Key result
Multicenter registry protocol evaluates machine learning model to predict hemodynamically significant PDA in preterm infants.
Why the study?
Systematic multicenter research on patent ductus arteriosus (PDA) in very preterm infants is scarce in China, and no dedicated online PDA registry exists nationally or internationally.
Can a machine learning-based predictive model using early clinical and echocardiographic variables accurately predict haemodynamically significant patent ductus arteriosus in very preterm infants?
Cohort (n=300)
Yes
Can a machine learning-based predictive model using early clinical and echocardiographic variables accurately predict haemodynamically significant patent ductus arteriosus in very preterm infants?
This study protocol establishes the first dedicated multicenter online registry and biobank for patent ductus arteriosus in very preterm infants in China, aiming to develop and validate a machine learning model for early risk stratification of haemodynamically significant PDA.
Answers are grounded in this paper and cite their sources.
Should not change hsPDA management; protocol leaves open ML model utility pending validation in very preterm infants.
Background Patent ductus arteriosus (PDA) is a common congenital cardiovascular condition in very preterm infants, with a reported incidence of up to 70%. Haemodynamically significant PDA (hsPDA) is associated with increased risks of intraventricular hemorrhage, bronchopulmonary dysplasia, necrotizing enterocolitis, and mortality. Systematic multicenter research on PDA in very preterm infants is scarce in China, and no dedicated online PDA registry exists either nationally or internationally. This study aims to establish a multicenter, prospective, observational cohort with a web-based registry to investigate the incidence, risk factors, management strategies, and outcomes of PDA in very preterm infants in Shenzhen, and to develop and internally validate a predictive model for hsPDA. Methods This multicenter, prospective, observational cohort study will be conducted across four tertiary hospitals in Shenzhen, China, from January 1, 2026 to December 31, 2028. Very preterm infants (gestational age < 32 weeks or birth weight < 1,500 g) with echocardiographically confirmed PDA will be enrolled. An online platform (Perinatal Cloud; https://perinatalcloud.com/) will serve as the central registry for standardized data collection. The primary objective is to develop and internally validate a machine learning-based predictive model for hsPDA. The target sample size of 300 infants with PDA was determined using the events-per-variable (EPV) criterion, assuming 10–15 candidate predictors and an hsPDA rate of approximately 50% among PDA cases, yielding ≥ 10 EPV. Internal validation will be performed via 10-fold cross-validation, with model performance assessed by area under the receiver operating characteristic curve (AUC), calibration plots, and Brier scores. Secondary objectives include determining PDA/hsPDA incidence, identifying risk factors, evaluating neurodevelopmental outcomes at 18–24 months corrected age, and exploring putative biomarkers. Biological specimens (blood, stool, urine, breast milk) will be collected and stored in a dedicated biobank. A nested case–control analysis will match PDA cases to non-PDA controls (1:2 ratio) on gestational age, birth weight, sex, and center. Discussion This study will establish the first dedicated online PDA registry in China, providing comprehensive, prospectively collected epidemiological data on PDA in very preterm infants. The development and internal validation of an hsPDA prediction model may enable early risk stratification and inform preventive strategies. The linked biobank will support future mechanistic and biomarker studies. Findings will be disseminated through peer-reviewed publications and conference presentations.
No takes yet. Share an insight, caveat, or question.
Qi et al. (2026) conducted a cohort in Patent ductus arteriosus (PDA) in very preterm infants (n=300). Patent ductus arteriosus (PDA) and clinical/echocardiographic risk factors vs. Non-hsPDA or matched non-PDA controls (1:2 nested case-control) was evaluated on Development and internal validation of a predictive model for haemodynamically significant patent ductus arteriosus (hsPDA, grade E3/E4). The Shenzhen online PDA registry protocol will enroll 300 very preterm infants with PDA across 4 centers (2026-2028) to develop and validate a machine learning model for predicting hsPDA.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: