Using machine learning to impact on long-term clinical care: principles, challenges, and practicalities

Area of Science:

• Pediatric Medicine
• Artificial Intelligence
• Machine Learning

Background:

• Machine learning (ML) applications are increasingly prevalent in daily life but underutilized in healthcare.
• Pediatricians manage complex, heterogeneous long-term conditions, presenting opportunities for ML-driven insights.
• Understanding ML foundations and applications is currently limited within the medical community.

Purpose of the Study:

• To review the fundamental principles and algorithms of machine learning.
• To highlight the importance of proper data preparation and external validation in ML applications.
• To explore the current and potential utility of ML in pediatrics using clinical examples.

Main Methods:

• Review of machine learning foundations and algorithms.
• Discussion of ML applications using case studies in preterm infant nutrition and pediatric inflammatory bowel disease.
• Examination of challenges and ethical considerations for AI in pediatrics.

Main Results:

• Machine learning holds significant promise for improving pediatric care through predictive models and patient subgroup identification.
• Clinical examples demonstrate the potential utility of ML in areas like preterm infant nutrition and pediatric inflammatory bowel disease.
• Effective ML implementation requires careful data preparation and external validation.

Conclusions:

• Machine learning has the potential to significantly advance pediatric medicine, particularly for complex conditions.
• Further research and careful implementation are needed to overcome challenges and ethical considerations.
• This review aims to bridge the understanding gap regarding ML in pediatrics and neonatal medicine.