RespiratoryPrediction Model

BPD-Free Survival Prediction

A machine-learning tool estimating the probability of bronchopulmonary dysplasia–free survival in very preterm infants, from perinatal factors alone and after the first 1, 7, and 14 days of respiratory support. The model — random forests combined by an ensemble — was developed at the Loma Linda University Children's Hospital and Riverside University Health System NICUs (Chou et al., 2022) and runs here as a faithful in-browser port verified against the original.

Trained on infants GA 22w4d–30w3d (rounded to the closest full week). Predictions update at four stages; the respiratory stages require continuous data from DOL1 (DOL7 needs DOL1–7, DOL14 needs DOL1–14). Educational tool — verify against the clinical picture.

Perinatal features

GA — week (22–30)

GA — day (0–6)

Birth weight (g)

Sex

Race / ethnicity

In-utero smoking

Respiratory support mode by day of life

DOL 1

DOL 2

DOL 3

DOL 4

DOL 5

DOL 6

DOL 7

DOL 8

DOL 9

DOL 10

DOL 11

DOL 12

DOL 13

DOL 14

Chou F-S, Pham A, Leigh R, et al. A machine-learning prediction model for bronchopulmonary dysplasia. BMC Pediatr. 2022;22:546. doi:10.1186/s12887-022-03602-w

BPD definition: Jensen EA, et al. Am J Respir Crit Care Med. 2019;200:751–759. Models: random forests (ranger/caret) combined by a generalized additive ensemble. Loma Linda University Children's Hospital & Riverside University Health System NICUs.

Frequently asked questions

What does this model predict?

The probability of survival without BPD, using a faithful in-browser port of the LLU/RUHS random-forest ensemble.

Is it a substitute for clinical judgment?

No — it is an educational risk estimate; management decisions require full clinical assessment.

Be the first to react