Aim: Using the expanded Apgar score, it was determined whether interventions performed in the ward could reliably predict the occurrence of retinopathy of prematurity (ROP). Materials and methods: A single-centre study based on a retrospective analysis of medical records of 287 children with a gestational age of ≤32 weeks. Artificial intelligence algorithms and chisquare automatic interaction detection (CHAID) were used to analyse the Apgar score, and the course of resuscitation was documented according to the expanded Apgar score. The CHAID classification tree creates a transparent knowledge system based on decision rules. Results: The mean gestational age of newborns with severe ROP (sROP) was 25.6 weeks, compared to 28.7 weeks (p < 0.001). No significant sex-related differences in sROP occurrence were observed in the study group. Analysis of the expanded Apgar score parameters showed that the fraction of inspired oxygen – oxygen concentration in the gas mixture (FiO₂) – received by the newborn at 10 minutes was a statistically significant parameter. The cut-off point was a FiO₂ of 0.8. Of the newborns with sROP, 29.6% had FiO₂ above this concentration, compared to only 6.5% below it (p < 0.001). The analysis also showed that 15.1% of babies requiring intubation within 3 minutes of life developed sROP, indicating a significant effect of this expanded Apgar score parameter. Only 3.7% of non-intubated newborns had sROP (p = 0.001). Chest compression and epinephrine administration showed no significant effect on sROP occurrence. Conclusions: Considering intubation and oxygen concentration supplied during resuscitation significantly improves the prognosis of retinopathy of prematurity compared to using the Apgar score alone.

algorithm, Apgar score, expanded Apgar score, chi-square automatic interaction detection, retinopathy of prematurity