Establishment and evaluation of nomogram for predicting intraventricular hemorrhage in neonatal acute respiratory distress syndrome

Abstract

Background: Intraventricular hemorrhage (IVH) is the most common type of brain injury in newborns, especially in newborns with Neonatal acute respiratory distress syndrome (ARDS). IVH can cause brain parenchyma damage and long-term neurological sequelae in children. Early identification and prevention of sequelae are essential. This study aims to establish a predictive nomogram for the early prediction of IVH in newborns with ARDS.

Methods: From 2019 to 2021, we collected data from 222 infants diagnosed with ARDS in the Department of Neonatology, First Affiliated Hospital of Xinjiang Medical University. Infants have been randomly assigned to the training set (n = 161) or the validation set (n = 61) at a ratio of 7:3. Variables were screened using the Least Absolute Contract and Selection Operator (LASSO) regression to create a risk model for IVH in infants with ARDS. The variables chosen in the LASSO regression model were used to establish the prediction model using multivariate logistic regression analysis.

Results: We recognized 4 variables as independent risk factors for IVH in newborns with ARDS via LASSO analysis, consisting of premature rupture of membranes (PROM), pulmonary surfactant (PS) dosage, PH¹ and Arterial partial pressure of oxygen (PaO₂¹). The C-Index for this dataset is 0.868 (95% CI: 0.837-0.940) and the C index in bootstrap verification is 0.852 respectively. The analysis of the decision curve shows that the model can significantly improve clinical efficiency in predicting IVH. We also provide a website based on the model and open it to users for free, so that the model can be better applied to clinical practice.

Conclusion: In conclusion, the nomogram based on 4 factors shows good identification, calibration and clinical practicability. Our nomographs can help clinicians make clinical decisions, screen high-risk ARDS newborns, and facilitate early identification and management of IVH patients.

Keywords: Intraventricular hemorrhage (IVH); Neonatal acute respiratory distress syndrome (ARDS); Nomogram; Prediction model; Risk factors.

Fig. 1

Flowchart of patient selection. ARDS = acute respiratory distress syndrome, TTN = transient tachypnoea of the neonate, IVH = intraventricular hemorrhage

Fig. 2

Feature selection based on LASSO binary logistic regression analysis. A The penalty coefficients in the LASSO model are adjusted using cross-validation and minimum criteria. B The prediction factors are determined by the cable regression method

Fig. 3

Nomogram of IVH prediction in newborns with ARDS. PROM = premature rupture of membrane, PS = pulmonary surfactant, PH¹ and O₂¹ respectively represent the PH value and oxygen partial pressure value of the first blood gas analysis

Fig. 4

ROC curves. A Training cohort, B Validation cohort, ROC = receiver operating characteristic, AUC = area under the ROC curve

Fig. 5

Case probability density plots (A) IVH prediction in training set, (B) NO-IVH prediction in training set, (C) IVH prediction in Validate set, (D) NO-IVH prediction in Validate set

Fig. 6

Calibration curve for predicting the probability of IVH in patients with ARDS. A Training cohort, B Validation cohort

Fig. 7

Nomogram decision curve for Intraventricular hemorrhage in the training cohort

Fig. 8

Web Application Screenshot