Development and validation of a nomogram to predict intracranial haemorrhage in neonates

Abstract

Background

The aim of this study was to establish and validate a Susceptibility-weighted imaging (SWI)-based predictive model for neonatal intracranial haemorrhage (ICH).

Methods

A total of 1190 neonates suspected of ICH after cranial ultrasound screening in a tertiary hospital were retrospectively enrolled. The neonates were randomly divided into a training cohort and a internal validation cohort by a ratio of 7:3. Univariate analysis was used to analyze the correlation between risk factors and ICH, and the prediction model of neonatal ICH was established by multivariate logistic regression based on minimum Akaike information criterion (AIC). The nomogram was externally validated in another tertiary hospital of 91 neonates. The performance of the nomogram was evaluated in terms of discrimination by the area under the curve (AUC), calibration by the calibration curve and clinical net benefit by the decision curve analysis (DCA).

Results

Univariate analysis and min AIC-based multivariate logistic regression screened the following variables to establish a predictive model for neonatal ICH: Platelet count (PLT), gestational diabetes, mode of delivery, amniotic fluid contamination, 1-min Apgar score. The AUC was 0.715, 0.711, and 0.700 for the training cohort, internal validation cohort, and external validation cohort, respectively. The calibration curve showed a good correlation between the nomogram prediction and actual observation for ICH. DCA showed the nomogram was clinically useful.

Conclusion

We developed and validated an easy-to-use nomogram to predict ICH for neonates. This model could support individualized risk assessment and healthcare.