Pilot Study of Hypertension Screening and Machine-Learning Prediction Using Community Outreach Data from Nkpokiti, Enugu, Nigeria Short Title: Machine Learning Prediction of Hypertension using Community Blood Pressure Data in Nigeria

Pilot Study of Hypertension Screening and Machine-Learning Prediction Using Community Outreach Data from Nkpokiti, Enugu, Nigeria Short Title: Machine Learning Prediction of Hypertension using Community Blood Pressure Data in Nigeria | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Pilot Study of Hypertension Screening and Machine-Learning Prediction Using Community Outreach Data from Nkpokiti, Enugu, Nigeria Short Title: Machine Learning Prediction of Hypertension using Community Blood Pressure Data in Nigeria Godswill Uzoechina, Winnifred Njideka Adiri, Osajiuba Treasure This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8088844/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background Hypertension is often undiagnosed in many low-income countries. While machine learning (ML) may enhance triage during community screening, there is, however, limited evidence from outreach programs across African settings. This pilot investigated the prevalence and feasibility of ML prediction using minimal data collected locally in a Nigerian outreach. Methods Cross-sectional analysis of anonymized data from a community outreach in Nkpokiti, Enugu, was performed. Eligible records (n = 115) included age, sex, and at least one paired systolic/diastolic blood-pressure (BP) measurement. Hypertension was also defined as mean SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg. Predictors were age, sex, first SBP, and pulse pressure. We trained penalized logistic regression (primary), random forest, and gradient-boosting models using nested 5-fold cross-validation for hyperparameter tuning; final illustrative results are reported on a stratified 80/20 hold-out. Discrimination (AUC), calibration (Brier score), and classification metrics were calculated with bootstrap confidence intervals. Results Median age was 33 years (mean 36.4, SD 14.4); 71.3% were female. The prevalence of hypertension was 25.2% (29/115), increasing with age from 12.7% (< 40y) to 52.0% (40–59y) and 54.5% (≥ 60y). SBP alone yielded an AUC of 0.865 (95% CI 0.777–0.941). On the hold-out set (n = 23; 6 positives), penalized logistic regression achieved an AUC of 0.941 (bootstrap mean 0.939; 95% CI 0.800–1.000), accuracy of 0.783, and Brier score of 0.094. Random forest: AUC 0.961, accuracy 0.826, Brier 0.088. A Gradient boosting method showed perfect discrimination on this small hold-out set (AUC 1.000) with a Brier score of 0.038, probably reflecting optimistic estimation due to small sample size. Conclusion In this pilot study, ML models comprising age, sex, and simple BP measures demonstrated excellent discrimination for hypertension, supporting the feasibility of context-specific, low-cost triage tools. These findings are exploratory; external validation with larger, representative African samples is needed before deployment. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16 Figure 17 Introduction High blood pressure is the single leading risk factor for cardiovascular and kidney disease worldwide and premature death, and its effective and early detection is a cornerstone of prevention ( 1 ). Globally, more than one billion adults have hypertension, and control rates remain low, especially in low- and middle-income countries with weak and fragile health systems ( 1 , 2 ). In Nigeria, the burden is significant and rising: national reviews report prevalence figures often within 20–40% with significant regional disparities (with rates higher in the southern geopolitical zones compared to the north) and consistent deficits in awareness, treatment, and control ( 3 , 4 ). Community-based screening and outreach activities are a cost-effective approach to increase detection and awareness, as demonstrated by large campaigns such as May Measurement Month that have screened millions and highlighted the scalability and public-health utility of opportunistic blood pressure (BP) screening ( 5 ). Yet outreach venues are frequently subjected to