Implementing structured follow-up of neonatal and paediatric patients: an evaluation of three university hospital case studies using the functional resonance analysis method.

BACKGROUND: In complex critical neonatal and paediatric clinical practice, little is known about long-term patient outcomes and what follow-up care is most valuable for patients. Emma Children's Hospital, Amsterdam UMC (Netherlands), implemented a follow-up programme called Follow Me for neonatal and paediatric patient groups, to gain more insight into long-term outcomes and to use such outcomes to implement a learning cycle for clinical practice, improve follow-up care and facilitate research. Three departments initiated re-engineering and change processes. Each introduced multidisciplinary approaches to long-term follow-up, including regular standardised check-ups for defined age groups, based on medical indicators, developmental progress, and psychosocial outcomes in patients and their families. This research evaluates the implementation of the three follow-up programmes, comparing predefined procedures (work-as-imagined) with how the programmes were implemented in practice (work-as-done). METHODS: This study was conducted in 2019-2020 in the outpatient settings of the neonatal intensive care, paediatric intensive care and paediatric surgery departments of Emma Children's Hospital. It focused on the organisational structure of the follow-up care. The functional resonance analysis method (FRAM) was applied, using documentary analysis, semi-structured interviews, observations and feedback sessions. RESULTS: One work-as-imagined model and four work-as-done models were described. The results showed vast data collection on medical, developmental and psychosocial indicators in all work-as-done models; however, process indicators for programme effectiveness and performance were missing. In practice there was a diverse allocation of roles and responsibilities and their interrelations to create a multidisciplinary team; there was no one-size-fits-all across the different departments. Although control and feedback loops for long-term outcomes were specified with respect to the follow-up groups within the programmes, they were found to overlap and misalign with other internal and external long-term outcome monitoring practices. CONCLUSION: Implementing structured long-term follow-up may provide insights for improving daily practice and follow-up care, with the precondition of standardised measurements. Lessons learned from practice are (1) to address fragmentation in data collection and storage, (2) to incorporate the diverse ways to create a multidisciplinary team in practice, and (3) to include timely actionable indicators on programme effectiveness and performance, alongside medical, developmental and psychosocial indicators.

Machine Learning Prediction Models for Neurodevelopmental Outcome After Preterm Birth: A Scoping Review and New Machine Learning Evaluation Framework.

BACKGROUND AND OBJECTIVES: Outcome prediction of preterm birth is important for neonatal care, yet prediction performance using conventional statistical models remains insufficient. Machine learning has a high potential for complex outcome prediction. In this scoping review, we provide an overview of the current applications of machine learning models in the prediction of neurodevelopmental outcomes in preterm infants, assess the quality of the developed models, and provide guidance for future application of machine learning models to predict neurodevelopmental outcomes of preterm infants. METHODS: A systematic search was performed using PubMed. Studies were included if they reported on neurodevelopmental outcome prediction in preterm infants using predictors from the neonatal period and applying machine learning techniques. Data extraction and quality assessment were independently performed by 2 reviewers. RESULTS: Fourteen studies were included, focusing mainly on very or extreme preterm infants, predicting neurodevelopmental outcome before age 3 years, and mostly assessing outcomes using the Bayley Scales of Infant Development. Predictors were most often based on MRI. The most prevalent machine learning techniques included linear regression and neural networks. None of the studies met all newly developed quality assessment criteria. Studies least prone to inflated performance showed promising results, with areas under the curve up to 0.86 for classification and R2 values up to 91% in continuous prediction. A limitation was that only 1 data source was used for the literature search. CONCLUSIONS: Studies least prone to inflated prediction results are the most promising. The provided evaluation framework may contribute to improved quality of future machine learning models.