Research Article
Predicting Hypertension Medication Uptake Using Explainable Artificial Intelligence: Evidence from a Kenyan Population-based Study
Issue:
Volume 11, Issue 2, June 2026
Pages:
40-59
Received:
6 May 2026
Accepted:
15 May 2026
Published:
2 June 2026
Abstract: Hypertension is a major contributor to cardiovascular morbidity and mortality worldwide, more so in Kenya, with limited progress towards achieving Africa's 2030 fast-track hypertension targets, especially in management. This study aimed to build a machine learning model to predict hypertension medication uptake in Kenya. Using data from 4,687 female and 5,269 male respondents from the 2022 Kenya Demographic and Health Survey, we applied Extreme Gradient Boosting, Support Vector Machine, Random Forest, and Elastic Net models. Data from 15 counties were split into training (80%) and testing (20%) sets, with class imbalance addressed using the Synthetic Minority Oversampling Technique and validation through leave-one-county-out cross-validation. The best-performing model, based on mean f1-score, was retrained using features selected through Sequential Forward Floating Selection. SHapley Additive exPlanations were used to interpret feature importance and directionality by sex. Treatment coverage remained suboptimal, with 26.6% of hypertensive males and 32.4% of females untreated. The XGBoost model achieved the best performance (78% males; 81% females). The most predictive features in both sexes were age, household size, sedentary time, income, exercise, wealth, residence duration, television viewership, and reproductive preferences among females. Interpretable machine learning revealed distinct sex-specific socio-behavioural predictors of hypertension treatment uptake in Kenya. Incorporating such data-driven insights can inform targeted, equitable interventions and strengthen hypertension control, especially in resource-limited settings where routine survey data can complement clinical assessments.
Abstract: Hypertension is a major contributor to cardiovascular morbidity and mortality worldwide, more so in Kenya, with limited progress towards achieving Africa's 2030 fast-track hypertension targets, especially in management. This study aimed to build a machine learning model to predict hypertension medication uptake in Kenya. Using data from 4,687 femal...
Show More
Research Article
Modelling Time-to-Failure in Presence of Competing Risks, the Sub-distribution Hazard Approach
Kipkirui Jephtah*
,
Tonui Benard Cheruiyot
Issue:
Volume 11, Issue 2, June 2026
Pages:
60-64
Received:
5 August 2025
Accepted:
27 November 2025
Published:
4 September 2026
DOI:
10.11648/j.bsi.20261102.12
Downloads:
Views:
Abstract: Background: Analysis of time-to-event data, generally called survival analysis, arise in many fields of study. Conventional methods of analyzing this type of data are historically well established and continue to be applied. The methods rely on the assumptions that subjects on follow-up can only fail from a well-defined single type of event, and that, conditioned on subject covariates, censoring time and the impending time-of-failure are independent. In real-world settings these assumptions are rarely fulfilled: subjects are typically exposed to multiple events that act concurrently to impede or modify the probability of failure from the event of interest. Events “compete” with each other, so the eventual failure of a subject can only be attributed to the first-occurring event – a competing risks scenario in which both the time to the first occurring event and the type of event that occurs at that time are of interest. Objective: To provide you the reader with a friendly and sufficient introduction to the theory underpinning the sub-distribution hazard model, based on the familiar or rather “traditional” description of survival data. This model describes the absolute risk of a subject experiencing an outcome while adjusting for right censoring, subject covariates, and naturally existing competing outcomes. Conclusions: In the competing risks framework, data may be modelled either through the cause-specific hazard model or the sub-distribution hazard model. Although both strategies are founded on the proportional hazards assumption, they differ in how risk sets are constituted and thus in what they measure, in their context of application, and in how the resulting hazard ratios are interpreted. For research questions where the sub-distribution hazard model is appropriate, the theoretical framework presented in this paper applies.
Abstract: Background: Analysis of time-to-event data, generally called survival analysis, arise in many fields of study. Conventional methods of analyzing this type of data are historically well established and continue to be applied. The methods rely on the assumptions that subjects on follow-up can only fail from a well-defined single type of event, and th...
Show More
