Related Experiment Video
Updated: Jan 14, 2026
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
7.4K
Machine learning-driven Diabetes Health Tracer (DHT): Optimizing prognosis using RaSK_GraDe and RaSK_GraDeL models
Muhammad Noman1, Maria Hanif1, Abdul Hameed2
1Department of Software Engineering and Artificial Intelligence, Iqra University, H-9, Islamabad, Pakistan.
Plos One
|October 21, 2025
Summary
Machine learning models show high accuracy in predicting diabetes mellitus, offering improved healthcare management. Ensemble methods like Voting Classifier and Stacking Model achieved over 98% accuracy on the Diabetes Health Tracer dataset.
Area of Science:
- Medical Informatics
- Computational Biology
- Health Data Science
Background:
- Diabetes mellitus is a major global health concern, with significant impact in South Asia.
- Traditional diabetes prediction methods have limitations in reliability and efficiency.
- Machine learning (ML) offers advanced capabilities for accurate disease prediction.
Purpose of the Study:
- To comparatively analyze various ML algorithms for diabetes prediction.
- To evaluate the performance of ensemble methods, including Voting Classifier (RaSK_GraDe) and Stacking Model (RaSK_GraDeL).
- To assess the effectiveness of ML on diverse datasets, including the proposed Diabetes Health Tracer (DHT) dataset.
Main Methods:
- Comparative analysis of ML algorithms: Random Forest, Decision Tree, SVM, KNN, Gradient Boosting.
- Application of ensemble techniques: Voting Classifier (RaSK_GraDe) and Stacking Model (RaSK_GraDeL).
- Data pre-processing: handling missing values, outliers, normalization, and class balancing (SMOTE).
- Hyperparameter tuning using cross-validation and Random Search.
Main Results:
- Ensemble methods achieved high predictive accuracy: RaSK_GraDe (98.03%) and RaSK_GraDeL (98.55%) on the DHT dataset.
- Pre-processing and hyperparameter tuning enhanced model robustness and performance.
- ML algorithms demonstrated superior performance compared to traditional methods.
Conclusions:
- Machine learning techniques are highly effective for diabetes mellitus prediction.
- Ensemble methods, particularly stacking, show significant promise for improving diagnostic accuracy.
- The findings support the advancement of personalized treatment and healthcare management for diabetes.
