Applied artificial intelligence for global child health: Addressing biases and barriers

Area of Science:

• Pediatric Healthcare Technology
• Artificial Intelligence in Medicine
• Machine Learning Applications

Background:

• Current artificial intelligence and machine learning (AI/ML) in healthcare inadequately addresses pediatric data specificities and vulnerabilities of children and young people (CYP).
• AI/ML applications for CYP are concentrated in high-income countries, with limited progress in lower- and middle-income countries (LMICs).
• Existing AI/ML efforts in LMICs are largely in the proof-of-concept stage, highlighting a global disparity in development and implementation.

Purpose of the Study:

• To identify global challenges and necessary shifts for effective AI/ML implementation in pediatric research and practice.
• To highlight the critical need for child-specific technical considerations throughout the AI/ML lifecycle.
• To advocate for equitable AI/ML deployment in child health, addressing governance and inclusivity.

Main Methods:

• This narrative review synthesizes existing literature on AI/ML in pediatric healthcare.
• It identifies intersecting challenges and proposes shifts for progress across multiple domains.
• The review emphasizes a human-centered and interdisciplinary approach.

Main Results:

• Child-specific technical considerations in AI/ML have been largely overlooked, impacting model effectiveness.
• Robust governance frameworks are essential for the safe and responsible deployment of AI/ML in CYP care.
• International collaboration, capacity building, and local grounding are crucial for equitable AI/ML adoption.

Conclusions:

• Realizing the universal benefits of AI/ML in child health requires greater international collaboration, capacity building, and strong oversight.
• AI/ML development must be inclusive of CYP and caregiver needs, particularly in LMICs, to avoid exacerbating inequalities.
• A human-centered, interdisciplinary approach grounded in local contexts is essential for sustainable and effective AI/ML tools in pediatric care.