甲巯咪唑
医学
儿科
机器学习
内分泌学
内科学
计算机科学
甲状腺
作者
Kanghyuck Lee,Joon Young Kim,Taehoon Ko,Kyungchul Song
摘要
This study developed a machine learning model using data from 142 children and adolescents with hyperthyroidism, with external validation conducted on 63 patients from another institution. Input variables included age, sex, height SDS, weight SDS, BMI SDS, T3, free T4, TSH, follow-up interval, and previous methimazole dose. SHAP analysis identified T3, TSH, and free T4 as the most influential factors. The model demonstrated robust performance with an RMSE of 5.15 mg (internal validation) and 3.54 mg (external validation), highlighting the potential of machine learning to optimize methimazole dose adjustment in pediatric hyperthyroidism.
科研通智能强力驱动
Strongly Powered by AbleSci AI