Abstract

Diabetes mellitus remains one of the fastest-growing chronic diseases worldwide, and early risk identification is essential for timely clinical intervention. This paper presents a comparative evaluation of two supervised machine learning algorithms, Logistic Regression (LR) and Random Forest (RF), for diabetes risk classification using the DiaBD dataset, a recently published clinical dataset of 5,288 patient records from Bangladesh. Chi-square and analysis-of-variance (ANOVA) feature selection identified six clinically relevant predictors: hypertension status, glucose level, systolic and diastolic blood pressure, age, and weight. The Synthetic Minority Over-sampling Technique (SMOTE) was applied to the training partition only, to correct for class imbalance without leaking information into the test set. Both models were trained on an 80/20 train-test split and evaluated using accuracy, precision, recall, F1-score, and the area under the receiver operating characteristic curve (AUC-ROC). Random Forest achieved a higher overall accuracy (0.88) and a marginally higher AUC (0.83) than Logistic Regression (0.81 accuracy, 0.82 AUC); Logistic Regression, however, achieved substantially higher recall for the diabetic class (0.66 versus 0.44). The two models' F1-scores were nearly identical (0.32 for both), which shows that a single blended metric conceals the clinically important difference in how each model distributes its errors. Precision remained modest for both models (0.21-0.25), indicating that a six-variable feature set is not sufficient for high-precision clinical deployment on its own. We argue that model selection for diabetes screening should be guided by the clinical cost of missed diagnoses rather than accuracy alone, and we recommend threshold calibration, cost-sensitive learning, and external validation on additional populations as directions for future work.

References

• World Health Organization, "Global Health Observatory: Diabetes - prevalence and treatment, 1990-2022," WHO, Geneva, Switzerland, 2024. [Online]. Available: https://www.who.int/data/gho/indicator-metadata-registry/imr-details/761
• M. J. Hossain, M. Al-Mamun, and M. R. Islam, "Diabetes mellitus, the fastest growing global public health concern: Early detection should be focused," Health Sci. Rep., vol. 7, no. 3, art. e2004, 2024.
• S. Safiri et al., "Prevalence, deaths and disability-adjusted life-years (DALYs) due to type 2 diabetes and its attributable risk factors in 204 countries and territories, 1990-2019," Front. Endocrinol., vol. 13, art. 838027, 2022.
• J. J. Khanam and S. Y. Foo, "A comparison of machine learning algorithms for diabetes prediction," ICT Express, vol. 7, no. 4, pp. 432-439, 2021.
• M. K. Hasan, M. A. Alam, D. Das, E. Hossain, and M. Hasan, "Diabetes prediction using ensembling of different machine learning classifiers," IEEE Access, vol. 8, pp. 76516-76531, 2020.
• C. C. Olisah, L. Smith, and M. Smith, "Diabetes mellitus prediction and diagnosis from a data preprocessing and machine learning perspective," Comput. Methods Programs Biomed., vol. 220, art. 106773, 2022.
• H. Zhou, Y. Xin, and S. Li, "A diabetes prediction model based on Boruta feature selection and ensemble learning," BMC Bioinformatics, vol. 24, art. 224, 2023.
• A. Doğru, S. Buyrukoğlu, and M. Arı, "A hybrid super ensemble learning model for the early-stage prediction of diabetes risk," Med. Biol. Eng. Comput., vol. 61, no. 3, pp. 785-797, 2023.
• T. Alghamdi, "Prediction of diabetes complications using computational intelligence techniques," Appl. Sci., vol. 13, no. 5, art. 3030, 2023.
• U. Ahmed, G. F. Issa, M. A. Khan, S. Aftab, M. F. Khan, R. A. T. Said, T. M. Ghazal, and M. Ahmad, "Prediction of diabetes empowered with fused machine learning," IEEE Access, vol. 10, pp. 8529-8538, 2022.
• H. M. Deberneh and I. Kim, "Prediction of type 2 diabetes based on machine learning algorithm," Int. J. Environ. Res. Public Health, vol. 18, no. 6, art. 3317, 2021.
• I. Tasin, T. U. Nabil, S. Islam, and R. Khan, "Diabetes prediction using machine learning and explainable AI techniques," Healthc. Technol. Lett., vol. 10, no. 1-2, pp. 1-10, 2023.
• M. T. García-Ordás, C. Benavides, J. A. Benítez-Andrades, H. Alaiz-Moretón, and I. García-Rodríguez, "Diabetes detection using deep learning techniques with oversampling and feature augmentation," Comput. Methods Programs Biomed., vol. 202, art. 105968, 2021.
• M. M. Bukhari, B. F. Alkhamees, S. Hussain, A. Gumaei, A. Assiri, and S. S. Ullah, "An improved artificial neural network model for effective diabetes prediction," Complexity, vol. 2021, art. 5525271, 2021.
• J. Zhao et al., "Attention-oriented CNN method for type 2 diabetes prediction," Appl. Sci., vol. 14, no. 10, art. 3989, 2024.
• T. T. Prama, M. J. Rahman, M. Zaman, F. Sarker, and K. A. Mamun, "DiaBD: A diabetes dataset for enhanced risk analysis and research in Bangladesh," Data Brief, art. 111746, 2025, doi: 10.1016/j.dib.2025.111746.
• A. Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 2nd ed. Sebastopol, CA, USA: O'Reilly Media, 2019.
• N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, "SMOTE: Synthetic minority over-sampling technique," J. Artif. Intell. Res., vol. 16, pp. 321-357, 2002.
• D. W. Hosmer, S. Lemeshow, and R. X. Sturdivant, Applied Logistic Regression, 3rd ed. Hoboken, NJ, USA: Wiley, 2013.
• L. Breiman, "Random forests," Mach. Learn., vol. 45, no. 1, pp. 5-32, 2001.
• T. Zhu, K. Li, P. Herrero, and P. Georgiou, "Deep learning for diabetes: A systematic review," IEEE J. Biomed. Health Inform., vol. 25, no. 7, pp. 2744-2757, 2021.