The relationship between childhood obesity and bone age advancement based on artificial intelligence digital radiography image analysis algorithm

Artificial intelligence (AI) shows promise in medical imaging, but its performance in obese pediatric bone age requires validation. This study aimed to develop an AI-aided system using digital radiography (DR) and to investigate the relationship between childhood obesity and bone age advancement. In this cross-sectional study, 442 children and adolescents (6-18 years) were enrolled between December 2023 and February 2025. Participants were categorized into normal weight, overweight, and obese groups based on BMI. Left-hand wrist DR images were obtained. Bone age was assessed automatically using United Imaging Intelligence Greulich-Pyle (G-P) atlas AI software and manually verified by two physicians. Physical, metabolic, and hormonal data were collected. Bone age advancement (bone age - chronological age) was analyzed against obesity-related indicators. AI assessment averaged (2.2 ± 0.6) seconds and showed high consistency with manual results (ICC = 0.992). Bone age advancement was greatest in the obese group (1.13 ± 0.88 years), followed by overweight (0.69 ± 0.72 years) and normal weight (0.16 ± 0.72 years). Advanced bone age (≥1 year) occurred in 67.8% of obese participants, significantly higher than the 9.8% in normal weight participants. Bone age advancement positively correlated with BMI, waist circumference, and HOMA-IR. Multivariable logistic regression identified overweight (OR = 3.85) and obesity (OR = 12.63) as independent risk factors for accelerated bone age. ROC analysis indicated HOMA-IR had moderate predictive ability for bone age progression. The AI-assisted DR bone age assessment system demonstrated high efficiency, accuracy, and reliability in obese children, supporting its use in large-scale screening. Obesity, especially with central adiposity and insulin resistance, was strongly associated with accelerated bone age.