Related Experiment Video
Updated: Jun 30, 2025
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
An interpretable predictive deep learning platform for pediatric metabolic diseases
Hamed Javidi1,2,3, Arshiya Mariam1,3, Lina Alkhaled3,4
1Department of Quantitative Health Sciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH 44195, United States.
Insights
Early detection of pediatric metabolic diseases like type 2 diabetes is crucial. A deep learning model using longitudinal data, including BMI trajectories, accurately predicts disease onset, improving upon models using only recent data.
Area of Science:
- Pediatric Endocrinology
- Computational Health Science
- Machine Learning in Medicine
Background:
- Childhood metabolic diseases, including prediabetes, type 2 diabetes (T2D), and metabolic syndrome, are rising globally.
- These conditions significantly impair quality of life and increase the risk of chronic comorbidities.
- Effective early detection tools are urgently needed for timely intervention in pediatric populations.
Purpose of the Study:
- To develop and validate an interpretable deep learning model for predicting the onset of prediabetes, T2D, and metabolic syndrome in children.
- To assess the utility of longitudinal clinical data, including body mass index (BMI) trajectories, for improving predictive accuracy.
Main Methods:
- Utilized interpretable deep learning on electronic health record data from a large integrated health system.
- Included longitudinal clinical measurements, demographical data, and diagnosis codes for a cohort of 49,517 children (aged 2-18) with overweight or obesity.
- Compared model performance using longitudinal data versus models relying solely on the most recent BMI data.
Main Results:
- The model achieved area under the receiver operating characteristic curve (AUC) accuracies of up to 0.87 for T2D, 0.79 for metabolic syndrome, and 0.79 for prediabetes.
- Incorporating longitudinal data significantly improved AUCs by 11-13% compared to models using only the most recent BMI.
- BMI trajectories were identified as a consistently influential feature in the predictive model.
Conclusions:
- Longitudinal data analysis, specifically BMI trajectories, provides a more comprehensive patient health characterization and enhances predictive accuracy for pediatric metabolic diseases.
- Interpretable deep learning models leveraging historical data offer a promising approach for early detection and intervention.
- This methodology highlights the importance of considering temporal health trends over static measurements for improved clinical prediction.
Objectives:
Metabolic 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 Methods:
No 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.
Results:
The 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).
Discussion:
Despite 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.
More Related Videos
09:34A Virtual Machine Platform for Non-Computer Professionals for Using Deep Learning to Classify Biological Sequences of Metagenomic Data
Published on: September 25, 2021
07:35A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports
Published on: October 13, 2023
