A Precision Medicine Approach to Patent Ductus Arteriosus in Preterm Infants
Integrating multimodal data and artificial intelligence to identify the right patient, treatment, and timing for optimal outcomes.
Transforming PDA Management Through Data-Driven Precision
Patent ductus arteriosus (PDA) is a frequent and vexing clinical challenge in the care of extremely preterm infants. Despite decades of research, there remains a profound lack of consensus on which infants benefit from treatment, the optimal timing of intervention, and the most effective therapeutic modality.
Affecting up to 70% of infants born before 28 weeks of gestation, the hemodynamic consequences of a large PDA shunt are associated with severe morbidities including bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), intraventricular hemorrhage (IVH), and increased mortality.
This proposal outlines a paradigm-shifting research program to develop, validate, and implement a precision medicine framework. We leverage multimodal data integration and artificial intelligence to create a holistic, dynamic understanding of each infant's risk profile and likely treatment response.
Specific Aims
Our overarching goal is to develop and validate a precision medicine framework to guide PDA management, improving outcomes and reducing treatment-related morbidity.
Develop and Validate a Multimodal AI Model
Create predictive models for spontaneous PDA closure and risk of hemodynamically significant PDA (hsPDA) by integrating EHR data, high-frequency physiological data, and advanced echocardiographic parameters.
A machine learning model can predict spontaneous PDA closure within 2 weeks with AUC > 0.85
AI models incorporating echocardiographic and biomarker data can identify infants who will develop hsPDA
Identify Novel Biomarkers and Genomic Signatures
Discover protein biomarkers and genetic variants that predict PDA persistence, hemodynamic significance, and treatment response through targeted and untargeted analyses.
Novel protein biomarkers related to vascular remodeling will differentiate persistent hsPDA from spontaneous closure
Genetic variants in prostaglandin and oxygen-sensing pathways predict treatment response
Develop and Evaluate a Clinical Decision Support Tool
Create a user-friendly CDS tool providing individualized, risk-stratified management recommendations and evaluate its impact through a multicenter randomized controlled trial.
The CDS tool will be feasible to implement and well-accepted by clinicians
CDS-guided management will significantly reduce death or moderate-to-severe BPD at 36 weeks PMA
Innovation
This proposal is highly innovative in its conceptual framework, integration of cutting-edge technologies, and goal of creating a learning healthcare system for PDA management.
Multimodal Data Integration
Integration of EHR data, high-frequency physiological waveforms, serial biomarkers, advanced echocardiography, and genomics using advanced machine learning.
Causal Inference Methods
Advanced causal inference techniques including propensity score matching and targeted learning to estimate individualized treatment effects.
Biomarker Discovery
Novel protein and genetic marker discovery to enhance predictive models and gain insights into PDA pathophysiology.
Clinical Decision Support
User-friendly CDS tool translating complex AI models into actionable bedside recommendations, validated through RCT.
Systems-Based Framework
Comprehensive framework considering the entire clinical trajectory, designed for continuous improvement as new data becomes available.
Physiological Integration
Deep integration of ductal closure physiology and molecular biology into model development for enhanced interpretability.
Multimodal Data Integration Framework
Clinical history, diagnoses, medications
ECG, SpO2, blood pressure waveforms
Ventilator modes, pressures, FiO2
AI-driven shunt quantification
Protein markers, genetic variants
Research Approach
A three-phase approach: prospective multicenter observational cohort study for model development and biomarker discovery, followed by a randomized controlled trial.
Available Cohorts for Model Development
| Cohort/Database | Description | Key Strengths |
|---|---|---|
| NICHD Neonatal Research Network (NRN) | Multi-center network across the US; Generic Database (1987-ongoing) | Large, well-characterized cohort; detailed clinical data; long-term follow-up |
| Vermont Oxford Network (VON) | Global Neonatal Database; long-term outcomes tracking for ELBW infants | Extensive international data; standardized data collection |
| National Neonatal Research Database (NNRD) - UK | Population-based cohort data from the UK | Links to outcome data; large sample size |
| ECHO Cohorts | Environmental Influences on Child Health Outcomes; three cohorts of very preterm infants | Longitudinal follow-up; demographic and neonatal characteristics |
Research Timeline
A comprehensive five-year research program from protocol development through trial completion and dissemination.
Foundation
- Finalize study protocol
- Obtain IRB approval
- Establish data infrastructure
- Begin prospective cohort enrollment
- Start retrospective data analysis
Development
- Continue prospective enrollment
- Complete model development using retrospective data
- Begin biomarker and genomic analysis
Refinement
- Complete prospective enrollment
- Refine AI models with prospective data
- Develop the CDS tool
- Conduct usability testing
Trial Launch
- Begin randomized controlled trial of CDS tool
- Continue long-term follow-up of observational cohort
Completion
- Complete the RCT
- Analyze the data
- Disseminate results
- Plan for implementation and scale-up
References
Ambalavanan N, Aucott SW, Salavitabar A, Levy VY, et al.. Patent Ductus Arteriosus in Preterm Infants. Pediatrics. 2025;155(5):e2025071425.
View SourceGupta S, Subhedar NV, Bell JL, Field D, et al. (Baby-OSCAR Collaborative Group). Trial of Selective Early Treatment of Patent Ductus Arteriosus with Ibuprofen. N Engl J Med. 2024;390:314-325.
View SourceBackes CH, Hill KD, Shelton EL, et al.. Patent Ductus Arteriosus: A Contemporary Perspective for the Pediatric and Adult Cardiac Care Provider. J Am Heart Assoc. 2022;11(17):e025784.
View SourceOvalı F. Molecular and Mechanical Mechanisms Regulating Ductus Arteriosus Closure in Preterm Infants. Front Pediatr. 2020;8:516.
View SourceSehgal A, Ruoss JL, Stanford AH, Lakshminrusimha S, McNamara PJ. Hemodynamic consequences of respiratory interventions in preterm infants. J Perinatol. 2022;42(9):1153-1160.
View SourceCucerea M, Marian R, Simon M, et al.. Serum Biomarkers in Patent Ductus Arteriosus in Preterm Infants: A Narrative Review. Biomedicines. 2025;13(3):670.
View SourceLalitha R, Bitar E, Hicks M, et al.. Multimodal Monitoring of Hemodynamics in Neonates With Extremely Low Gestational Age. JAMA Netw Open. 2025;8(4):e254101.
View SourcePark S, Moon J, Eun H, Hong JH, Lee K. Artificial Intelligence-Based Diagnostic Support System for Patent Ductus Arteriosus in Premature Infants. J Clin Med. 2024;13(7):2089.
View Source