Focused review on artificial intelligence for disease detection in infants

Area of Science:

• Medical Informatics
• Pediatric Health
• Artificial Intelligence

Background:

• Increasing research utilizes artificial intelligence (AI) for disease detection and prediction, particularly in vulnerable infant populations.
• The advent of large language models (LLMs) like ChatGPT signifies a new era in AI, with potential impacts on medical research yet to be fully understood.
• This study focuses on pre-ChatGPT developments in AI for infant disease detection and prediction.

Purpose of the Study:

• To summarize recent (2018-2022) advancements in automated disease detection and prediction in infants using AI, preceding the widespread impact of LLMs.
• To analyze trends in research activity, medical conditions, data types, AI methodologies, and performance metrics in infant-focused AI studies.

Main Methods:

• Systematic literature search of PubMed and IEEE Xplore for original articles published between 2018 and 2022.
• Inclusion of 154 articles from an initial search of 927.
• Analysis of research trends over time and detailed examination of 2022 articles for medical conditions, data types, tasks, AI approaches, and model performance.

Main Results:

• A consistent increase in research activity in AI for infant health was observed from 2018 to 2022.
• "Certain conditions originating in the perinatal period" was the most frequent medical focus.
• Clinical, demographic, and laboratory data were commonly used, with deep neural networks being the predominant AI approach, though traditional methods remain relevant.

Conclusions:

• AI demonstrates significant potential to aid diagnostic procedures for infants.
• The findings suggest AI tools could become valuable assets in infant healthcare.
• Future developments, potentially boosted by LLMs, are expected to further advance AI applications in this field.