{
  "abstract": "For over 30 years, there has been a near continuous effort to develop mathematical prediction models that will help clinicians and investigators distinguish between preterm infants who are likely to develop bronchopulmonary dysplasia (BPD) and those who are not.2 3 The predictive performance of many demographic and clinical characteristics, medication exposures, imaging results and diagnostic biomarkers has been examined for this purpose and variably included into published prediction models.2 4 Most models ultimately rely heavily on a few basic patient characteristics: the degree of prematurity at birth (often defined by gestational age, birth weight or both), patient sex, and the nature of the respiratory support administered at or near to the chronological age at risk prediction. With these inputs, potentially combined with a small number of additional variables, most prediction models that have been developed or tested in reasonably large patient populations have demonstrated moderate ability to discriminate between infants who do versus those who do not subsequently develop BPD.2 3 The incomplete prognostic performance of these models, combined with ongoing changes in perinatal and neonatal care practices, is an important driver that motivates continued efforts to generate more accurate and updated prediction tools.",
  "authors": [
    {
      "affiliations": [
        "Department of Pediatrics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA"
      ],
      "name": "Timothy D Nelin"
    },
    {
      "affiliations": [
        "Stanford University School of Medicine, Stanford, California, USA"
      ],
      "name": "Erik A Jensen"
    }
  ],
  "title": "Consideration of centre effects in BPD risk prediction",
  "uid": "2c1c52cf-45b4-544c-be05-09b274509362"
}