practical limitations, such as restricted time, scarcity of trained personnel, and dependence on single BP measurements, which heighten the risk of misclassification due to measurement variability, the white-coat phenomenon, or temporary influences (e.g., exercise, caffeine) ( 6 ). As a result, consensus guidelines and measurement sciences literature advocate for repeat measurements and computation of averaged values in order to enhance diagnostic accuracy, but repeat measurements are often not feasible during rapid, mass screening efforts ( 6 ). Machine-learning (ML) methodologies offer an alternative and complementary option for triage in resource-constrained screening by leveraging predictive signals from basic, routinely collected variables (age, sex, first BP reading) to direct individuals for confirmatory testing or referral. Systematic and empirical research demonstrates that ML can enhance the prediction of cardiovascular risk compared to conventional models; however, there is significant heterogeneity among studies in terms of outcome definitions, predictors utilized, and the reliability of validation methods. Calibration and transparency in reporting are often insufficient; constraints that should be overcome before clinical deployment and application ( 7 ). The recent TRIPOD + AI guidance stresses strong internal and external validation, transparent reporting of model development and calibration, along with sharing of code and model specification for reproducibility ( 8 ). There is a significant gap in evidence for feasibility studies assessing the performance of ML-based hypertension prediction, specifically in real-world outreach screening settings across sub-Saharan Africa, utilizing minimal or low-cost predictors. Such pilots in these settings are needed: (a) to estimate local prevalence under outreach conditions, (b) to measure misclassification induced by use of single as opposed to mean BP recordings, and (c) to determine whether ML techniques achieve useful and adequate discrimination and calibration under field conditions. Actual feasibility evidence will guide the development of larger, multicenter validation studies and the possible adoption of ML-enabled decision tools in the community screening workflow. Therefore, this pilot study is designed to ( 1 ) determine the prevalence of hypertension in a community-based outreach population in Nkpokiti, Enugu; ( 2 ) measure concordance and misclassification between single and mean BP measurements; and ( 3 ) construct and internally validate ML models (regularised logistic regression as primary model, with tree-based ensemble models: random forest and gradient boosting as comparators) for predicting hypertension using age, sex and mean BP reading through nested cross-validation, reporting discrimination, calibration, and decision-analytic metrics in accordance with TRIPOD + AI guidelines ( 8 ). Materials and Methodology Study design and setting This was a cross-sectional pilot study conducted using de-identified data obtained from a community hypertension screening in Nkpokiti, Enugu State, Nigeria, by the Nigerian Medical Students’ Association (NiMSA) and University of Nigeria Medical Students’ Association (UNMSA) Standing Committees on Policy Implementation (SCOPI), and Research Exchange (SCORE). Data collection took place on 17th May 2025. The data consisted of demographic information (age, sex) and paired systolic and diastolic blood pressure (BP) readings for each subject. No new data collection was conducted for this secondary analysis. Participants and eligibility All outreach attendees with a documented age, sex, and at least one paired systolic (SBP) and diastolic (DBP) BP measurement were included. Records were excluded from the study if age or sex were missing or if BP values were implausible (SBP 250 mmHg, DBP 150 mmHg) ( 9 ). Following quality control, the sample size was 115 participants. Outcome definition The outcome of interest was hypertension (systolic blood pressure [SBP] ≥ 140 mmHg and/or diastolic blood pressure [DBP] ≥ 90 mmHg) according to World Health Organization (WHO) cutoff points in the setting of community-based screening ( 1 ). When