Construction of an early diagnostic model for brain injury in premature infants based on combined amplitude-integrated electroencephalography and cranial ultrasound parameters.

OBJECTIVE

To develop and evaluate an early diagnostic model for brain injury in premature infants (BIPI) using combined amplitude-integrated electroencephalography (aEEG) and cranial ultrasound (CUS) parameters, aiming to enhance the accuracy of early BIPI detection.

METHODS

This single-center retrospective cohort study included 350 premature infants admitted to the Neonatal Intensive Care Unit (NICU) of the First Affiliated Hospital of Xi'an Medical University between August 2018 and October 2023. Key aEEG parameters (upper boundary voltage, lower boundary voltage, narrow bandwidth, and aEEG score) and CUS parameters (systolic blood flow velocity, diastolic blood flow velocity, and resistance index) were collected. Feature selection was performed using Lasso regression, and a combined risk prediction model was developed. Model performance was assessed using receiver operating characteristic (ROC) curves and the area under the curve (AUC).

RESULTS

Significant differences were observed in both aEEG and CUS parameters between the brain injury group (n = 145) and the non-injury group (n = 205) (all P < 0.05). Lasso regression identified seven key parameters for model construction. The combined model achieved an AUC of 0.89, with a sensitivity of 86.21% and specificity of 79.51%, significantly outperforming models based on aEEG or CUS parameters alone (P < 0.001).

CONCLUSION

The combined aEEG and CUS model significantly improves the early detection of BIPI and may facilitate timely interventions to reduce the risk of long-term neurodevelopmental impairments in premature infants.