Related Experiment Video
Updated: Aug 16, 2026
Biochemical Titration of Glycogen In vitro
Published on: November 24, 2013
A machine learning model accurately identifies glycogen storage disease Ia patients based on plasma acylcarnitine
Joost Groen1, Bas M de Haan2, Ruben J Overduin3
1Laboratory of Metabolic Diseases, Department of Laboratory Medicine, University Medical Center Groningen, University of Groningen, Hanzeplein 1, Postbus, Groningen, 30001 - 9700 RB, the Netherlands. j.groen@umcg.nl.
Insights
Machine learning accurately identifies Glycogen storage disease (GSD) Ia using plasma acylcarnitine profiles, offering a potential biomarker for this rare metabolic disorder. This approach could aid in early diagnosis and newborn screening for GSD Ia.
Area of Science:
- Biochemistry
- Genetics
- Computational Biology
Background:
- Glycogen storage disease (GSD) Ia is a rare inherited metabolic disorder affecting carbohydrate metabolism.
- Diagnosis currently relies on biomarkers and genetic confirmation, lacking a specific, reliable marker.
- Altered lipid metabolism and mitochondrial function in GSD Ia patients suggest potential for acylcarnitine profiling.
Purpose of the Study:
- To develop and validate a machine learning model for identifying GSD Ia patients using plasma acylcarnitine profiles.
- To address the challenge of class imbalance in ultra-rare disease datasets.
Main Methods:
- Collected plasma acylcarnitine profiles from 3958 patients, including 31 with GSD Ia.
- Employed gradient-boosted tree models with hyperparameter tuning and feature selection.
- Utilized nested cross-validation and a held-out test set to assess model generalization.
Main Results:
- The machine learning model accurately identified 5/6 GSD Ia patients in the test set with minimal false positives.
- Achieved high performance metrics, including a mean ROC AUC of 0.955 and PR AUC of 0.674 in nested cross-validation.
- Identified key predictive acylcarnitine features: C16-carnitine, C14OH-carnitine, total carnitine, and acetylcarnitine.
Conclusions:
- Machine learning effectively identifies GSD Ia patients via plasma acylcarnitine analysis, leveraging subtle metabolic alterations.
- The model demonstrates high sensitivity, specificity, and interpretability, suggesting its utility for early detection.
- This approach holds promise for integrating GSD Ia into newborn screening programs, addressing underrepresentation in ML studies.
Background:
Glycogen storage disease (GSD) Ia is an ultra-rare inherited disorder of carbohydrate metabolism. Patients often present in the first months of life with fasting hypoketotic hypoglycemia and hepatomegaly. The diagnosis of GSD Ia relies on a combination of different biomarkers, mostly routine clinical chemical markers and subsequent genetic confirmation. However, a specific and reliable biomarker is lacking. As GSD Ia patients demonstrate altered lipid metabolism and mitochondrial fatty acid oxidation, we built a machine learning model to identify GSD Ia patients based on plasma acylcarnitine profiles.
Methods:
We collected plasma acylcarnitine profiles from 3958 patients, of whom 31 have GSD Ia. Synthetic samples were generated to address the problem of class imbalance in the dataset. We built several machine learning models based on gradient-boosted trees. Our approach included hyperparameter tuning and feature selection and generalization was checked using both nested cross-validation and a held-out test set.
Results:
The binary classifier was able to correctly identify 5/6 GSD Ia patients in a held-out test set without generating significant amounts of false positive results. The best model showed excellent performance with a mean received operator curve (ROC) AUC of 0.955 and precision-recall (PR) curve AUC of 0.674 in nested CV.
Conclusions:
This study demonstrates an innovative approach to applying machine learning to ultra-rare diseases by accurately identifying GSD Ia patients based on plasma free carnitine and acylcarnitine concentrations, leveraging subtle acylcarnitine abnormalities. Acylcarnitine features that were strong predictors for GSD Ia include C16-carnitine, C14OH-carnitine, total carnitine and acetylcarnitine. The model demonstrated high sensitivity and specificity, with selected parameters that were not only robust but also highly interpretable. Our approach offers potential prospect for the inclusion of GSD Ia in newborn screening. Rare diseases are underrepresented in machine learning studies and this work highlights the potential for these techniques, even in ultra-rare diseases such as GSD Ia.
Related Concept Videos
01:22Diabetes Mellitus: Overview and Type I Subtype
Type 1 diabetes is an autoimmune disease in which the immune system mistakenly attacks and destroys the insulin-producing beta cells in the pancreas. As a result, the body is unable to produce sufficient insulin, and individuals with...
01:26Type II Diabetes I: Introduction
01:25Type II Diabetes Mellitus III: Clinical Manifestations and Diagnosis