Cohen-Sela Eyal, Lebenthal Yael, Brener Avivit, Regev Ravit, Hagenäs Lars
The Institute of Pediatric Endocrinology, Diabetes and Metabolism, Dana-Dwek Children's Hospital, Tel Aviv Sourasky Medical Center, 64239-06, Tel Aviv, Israel.
The School of Medicine, Gray Faculty of Medical & Health Sciences, Tel Aviv University, Tel Aviv, Israel.
Eur J Pediatr. 2025 Jul 18;184(8):490. doi: 10.1007/s00431-025-06321-3.
Growth assessment in achondroplasia requires disorder-specific growth charts incorporating sex- and age-specific values. Manual calculations are tedious and subject to error. We present an artificial intelligence (AI)-assisted tool that automates z-score calculations for pediatric patients with achondroplasia. The tool integrates European Lambda-Mu-Sigma (LMS) growth reference data for 9 anthropometric parameters: height, weight, body mass index, head circumference, sitting height, leg length, arm span, relative sitting height, and foot length. It inputs anthropometric measurements and transforms them into sex- and age-specific z-scores and percentiles in real time. Ten pediatric endocrinologists independently calculated anthropometric z-scores for 3 patients with achondroplasia using both the manual growth charts and the automated tool. Time-to-completion and accuracy were recorded and compared. The mean time required by the AI-assisted tool to calculate z-scores for all 9 parameters was significantly shorter than that required by manual calculation (23.4 ± 5.8 vs. 10.1 ± 2.8 min, p < 0.001). The tool demonstrated 100% agreement with manual LMS-based calculations and eliminated human errors to which manual calculations are subject, with significantly higher median absolute z-score deviation compared to the smart tool (0.17 [0.07-0.30] vs. 0 [0-0.01], p < 0.001).
This AI-assisted tool provides a user-friendly, accessible, and highly accurate method for automated growth assessment in pediatric achondroplasia. It facilitates efficient clinical and research applications, with potential for future integration into electronic health records and web-based platforms.
•Growth monitoring in achondroplasia requires syndrome-specific Lambda-Mu-Sigma based charts. •Manual z-score calculations are time-consuming and subject to error.
•We present an AI-assisted Excel tool that automates z-scores and percentile calculations for 9 anthropometric parameters. •Performance and inter-user reliability testing by 10 pediatric endocrinologists showed significantly improved speed and accuracy over manual methods.
软骨发育不全的生长评估需要特定疾病的生长图表，其中包含性别和年龄特异性值。手动计算繁琐且容易出错。我们展示了一种人工智能（AI）辅助工具，可自动为软骨发育不全的儿科患者计算z评分。该工具整合了9个人体测量参数的欧洲Lambda-Mu-Sigma（LMS）生长参考数据：身高、体重、体重指数、头围、坐高、腿长、臂展、相对坐高和足长。它输入人体测量数据，并实时将其转换为性别和年龄特异性的z评分和百分位数。十位儿科内分泌学家使用手动生长图表和自动化工具分别为3名软骨发育不全患者独立计算人体测量z评分。记录并比较完成时间和准确性。AI辅助工具计算所有9个参数的z评分所需的平均时间明显短于手动计算所需的时间（23.4±5.8分钟对10.1±2.8分钟，p<0.001）。该工具与基于LMS的手动计算显示出100%的一致性，并消除了手动计算易出现的人为误差，与智能工具相比，中位绝对z评分偏差明显更高（0.17[0.07 - 0.30]对0[0 - 0.01]，p<0.001）。
这种AI辅助工具为儿科软骨发育不全的自动生长评估提供了一种用户友好、易于使用且高度准确的方法。它有助于高效的临床和研究应用，未来有可能集成到电子健康记录和基于网络的平台中。
•软骨发育不全的生长监测需要基于特定综合征的Lambda-Mu-Sigma图表。
•手动z评分计算耗时且容易出错。
•我们展示了一种AI辅助的Excel工具，可自动计算9个人体测量参数的z评分和百分位数。
•10位儿科内分泌学家进行的性能和用户间可靠性测试表明，与手动方法相比，速度和准确性有显著提高。