available, the mean of at least two readings was used for classification; otherwise, one valid reading was used. Data preparation and derived variables Data cleaning and transformation were performed with reproducible scripts in Python 3.9. Calculated variables were pulse pressure (PP = SBP – DBP, mmHg) and mean arterial pressure (MAP = DBP + [SBP – DBP]/3, mmHg) ( 10 ). For descriptive purposes, age groups were stratified as < 40, 40–59, and ≥ 60 years. Erroneous, duplicate, and implausible submissions were excluded. Analytic variables had no missingness. Descriptive statistical analysis Continuous variables were presented using mean ± standard deviation (SD) or medians (interquartile range [IQR]), and categorical variables as frequencies and percentages. Normality was tested using the Shapiro–Wilk test and visual inspection ( 11 ). Between-group comparisons were made with Student’s t test or Mann–Whitney U test as appropriate. Univariable logistic regression models calculated the unadjusted odds ratios (ORs) and 95% confidence intervals (CIs) for each covariate (age, sex, SBP, DBP, PP, MAP). The variance inflation factors (VIFs) were used to assess and detect multicollinearity among BP measures, and highly collinear predictors were removed from the multivariable models ( 12 ). Machine-learning framework The machine- learning (ML) analysis was exploratory to assess feasibility and internal validity, and not for definitive model development. The predictors were age, sex, SBP, and PP; these had been selected for parsimony and a low level of collinearity ( 13 ). The following three model classes were trained: Penalized logistic regression (L2 regularisation; inverse strength C = 0.01 – 10). Random forest classifier (n_estimators = 200; max_depth = 3 – None). Gradient boosting classifier (n_estimators = 100–200; learning_rate = 0.05–0.1; max_depth = 2–3). Each of the models was used with class_weight = ‘balanced’ to address a relatively moderate class imbalance (prevalence ≈ 25%). No standalone single tree classifier was used. Cross-validation and hyperparameter tuning A nested 5-fold stratified cross-validation (CV) was implemented to avoid overfitting ( 14 ). Each outer fold estimated generalisation performance, while inner folds tuned hyperparameters by grid search using the area under the receiver operating characteristic curve (AUC) as the selection metric. To avoid data leakage, all preprocessing procedures (scaling, encoding, and imputation) were performed within the inner-CV loops ( 15 ). Continuous variables were standardized (zero mean and unit variance) using the training fold statistics, and categorical variables were binary encoded. Model evaluation Performance was evaluated using discrimination, calibration, and classification indices. Discrimination was reported in terms of AUC and average precision ( 16 ). Calibration was analysed with the Brier score and reliability plots ( 17 ). Accuracy, sensitivity, specificity, precision, F1-score, and balanced accuracy were averaged across the outer folds. A bootstrap resampling with 2,000 iterations was then used to compute 95% CIs for the AUC and Brier scores ( 18 ). Model interpretability Transparency of the model was prioritised in concordance with TRIPOD-AI guidance ( 8 ). For logistic regression, the coefficients were presented as odds ratios (ORs) over 10 mmHg or per standard-deviation increase with bootstrap 95% CIs. For the ensemble models, feature importance scores were averaged across outer folds, and the top predictors (age, SBP) were plotted using partial dependence plots (PDP) for visualising marginal effects. SHAP (SHapley Additive exPlanations) values were calculated to assess local and global feature contributions for the tree-based models ( 19 ). Sensitivity analyses We repeated analyses (a) replacing SBP with MAP; (b) excluding derived variables (PP, MAP) to assess the sensitivity of collinearity; (c) employing a different hypertension threshold (≥ 130/80 mmHg); and (d) using different random seeds and CV folds (5-fold vs 10-fold) to confirm the stability of the metrics. Statistical software and reproducibility All analyses were conducted using Python 3.9 with