Collaboration and Future Directions in Artificial Intelligence for Neonatology

Abstract

Amidst the dual crisis of increasing birth rate of high-risk newborns and shortage of neonatologists in Korea, neonatal intensive care units (NICU) have become a complex medical environment that generates vast amounts of data. Artificial Intelligence (AI) technology is gaining attention as an innovative solution in data-driven environments, enabling the early prediction of diseases, optimization of treatment decisions, and reduction of medical staff workload. This review aimed to examine the historical development of AI and explore areas where AI can be applied in various clinical fields of neonatology. Furthermore, it analyzes Korean medical AI research trends and related policies to shed light on their status, proposes effective measures for establishing an AI research infrastructure utilizing NICU data and interdisciplinary collaboration models, and suggests future directions for AI research in the field of neonatology. AI has evolved through two 'winters' into the current era of generative AI, with core technologies such as machine learning, deep learning, and large-language models being applied in neonatology. Key application areas include the early prediction of conditions such as sepsis, necrotizing enterocolitis, and bronchopulmonary dysplasia using heart rate characteristics, medical imaging, and multimodal data, as well as treatment optimization through Clinical Decision Support Systems. In Korea, R&D is being actively promoted under government leadership, including the establishment of the 'Korean Specialized Big Data for Critical Care (K-MIMIC)' and the 'Advanced Research Projects Agency for Health (ARPA-H) Project,' while an institutional foundation is also being established with laws like the 'Digital Medical Products Act.' For the successful implementation of AI, a standardized multicenter data infrastructure such as the Korea Neonatal Network and a clinician-led industry– academia–research–hospital collaboration model are essential. Future AI research in neonatology must move beyond static risk prediction toward a dynamic intervention recommendation system that suggests optimal treatments based on real-time data. Ultimately, the goal is to establish a Smart NICU that realizes precision, automation, and remote capabilities for neonatal care. To this end, by having medical professionals lead the research ecosystem and strengthen interdisciplinary collaboration, AI will play a decisive role in saving the lives of newborns and ensuring their healthy future.

Ähnliche Arbeiten