Can artificial intelligence bridge critical gaps in hypertension prediction and risk assessment in low- and...

According to the WHO, as of 2025, 1.4 billion adults between the ages 30-79 were living with hypertension, with approximately two-thirds of that number from low- and middle-income countries (LMICs). By 2025, the number of individuals with hypertension is expected to have reached 1.5 billion, with sub-Saharan Africa alone projected to have 74.7 million cases. In recent years, artificial intelligence (AI) has demonstrated the potential to improve the accuracy of risk assessment tools for predicting, managing and diagnosing diseases before they escalate. This paper aims to conduct a scoping review to identify current research, innovations and developments in the application of AI-based tools for hypertension prediction and risk assessment, specifically in LMICs.

This scoping review used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines to ensure proper representation and organisation of the literature. A search was conducted across multiple databases, including PubMed, Elsevier, Google Scholar, Scopus, IEEE Xplore and the ACM Digital Library. Two independent reviewers used the search terms in Appendix A to identify relevant literature. Initial scans were followed by a detailed evaluation based on well-structured and clear inclusion and exclusion criteria. A third, more experienced reviewer resolved any selection discrepancies flagged from the first two reviews.

The literature that would be included within this study will primarily be sourced from the following scientific literature repositories: PubMed, Scopus, Elsevier, Google Scholar, IEEE and ACM Digital Library.

The inclusion and exclusion criteria defined and used to guide the literature selection process within this review included the following:The literature should primarily be about the study of the deployment of AI algorithms or tools in the diagnosis, detection or risk assessment of hypertension in an LMIC.The data used in the literature to train and assess any implemented AI algorithms should be obtained from either an LMIC.The literature should have been published within the last 10 years, that is, 2014-2025.

The scoping review process followed the steps first outlined by Arksey and O'Malley in 2005. These steps are:Identifying the research question.Identifying the relevant studies within the literature.Selecting the studies within the relevant studies that most align with the research question.Charting the data that is present within these selected studies to get a deeper understanding of them.Collating, summarising and reporting the information found in the selected literature.From an initial result of 1371 papers, 5 were selected based on predefined inclusion and exclusion criteria. These studies solely focused on AI-based risk assessment and prediction of hypertension in LMICs.

The review identified significant research on AI applications for hypertension risk assessment and prediction in LMICs. Notable studies include an Ethiopian study on using machine learning algorithms, achieving an accuracy of 88.81%, precision of 89.62% and recall of 97.04% in hypertension risk prediction. Another research study involving hypertensive patients from South Asian countries (Bangladesh, Nepal, India) employed non-invasive information and machine learning tools for accurate hypertension prediction. The reported prediction accuracy of the studies found ranged from 78% to 97% depending on dataset size and AI model used. Algorithms discussed in these papers include logistic regression, decision trees and naïve Bayes.

The reviewed studies highlight the promising potential of AI to enhance hypertension prevention and management in LMICs. Successful implementation of AI tools requires the development of locally contextualised models and the availability of validated local data. AI can significantly impact hypertension outcomes in LMICs, provided these conditions are met.