Jun 2026· Journal of Engineering Research and Reports· Vol 28, pp. 1-15· 0 citations
TL;DR
The findings indicate that supervised machine learning methods can support early diabetes risk classification.
Abstract
Aims: This study evaluated the effectiveness of supervised machine learning algorithms for early diabetes risk classification using demographic, behavioural, cardiovascular, and general health-related indicators. It also examined the variables most strongly associated with diabetes occurrence patterns.
Study Design: A quantitative experimental design was used, based on supervised multiclass classification and comparative machine learning evaluation.
Place and Duration of Study: The experimental analysis was conducted using the Diabetes Health Indicators BRFSS2015 dataset between March 2026 and mid-May 2026.
Methodology: The dataset contained demographic, lifestyle, cardiovascular, and general health-related variables associated with diabetes conditions. Before model implementation, duplicate inspection, exploratory data analysis, feature standardisation, and variable consistency evaluation were performed to improve analytical stability. Diabetes status was used as the target variable in a multiclass classification framework. Logistic Regression, K-Nearest Neighbours, Naïve Bayes, and AdaBoost classifiers were implemented using Python-based machine learning libraries. The dataset was divided into training and testing subsets using an 80:20 ratio. Model performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC. Feature importance analysis and ROC curve comparison were also performed to assess classification behaviour and variable contribution patterns.
Results: Logistic Regression achieved the highest ROC-AUC value of 0.814 and demonstrated stable discrimination across diabetes categories. AdaBoost achieved the highest accuracy score of 0.847 and produced competitive precision, recall, and F1-score values. K-Nearest Neighbours showed moderate classification capability, whereas Naïve Bayes demonstrated comparatively weaker classification consistency. Feature importance analysis identified HighBP, GenHlth, Age, BMI, CholCheck, and HighChol as influential variables.
Conclusion: The findings indicate that supervised machine learning methods can support early diabetes risk classification. Cardiovascular conditions, obesity-related indicators, and general health variables were important contributors to classification behaviour within the implemented framework.
Background: Early detection of diabetes and prediabetes is important for reducing long-term complications.
Aim: This study comparatively evaluated four machine learning algorithms for three-class diabetes classification using routinely available clinical indicators from the National Health and Nutrition Examination Survey.
Methods: The analytical sample included 2,029 participants classified as normal (62.0%), prediabetic (26.7%), or diabetic (11.3%). Recursive Feature Elimination with 10-fold cross-validation was used to select predictors from 27 candidate variables. Six features were retained: fasting glucose, age, diabetes history, insulin level, waist circumference, and systolic blood pressure. Multinomial Logistic Regression, Decision Trees, Random Forest, and XGBoost were trained using a 70/30 stratified train-test split and evaluated using accuracy, Cohen’s Kappa, and class-specific performance metrics.
Results: Random Forest achieved the highest overall test performance, with 78.1% accuracy and a Cohen’s Kappa of 0.471. XGBoost, Multinomial Logistic Regression, and Decision Trees achieved accuracies of 72.6%, 73.0%, and 72.4%, respectively. All models showed high specificity for diabetes detection, exceeding 97%. Prediabetes classification remained difficult across algorithms, with sensitivity ranging from 38% to 42%.
Conclusion: Random Forest provided the best overall performance for three-class diabetes classification in this analytical sample. However, modest agreement, low prediabetes sensitivity, potential label leakage from fasting glucose, and the absence of external validation indicate that further evaluation is required before clinical application.
Ayeni Taiwo Michael, Odukoya Ayooluwa, Ilesanmi Opeyemi· Asian Research Journal of Ma...· 0 citations
This study addresses the challenge of delayed diagnosis of diabetes, a condition that often leads to severe complications if not detected early. The primary objective is to evaluate and compare the performance of three machine learning classifiers XGBoost, Support Vector Machine (SVM), and Random Forest for early diabetes prediction using clinical and lifestyle data. The study utilizes the Diabetes Health Indicators dataset, which includes features such as body mass index (BMI), blood pressure, cholesterol levels, and physical activity. The dataset was sourced from a publicly available repository and preprocessed through handling missing values, feature scaling, and encoding categorical variables. The models were trained on the processed dataset and evaluated using accuracy, precision, recall, and F1-score metrics, alongside exploratory data analysis to understand feature relationships. Results show that all three models performed effectively, with XGBoost achieving the highest accuracy of 85.11%, followed by SVM at 84.82%, and Random Forest at 83.16%. These findings highlight the strength of ensemble and boosting techniques in handling complex health data and accurately predicting diabetes risk. In conclusion, machine learning models demonstrate strong potential for supporting early diabetes diagnosis and improving clinical decision-making. It is recommended that healthcare systems adopt XGBoost-based predictive models in clinical decision support tools for early screening, while future studies should validate these models using real-world clinical data to enhance reliability and generalizability.
