central retinal vein. On the basis of informative indicators, a reliable prognostic model of the type of active ROP was built, which has two independent variables: the area of the foveal avascular zone and the end diastolic blood flow velocity in the ophthalmic artery. The dependent variable can take values close to 0, which determine the favorable type of the disease, or close to 1, which determines the unfavorable type of the disease.Conclusion. The introduction of the obtained model into clinical practice will allow to objectively determine the unfavorable course of the pathological process at the earliest stages, as well as to carry out early treatment of active ROP.

Keywords

• predictive model
• optical coherence tomography of the retina in angio-mode
• color duplex scanning