A Percentile Rank-Based Multi-Criteria Decision-Making Algorithm for Interpretable Hypertension Risk Stratification in Clinical Decision Support Systems

Hypertension risk prediction remains a critical challenge in data-driven healthcare systems, particularly in low- and middle-income countries (LMICs). In many of these settings, timely monitoring and diagnosis remain difficult. Machine learning (ML) methods have been applied for hypertension risk prediction; however, their complexity and “black box” nature limit their integration into clinical decision support systems. This study presents a Percentile Rank-Based Multi-Criteria Decision Making (PRBMCDM) model for hypertension monitoring and risk stratification. Hypertension-related data were collected from public and private hospitals in the Kogi Eastern Senatorial District of Nigeria. The data were used to develop and validate the proposed model. The proposed algorithm calculates risk scores based on the percentile ranking of multiple risk criteria of each patient. The model is designed as a lightweight and interpretable framework suitable for deployment in resource-constrained digital health environments. A drop column method was used to rank features. The model performance was evaluated using 5-fold cross-validation. PRBMCDM achieved an AUC of 0.75, 0.70 and 0.72 high, medium, and low-risk respectively. The model demonstrated high interpretability, making it well-suited for deployment in clinical settings. The PRBMCDM algorithm offers a transparent, simple, and clinically relevant framework for hypertension risk stratification model. Key limitations of existing ML models were addressed. Future work will focus on validating PRBMCDM against established risk scores on larger and more divers dataset to further strengthen its clinical applicability.

Hypertension, Decision Support Systems, Multi-Criteria Decision Making, Percentile Ranking

1. Unger T. et al. 2020 International Society of Hypertension global hypertension practice guidelines. Hypertension. 2020;75 (6): 1334-1357. https://doi.org/ 10.1161/HYPERTENSIONAHA.120.15026. [Google Scholar] [Crossref]

2. Chaturvedi A, Zhu A , Gadela N V , Prabhakaran D, and Jafar T H. Social determinants of health and disparities in hypertension and cardiovascular diseases. Hypertension. 2024; 81( 3): 387-399. https:// doi.org/ 10.1161/HYPERTENSIONAHA.123.21354. [Google Scholar] [Crossref]

3. Mills KT, Stefanescu A, and He J. The global epidemiology of hypertension. Nature Reviews Nephrology. 2020; 16(4): 223-237. https://doi.org/ 10.1038/s41581-019-0244-2. [Google Scholar] [Crossref]

4. Cheng Y B. et al. Risk stratification by cross-classification of central and brachial systolic blood pressure. Hypertension. 2022;79 (5): 1101-1111. https://doi.org/ 10.1161/HYPERTENSIONAHA.121.1 8773. [Google Scholar] [Crossref]

5. Boutilier JJ, Chan T C Y, Ranjan M, and Deo S. Risk stratification for early detection of diabetes and hypertension in resource-limited settings : machine learning analysis.Journal of Medical Internet Research. 2021; 23:20123. https://doi.org/ 10.2196/20123. [Google Scholar] [Crossref]

6. Forte C. et al. Development and validation of decision rules models to stratify coronary artery disease , diabetes , and hypertension risk in preventive care : cohort study of returning UK Biobank participants. Journal of personalized medicine. 2021;11(2); https://doi.org/ 10.3390/jpm11121322. [Google Scholar] [Crossref]

7. Rudin C, Chen C, Chen Z, Huang H, and Semenova L. Interpretable machine learning : fundamental principles and 10 grand challenges. Statistical Surveys. 2021; 16: 1-85. https://doi.org/10.1214/21-SS133 [Google Scholar] [Crossref]

8. Tonekaboni S, Joshi S, McCradden MD, and Goldenberg A. What clinicians want: contextualizing explainable machine learning for clinical end use. Proceeding of Machine Learning Research. 2019; 106:359-380. https://doi.org/ 10.48550/arXiv.1905.05134. [Google Scholar] [Crossref]

9. Ahmad M A, Eckert C, and Teredesai A. Interpretable machine learning in healthcare. IEEE Intelligent Informatics Bulleting. 2018; 9(1): 559-560. https://doi.org/ 10.1145/3233547.3233667. [Google Scholar] [Crossref]

10. Kolyshkina I, and Simoff S. Interpretability of machine learning solutions in public healthcare : The CRISP-ML Approach. Frontiers in Big Data. 2021; 4. https:// doi.org/ 10.3389/fdata.2021.660206. [Google Scholar] [Crossref]

11. Drnovšek R, Kapun MM, Šteblaj S, and Rajkovič U. Multicriteria risk evaluation model: utilizing fuzzy logic for improved transparency and quality of risk evaluation in healthcare. Risk Management. Healthcare Policy. 2025; 637-653. https:// doi.org/10.2147/RMHP.S490598. [Google Scholar] [Crossref]

12. Pollock BD, Ubl DS, Devkaran S, and Dowdy SC. Simulating the overall hospital quality star ratings with random measure weights. JAMA Network Open. 2025; 8 (7): 1-8. https:// doi.org/ 10.1001/jamanetworkopen.2025.19029. [Google Scholar] [Crossref]

13. Wu Y, Li S, Yuan J, Zhang H, Wang M. and Zhang Z. Benchmarking : a novel measuring tool for outcome comparisons in surgery. International Journal of Surgery. 2023;: 419-428. http://doi.org/10.1097/JS9.0000000000000212 [Google Scholar] [Crossref]

14. Parsons RE, Liu X, Collister JA, Clifton DA, Cairns BJ, and Clifton L. Independent external validation of the QRISK3 cardiovascular disease risk prediction model using UK Biobank. Heart. 2023; 109 (22); 1690-1697. https//doi.org/ 10.1136/heartjnl-2022-321231. [Google Scholar] [Crossref]

15. Van Trier TJ. et al. Validation of systematic coronary risk evaluation 2 ( SCORE2 ) and SCORE2-Older Persons in the EPIC-Norfolk prospective population cohort. European. Journal of Preventive Cardiology. 2024; 31(2): 182-189. https://doi.org/ 10.1093/eurjpc/zwad318. [Google Scholar] [Crossref]

• What the Desert Fathers Teach Data Scientists: Ancient Ascetic Principles for Ethical Machine-Learning Practice
• Comparative Analysis of Some Machine Learning Algorithms for the Classification of Ransomware
• Comparative Performance Analysis of Some Priority Queue Variants in Dijkstra’s Algorithm
• Transfer Learning in Detecting E-Assessment Malpractice from a Proctored Video Recordings.
• Dual-Modal Detection of Parkinson’s Disease: A Clinical Framework and Deep Learning Approach Using NeuroParkNet