Idehen Emmanuel Imafidon, Chikere Obinna Munachiso, Dominic Evans Onyebuchi et al.· International Journal of Sci...· 0 citations
Diabetes is a chronic metabolic disease characterized by high blood glucose levels that can lead to severe complications affecting various organs. The increasing prevalence of diabetes requires an accurate and efficient diagnostic approach to support early detection and medical decision-making. The development of machine learning technology provides opportunities to assist disease diagnosis by analyzing patient health data and identifying patterns associated with diabetes conditions. This study aims to compare the accuracy performance of the Naive Bayes and CatBoost algorithms in diagnosing diabetes using the confusion matrix evaluation method. The research method consists of several stages, including data collection, data preprocessing, model selection, model training, and model evaluation. The dataset used was obtained from Kaggle, consisting of 768 records with nine attributes, including pregnancies, glucose level, blood pressure, skin thickness, insulin, body mass index, diabetes pedigree function, age, and outcome classification. The models were trained using 80% training data and 20% testing data. The evaluation results showed that the Naive Bayes algorithm achieved an accuracy of 79%, while the CatBoost algorithm obtained a higher accuracy of 81%. These findings indicate that CatBoost provides better classification performance compared to Naive Bayes for diabetes diagnosis in this dataset. Therefore, CatBoost can be considered a more effective machine learning approach to support diabetes prediction. Future research is recommended to utilize larger datasets, optimize model parameters, and explore other advanced algorithms to improve diagnostic accuracy.
Basyach Aryodito Tranggono Heryanto, Onny Marleen· Journal of social research· 0 citations
Diabetes mellitus has become a major global health threat, and many undiagnosed cases remain undetected due to some limitations of the conventional diagnostic methods. Despite the promising results of machine learning (ML) for early diabetes diagnosis, the majority of the current research assessing algorithms either uses insufficient metrics or does not follow a consistent assessment approach. This paper addresses that gap by utilising an integrated evaluation framework. The framework includes feature importance analysis, Pearson correlation assessment, confusion matrix decomposition, and ROC-AUC comparison. It applies this framework to the Pima Indians Diabetes Dataset (mde) and four popular ML classification algorithms: Naive Bayes, Decision Tree, Random Forest, and Logistic Regression. The most significant predictors, according to our feature analysis, were glucose (27.6%), body mass index (16.0%), age (12.7%), and diabetes pedigree function (12.7%). Among the classifiers, Random Forest exhibited the greatest accuracy (76.0%) and precision (68.1%), Naive Bayes the best recall (64.8%), and Logistic Regression the highest AUC-ROC (82.3%). For patients at high risk, the models' virtual projections across all three risk profiles were in agreement. Model selection should be determined by the unique clinical screening aim, since these findings suggest that there is no one better universal method. Random Forest and Logistic Regression are the most promising for assisting in preliminary diabetes prediction, although further validation on diversity datasets is needed prior to clinical deployment.
F. Setiawan, Tri Sugihartono· SinkrOn· 0 citations
Six supervised learning models were developed and compared for diabetes prediction using a dataset and compared for diabetes prediction using a 100k patients records with eight clinical features including gender, age, hypertension, smoking history, heart disease, BMI, HbA1c level, and blood glucose level.
Diabetes is one of the leading causes of morbidity and mortality worldwide. To avoid difficult management of the condition, there is need to predict early onset of the condition. This study investigated the application of machine learning techniques and exploratory data analysis to forecast early-onset of diabetes using the PIMA Indian dataset. Preprocessing included handling missing values and standardization, leading to the development and evaluation of 5 models which include Logistic Regression, K-Nearest Neighbors, Random Forest, Support Vector Machine, and Neural Network. Exploratory analysis identified glucose concentration, body mass index, and age as the most influential features. Random Forest achieved the highest accuracy (0.74%) while both Random Forest and Logistic Regression attained the best ROC-AUC score of 0.81%. Feature importance analysis emphasized the predictive significance of glucose and BMI, aligning with clinical knowledge of diabetes risk factors. Despite the promising results the study acknowledged limitations related to the PIMA dataset's demographic scope and the moderate complexity of neural networks, highlighting areas for future enhancement. Ethical considerations, including data privacy and algorithmic bias, were addressed to ensure responsible model development.
Abdulrahman Nasiru Sada, Eli. A. Jiya, Yahaya Muhammad Umar et al.· Journal of Science Research...· 0 citations