Abstract

Bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), and retinopathy of prematurity (ROP) are severe complications affecting Very Low Birth Weight (VLBW) infants. Our findings show that data gathered in the intensive care unit during the first 24 or 72 hours of care can be used to predict whether a VLBW infant is at risk of developing BPD. Using Gaussian process classification, we achieved classification results with areas under the receiver operator characteristic curve of 0.85 (standard error (SE) 0.05) for 24h and 0.87 (SE 0.06) for 72h BPD data. This compares favourably with results achieved using the clinical standard SNAP-II and SNAPPE-II scores. Sensitivity for BPD was 0.52 (SE 0.06). Sensitivity for NEC and ROP was close to zero, suggesting that NEC and ROP can not be reliably predicted with this approach from our data set.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call