An interpretable predictive deep learning platform for pediatric metabolic diseases.
An interpretable predictive deep learning platform for pediatric metabolic diseases.
复制标题
针对儿科代谢疾病的可解释预测深度学习平台。
DOI：
10.1093/jamia/ocae049
复制
发表时间：
2024
期刊：
影响因子：
--
通讯作者：
Rotroff,DanielM
中科院分区：
文献类型：
--
作者：
Javidi,Hamed;Mariam,Arshiya;Alkhaled,Lina;Pantalone,KevinM;Rotroff,DanielM
目的：儿童代谢性疾病在全球范围内日益增多，易诱发多种慢性共病，严重影响生活质量。需要工具进行早期检测，及时干预，以防止或减缓这些长期complications.Materials和MethodsNo临床上可用的工具的发展，目前正在广泛使用，可以预测代谢性疾病在儿科患者的发病。在这里，我们使用可解释的深度学习，利用纵向临床测量，人口统计学数据和来自大型综合卫生系统的电子健康记录数据的诊断代码来预测儿科队列中前驱糖尿病，2型糖尿病（T2 D）和代谢综合征的发病。结果该队列包括49 517名2-18岁超重或肥胖儿童（54.9%男性，73%白人），中位随访时间为7.5年，平均体重指数（BMI）百分位数为88.6%。我们的模型表明，在预测T2 D、代谢综合征和前驱糖尿病时，受试者工作特征曲线（AUC）下的面积准确度分别高达0.87、0.79和0.79。尽管大多数风险计算器仅使用最近可用的数据，但与使用最新BMI的模型相比，合并纵向数据后，T2 D、综合征和前驱糖尿病的AUC分别提高了13.04%、11.48%和11.67（P<2.2 × 10-16）。讨论尽管大多数风险计算器只使用最新数据，合并纵向数据提高了模型的准确性，因为利用轨迹提供了患者健康史的更全面表征。我们的可解释模型表明，BMI轨迹一直被认为是最有影响力的预测特征之一，突出了在可用时纳入纵向数据的优势。
ObjectivesMetabolic disease in children is increasing worldwide and predisposes a wide array of chronic comorbid conditions with severe impacts on quality of life. Tools for early detection are needed to promptly intervene to prevent or slow the development of these long-term complications.Materials and MethodsNo clinically available tools are currently in widespread use that can predict the onset of metabolic diseases in pediatric patients. Here, we use interpretable deep learning, leveraging longitudinal clinical measurements, demographical data, and diagnosis codes from electronic health record data from a large integrated health system to predict the onset of prediabetes, type 2 diabetes (T2D), and metabolic syndrome in pediatric cohorts.ResultsThe cohort included 49 517 children with overweight or obesity aged 2-18 (54.9% male, 73% Caucasian), with a median follow-up time of 7.5 years and mean body mass index (BMI) percentile of 88.6%. Our model demonstrated area under receiver operating characteristic curve (AUC) accuracies up to 0.87, 0.79, and 0.79 for predicting T2D, metabolic syndrome, and prediabetes, respectively. Whereas most risk calculators use only recently available data, incorporating longitudinal data improved AUCs by 13.04%, 11.48%, and 11.67% for T2D, syndrome, and prediabetes, respectively, versus models using the most recent BMI (P<2.2 × 10–16).DiscussionDespite most risk calculators using only the most recent data, incorporating longitudinal data improved the model accuracies because utilizing trajectories provides a more comprehensive characterization of the patient’s health history. Our interpretable model indicated that BMI trajectories were consistently identified as one of the most influential features for prediction, highlighting the advantages of incorporating longitudinal data when available.