Abstract
IMPORTANCE: Artificial intelligence (AI)-based technologies hold promise for faster and more accurate devices in health care; however, little is known about their availability to pediatric patients. A comprehensive analysis of US Food and Drug Administration (FDA) regulatory submissions is necessary to identify technologies with pediatric labeling. OBJECTIVE: To characterize AI devices marketed in the US, identify those with pediatric indications, and provide clues on AI-specific barriers to pediatric device innovation. DESIGN, SETTING, AND PARTICIPANTS: This retrospective descriptive cross-sectional study was conducted using public data from the FDA's AI-Enabled Medical Device List. Marketing submissions reviewed by the FDA between November 1995 and June 2024 were analyzed. MAIN OUTCOMES AND MEASURES: Prevalence of pediatric AI devices and their main characteristics (eg, clinical area, review time, and year of FDA marketing decision). RESULTS: Among 952 submissions, 42 (4.4%) included pediatric age ranges (0-17 years). The first pediatric-inclusive device was cleared in 2015, and 5 exclusively pediatric technologies were introduced between 2020 and 2024. Of 18 clinical areas, radiology comprised 723 of all devices (75.9%) but only 18 of the devices specifically labeled for pediatrics (42.9%), whereas neurology comprised 34 devices overall (3.6%) vs 13 pediatric devices (31.0%); 10 clinical areas (55.6%) were missing among pediatric devices. The median (IQR) FDA review time was significantly longer for pediatric than for nonpediatric devices (162 [114-228] days; 95% CI, 151-212 days vs 134 [87-214] days; 95% CI, 149-162 days; 2-sided Mann-Whitney U test P = .049). Based on National Clinical Trial Identifiers in FDA summaries, clinical trial registration was noted in 6 pediatric (14.3%) vs 20 of 906 nonpediatric (2.2%) submissions. CONCLUSIONS AND RELEVANCE: In this study, pediatric devices were rare, emerged recently, and had longer review times and a higher proportion of registered clinical trials compared with nonpediatric devices, suggesting expectations for more pediatric-specific evidence despite unchanged statutory standards. To address gaps in pediatric device development, the FDA should standardize age labeling and validation requirements for AI-enabled technologies.