pandas, numpy, scikit-learn, scipy, statsmodels, matplotlib, and shap packages ( 20 ). Random seeds were set to ensure reproducibility. The entire analysis pipeline, including code and environment files, is deposited in a public GitHub repository. Anonymized data may be made available on reasonable request after approval from an ethics committee. Ethical approval and consent The consents for participation in the outreach and for the use of data in research were verbal and written. Permission for secondary analysis was obtained from the Ethics Committee of the University of Nigeria Teaching Hospital (NHREC/05/01/2008B-FW00002458-1RB00002327). All approaches conformed to the Declaration of Helsinki and the Nigerian National Code of Health Research Ethics ( 21 ). Reporting standards The study was conducted in accordance with the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis with Artificial Intelligence (TRIPOD-AI) guidance ( 8 ) to facilitate transparent methods, reproducibility, and explainability. Results Sample and data quality One hundred and fifteen (115) participant records met the inclusion criteria and were included in the analysis. Age, sex, systolic blood pressure (SBP), and diastolic blood pressure (DBP) were available for all records; there were no missing values across the analytic variables, and no duplicate records were found. There were no implausible BP values detected after pre-specified quality control procedures (no SBP < 70 or ≥ 250 mmHg; no DBP < 40 or ≥ 150 mmHg). Participant characteristics The mean age of participants was 36.4 years (SD 14.4; median 33; range 18–84). N = 33 were male (28.7%) and n = 82 (71.3%) female. The summary BP measures were: mean SBP 124.5 mmHg (SD 11.7; median 126; range 100–163), mean DBP 80.7 mmHg (SD 9.9; median 80; range 60–116), mean pulse pressure 43.8 mmHg (SD 9.8; median 44; range 20–80), and mean arterial pressure 95.3 mmHg (SD 9.5). Table 1 presents the baseline demographic and blood pressure characteristics of the study participants. Table 1 Participant characteristics Characteristic Overall (N = 115) Age, mean (SD) 36.38 (14.39) Age, median (IQR) 33 (—) Age, range 18–84 Sex — female, n (%) 82 (71.3%) Sex — male, n (%) 33 (28.7%) SBP, mean (SD) (mmHg) 124.50 (11.72) SBP, median (mmHg) 126 SBP, range (mmHg) 100–163 DBP, mean (SD) (mmHg) 80.73 (9.92) DBP, median (mmHg) 80 DBP, range (mmHg) 60–116 Pulse pressure, mean (SD) (mmHg) 43.77 (9.80) MAP, mean (SD) (mmHg) 95.32 (9.49) Hypertension (WHO definition), n (%) 29 (25.2%) Participant age distribution is shown in Fig. 1. Figure 1. Age distribution of participants Histogram showing the distribution of participant ages (n = 115). The distribution of systolic blood pressure (SBP) measurements is presented in Fig. 2, while diastolic blood pressure (DBP) is shown in Fig. 3. Figure 2. Distribution of systolic blood pressure (SBP) Histogram of systolic blood pressure measurements (mmHg) among all participants. Figure 3. Distribution of diastolic blood pressure (DBP) Histogram of diastolic blood pressure measurements (mmHg) in the study sample. Hypertension prevalence Application of the predefined criteria for SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg (WHO screening definition) resulted in 29 of 115 participants being classified as having high blood pressure, with a total prevalence of 25.2% (29/115) ( 2 ). By sex, prevalence was 21/82 = 25.6% for females and 8/33 = 24.2% for males. Table 2 shows the distribution of hypertension across age and sex groups, including normality and group comparison tests. Table 2 Hypertension prevalence by age and sex Stratum N Hypertensive n (%) Age < 40 years 79 10 (12.7%) Age 40–59 years 25 13 (52.0%) Age ≥ 60 years 11 6 (54.5%) Female (overall) 82 21 (25.6%) Male (overall) 33 8 (24.2%) Overall 115 29 (25.2%) A positive relationship between age and SBP is illustrated in Fig. 4. Figure 4. Relationship between age and systolic blood pressure Scatterplot showing a positive trend between participant age and SBP with linear fit. Figure 5 compares SBP levels between male and female participants, showing no significant difference (p = 0.8606). Figure 5. Systolic blood pressure by sex Boxplots