Related Experiment Video
Updated: Jun 15, 2025

Modeling and Evaluation of Murine Diabetic Cardiomyopathy Model
Published on: November 29, 2024
Machine learning-based reproducible prediction of type 2 diabetes subtypes
Hayato Tanabe1,2, Masahiro Sato1, Akimitsu Miyake3
1Department of Diabetes, Endocrinology, and Metabolism, Fukushima Medical University School of Medicine, Fukushima, Japan.
A new machine learning model accurately classifies type 2 diabetes subtypes, offering consistent, long-term predictions of complications. This approach aids personalized treatment by overcoming limitations of existing methods, even with missing clinical data.
Area of Science:
- Endocrinology and Metabolism
- Computational Biology and Bioinformatics
- Genetics and Genomics
Background:
- Clustering-based subclassification of type 2 diabetes (T2D) shows promise for personalized therapy by reflecting pathophysiology and genetic predisposition.
- Ahlqvist's classification is validated for predicting complications but lacks temporal consistency and requires non-routine HOMA2 indices.
- There is a need for a robust method to consistently classify T2D subtypes over time using routinely available clinical data.
Purpose of the Study:
- To develop and validate a machine learning (ML) model for consistent classification of T2D into Ahlqvist's subtypes.
- To assess the model's performance using readily available variables, including in cases with missing HOMA2 indices.
- To evaluate the long-term consistency and predictive ability of the ML-derived subtypes for diabetes complications.
Main Methods:
- A random forest (RF) algorithm (T2DRF15) was trained on 15 variables from a Japanese T2D cohort (Cohort 1) pre-classified using k-means clustering (T2Dkmeans).
- The model was externally validated on a separate cohort (Cohort 2) and tested on a subset with missing insulin-related variables.
- Kaplan-Meier analysis assessed complication risks, and subtype consistency was evaluated over time.
Main Results:
- T2DRF15 achieved 94% accuracy in predicting T2Dkmeans subtypes, with high performance in external validation (86.3%) and imputation scenarios (82.9%).
- The ML-derived subtypes showed distinct risks for diabetic retinopathy (SIDD) and chronic kidney disease (SIRD), mirroring T2Dkmeans findings.
- Excluding an 'undecidable' cluster improved predictive accuracy and demonstrated significantly higher subtype consistency over time compared to T2Dkmeans.
Conclusions:
- The developed ML model offers a consistent and clinically applicable method for classifying T2D subtypes using routinely available data.
- This approach facilitates personalized treatment strategies by predicting glycaemic control, complications, and outcomes with long-term stability.
- Further multiethnic validation is warranted to assess broader applicability in research and clinical practice.
More Related Videos
Related Concept Videos
Diabetes 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...
Diabetes Mellitus: Type 2 and Gestational
Pathophysiology of Diabetes
Type 1 diabetes is characterized by autoimmune-mediated destruction of pancreatic β cells, with environmental factors potentially triggering this process in genetically susceptible individuals. Despite many not having a family history, certain genes increase susceptibility,...
Carbohydrate Metabolism
Starch accounts for approximately 60% of the carbohydrates consumed by humans. Since amylase enzymes cannot function in the stomach's acidic environment, starch can only be digested in the mouth and small intestine. Simple sugars are found naturally in milk and fruits in...
Diabetes: Symptoms, Diagnosis, and Complications

