Data Modeling
Mathematical and Computational Data Modeling
1
Citations
7k
Views
37
Articles
Volume 1 Number 1 2020
RESEARCH ARTICLE (Open Access)
Predicting Prediabetes from Simple Clinical Questionnaires: A Machine Learning Approach Using K-Nearest Neighbors and Decision Tree Models
Kamruzzaman Mithu 1*
Data Modeling 1 (1) 1-11 https://doi.org/10.25163/data.1110858
Submitted: 25 September 2020 Revised: 12 November 2020 Accepted: 23 November 2020 Published: 25 November 2020
Abstract
Background. Prediabetes affects an estimated 96 million American adults, yet more than 80% remain unaware of their condition — a gap that matters, since prediabetes is largely reversible while type 2 diabetes generally is not. Despite this, most predictive modeling efforts in the diabetes literature have focused on undiagnosed diabetes broadly, leaving prediabetes-specific prediction comparatively underexplored, partly because of scarce labeled data and persistent class imbalance. Methods. This study used a retrospective dataset of 1,000 patient records sourced from Medical City Hospital and the Al-Kindy Teaching Hospital in Iraq, comprising demographic and routine laboratory variables (e.g., BMI, HbA1c, lipid profile, blood urea). Feature selection combined manual review with information-gain analysis; outliers were identified and removed using boxplot- and histogram-based screening; and class imbalance across non-diabetic, prediabetic, and diabetic groups was addressed using SMOTE-based oversampling. Two classifiers — K-Nearest Neighbors and Decision Tree — were trained on a 70/30 split and evaluated using accuracy, precision, recall, F1-score, and AUROC. Results. The Decision Tree model achieved 82% accuracy, modestly outperforming KNN's 79%. Both models classified diabetic patients more reliably than prediabetic ones, a pattern attributable largely to the underlying class imbalance that oversampling only partially corrected. Conclusion. Reasonably accurate prediabetes screening appears achievable using a small set of accessible, low-cost variables, though prediabetes-specific detection remains disproportionately difficult given current data limitations. Expanding future datasets to include lifestyle and family-history variables, alongside testing ensemble methods, may meaningfully improve model reliability for this often-overlooked, high-risk population.
Keywords: prediabetes; machine learning; K-nearest neighbors; decision tree; early detection
References
American Diabetes Association. (2019). Prediabetes. https://diabetes.org/diabetes/prediabetes
Centers for Disease Control and Prevention. (2019). Prediabetes: Your chance to prevent type 2 diabetes. https://www.cdc.gov/diabetes/basics/prediabetes.html
Centers for Disease Control and Prevention. (2018). National diabetes statistics report. https://www.cdc.gov/diabetes/data/statistics-report/index.html
Choi, S. B., Kim, W. J., Yoo, T. K., Park, J. S., Chung, J. W., Lee, Y., Kang, E. S., & Kim, D. W. (2014). Screening for prediabetes using machine learning models. Computational and Mathematical Methods in Medicine, 2014, Article 618976. https://doi.org/10.1155/2014/618976
Deberneh, H. M., & Kim, I. (2021). Prediction of type 2 diabetes based on machine learning algorithm. International Journal of Environmental Research and Public Health, 18(6), Article 3317. https://doi.org/10.3390/ijerph18063317
Erlin, Junadhi, Agustin, L., Fajri, N., & Sari, S. N. (2020). Early detection of diabetes using machine learning with logistic regression algorithm. Jurnal Nasional Teknik Elektro dan Teknologi Informasi. https://jurnal.ugm.ac.id/v3/JNTETI/article/view/3586/1647
Hasan, M. K., Alam, M. A., Das, D., Hossain, E., & Hasan, M. (2020). Diabetes prediction using ensembling of different machine learning classifiers. IEEE Access, 8, 76516-76531. https://doi.org/10.1109/ACCESS.2020.2989857
International Diabetes Federation. (2020). IDF diabetes atlas (10th ed.). https://diabetesatlas.org/
Islam, R. M., Islam, S. M. S., Ferrari, A. J., Rahman, M. M., & Rahman, M. A. (2021). Prevalence of diabetes and prediabetes among Bangladeshi adults and associated factors: Evidence from the demographic and health survey, 2017-18. medRxiv. https://doi.org/10.1101/2021.01.26.21250519
Mahboob Alam, T., Iqbal, M. A., Ali, Y., Wahab, A., Ijaz, S., Baig, T. I., Hussain, A., Malik, M. A., Raza, M. M., Ibrar, S., & Abbas, Z. (2019). A model for early prediction of diabetes. Informatics in Medicine Unlocked, 16, Article 100204. https://doi.org/10.1016/j.imu.2019.100204
Mayo Clinic Staff. (2018). Prediabetes: Symptoms and causes. Mayo Clinic. https://www.mayoclinic.org/diseases-conditions/prediabetes/symptoms-causes/syc-20355278
MedlinePlus. (2018). Prediabetes. National Library of Medicine. https://medlineplus.gov/prediabetes.html
Rashid, A. (2020). Diabetes dataset [Data set]. Mendeley Data. https://data.mendeley.com/datasets/wj9rwkp9c2/1
Tigga, N. P., & Garg, S. (2020). Prediction of type 2 diabetes using machine learning classification methods. Procedia Computer Science, 167, 706-716. https://doi.org/10.1016/j.procs.2020.03.336
Recommended articles
Fairness-Aware AI for Healthcare Supply Chain Equity: Integrating Social Determinants of Health, Causal Modeling, and Geospatial Optimization
Lung Cancer in Bangladesh: Epidemiological Burden, Tobacco and Environmental Risk Determinants, Diagnostic Delay, and Survival Challenges — A Systematic Narrative Evidence Synthesis
1
Save
Save
0
Citation
Citation
18
View
View
0
Share
Share