comparing SBP distributions between male and female participants; no significant difference observed (p = 0.8606). Distributional checks and group comparisons Shapiro–Wilk tests highlighted a digression from normality in some variables (Age: W = 0.871, p < 0.001; SBP: W = 0.977, p = 0.044; DBP: W = 0.950, p = 0.0003; MAP: W = 0.976, p = 0.0374), but there was no strong evidence against normality in pulse pressure (W = 0.983, p = 0.142). There was no evidence of a sex-based difference in SBP across the sample (two-sample test: t = 0.176, p = 0.861), and there was no significant association between sex and hypertensive status (χ² = 0.000, p = 1.000) via contingency testing, as SBP normality by sex: male SBP Shapiro p = 0.371; female SBP Shapiro p = 0.073. A Kruskal–Wallis test showed that there is a reliable difference in SBP across the predetermined age groups (p < 0.001), in accordance with a stepwise increase in prevalence with age. Correlations among age, SBP, DBP, pulse pressure (PP), and mean arterial pressure (MAP) are visualized in Fig. 6. Figure 6. Correlation matrix for age and blood pressure variables Annotated heatmap illustration for Age, SBP, DBP, Pulse Pressure (PP), and Mean Arterial Pressure (MAP). Univariable associations with hypertension SBP, DBP, and MAP demonstrated strong univariable associations with the hypertension label; sex did not. Because multiple BP formulations are deterministically related, these univariable findings were interpreted cautiously and steered collinearity checks. Table 3 summarizes the univariable associations between candidate predictors and hypertension status. Table 3 Univariable logistic regression result Predictor OR 95% CI p-value Age (per year) 1.056 1.024–1.090 0.0005 Sex (male vs female) 0.930 0.364–2.375 0.879 SBP (per mmHg) 1.208 1.116–1.307 < 0.00001 DBP (per mmHg) 1.681 1.314–2.151 0.000035 Pulse pressure 0.988 0.946–1.032 0.595 MAP 1.733 1.351–2.223 0.000015 Multicollinearity diagnostics The Variance inflation factor (VIF) analysis on Age, SBP, DBP, PP, and MAP showed very large or infinite VIFs for SBP/DBP/PP/MAP (highlighting deterministic collinearity) and a small VIF for Age (~ 1.30). Furthermore, multivariable models were limited to a few parsimonious variables (SBP kept as the principal BP predictor; PP investigated separately). Single-predictor discrimination (AUC with bootstrap CIs) Bootstrap AUCs (2000 resamples) for individual predictors were: SBP: AUC = 0.865; 95% CI = (0.777, 0.941). Age: AUC = 0.738; 95% CI = (0.615, 0.852). Pulse pressure: AUC = 0.455; 95% CI = (0.325, 0.588). SBP alone showed significant discrimination for the binary hypertension label, age showed modest discrimination, and pulse pressure alone performed weakly. Class balance and train/test split The overall class distribution (total) was [86 29] (negative, positive). We adopted a stratified 80/20 split for training and hold-out testing, which yielded: Training set: n = 92 samples with 23 hypertensives (shape reported as (92, 4) where 4 is the number of predictors/features used). Test set (hold-out): n = 23 samples with 6 hypertensives (shape ( 23 , 4 )). These numbers were used for the illustrative hold-out performance reported below; AUC optimization and nested CV (inner 5-fold) were used for hyperparameter tuning during development. Machine-learning model training, best hyperparameters, and hold-out performance Models evaluated were penalised logistic regression (primary), random forest (RF), and gradient boosting (GB). The perfect AUC for the GB model on the small hold-out (n = 23; 6 hypertensives) possibly demonstrates optimistic estimation bias due to the limited sample; bootstrap CIs of exactly 1.000 are an artefact of this small resampling distribution and should be interpreted cautiously. Model coefficients and feature importances were consistent: SBP had the greatest predictive importance in all models. Example LR coefficients (standardised predictors): SBP coef = 3.062 (OR 21.37 per SD), Age coef = − 0.046 (OR 0.955 per SD), Sex coef = − 0.044 (OR 0.957 per SD), PP coef = − 1.504 (OR 0.222 per SD). RF and GB feature-importance rankings assigned the biggest weight to systolic pressure (RF importance 0.554; GB importance 0.727). Table 4 reports the predictive performance metrics of the logistic regression, random forest, and gradient boosting models. Table 4 Model performance summary Model Best params (selected) Test AUC AUC_boot_mean 95% CI (AUC) Accuracy Sensitivity Specificity Brier Logistic regression C = 1, penalty = L2 0.941 0.939 0.800–1.000 0.783 0.667 0.824 0.094 Random forest n_estimators = 200, max_depth = None, min_samples_leaf = 1 0.961 0.960 0.857–1.000 0.826 0.500 0.941 0.088 Gradient boosting n_estimators = 200, lr = 0.1, max_depth = 3 1.000 1.000 1.000–1.000 0.957 1.000 0.941 0.038 Table 5 presents the comparative performance of the three machine learning models on the hold-out test set, showing that all achieved high discrimination, with gradient boosting performing best overall. Table 5 Model comparison summary (hold-out test set) model AUC AUC_boot_mean AUC_CI_low AUC_CI_high Accuracy Sensitivity Specificity LogisticRegression 0.941 0.939 0.800 1.000 0.783 0.667 0.824 RandomForest 0.961 0.960 0.857 1.000 0.826 0.500 0.941 GradientBoosting 1.000 1.000 1.000 1.000 0.957 1.000 0.941 The discriminative performance of the logistic regression, random forest, and gradient boosting models is shown by the ROC curves in Fig. 7. Figure 7. ROC curves for logistic regression, random forest, and gradient boosting models Receiver-operating characteristic curves comparing model discrimination on the hold-out test set. The robustness of the logistic regression model is further highlighted by the bootstrap ROC curve with a 95% confidence interval in Fig. 8. Figure 8. Bootstrap ROC for logistic regression with 95% confidence interval Mean ROC curve from 2000 bootstrap resamples showing logistic regression performance variability. Figure 9 displays the precision–recall curves, highlighting model performance under class-imbalance conditions. Figure 9. Precision–recall curves for all models Precision–recall plots illustrating model performance under class imbalance conditions. Calibration and probabilistic accuracy Calibration was evaluated using reliability plots and the Brier scores. Brier scores on the test set were LR = 0.094, RF = 0.088, GB = 0.038. Reliability diagrams indicated that GB demonstrated the best apparent calibration on the small test set, while LR and RF showed slight miscalibration in some probability bins. Due to the sample size, these assessments of calibration are descriptive and necessitate confirmation and definitiveness in larger samples. Calibration plots comparing predicted probabilities with observed outcomes for each model are provided in Fig. 10. Figure 10. Calibration plots for logistic regression, random forest, and gradient boosting Reliability diagrams comparing predicted probabilities with observed outcomes for each model. Model confusion matrices at the 0.5 probability threshold are shown in Figs. 11–13. Figure 11–13. Confusion-matrix heatmaps at threshold 0.5 Heatmaps showing counts of true/false positives and negatives for each model on the test set. Explainability and diagnostics ROC curves for all models also demonstrated good discrimination on the hold-out dataset, with higher discrimination for GB and RF when compared to LR. Partial dependence plots (GB) demonstrated a monotonic increase in predicted probability with rising SBP and a smaller positive effect for age. SHAP summaries for tree ensembles also confirmed SBP as the primary determinant for predictive analysis. Feature importance scores for the random forest and gradient boosting models, along with logistic regression coefficients (odds ratios), are summarized in Figs. 14–16. Figure 14–16. Feature importance and logistic regression coefficients Bar charts displaying feature importances (RF/GB) and corresponding odds ratios (LR), highlighting SBP as the top predictor. Sensitivity analyses Results were robust to alternate choices of predictors (MAP in place of SBP) and to the removal of derived variables (PP, MAP); SBP remained the strongest predictor, and performance estimates were directionally consistent across sensitivity checks. Variation of CV seeds and number of folds yielded stable qualitative results, but the numerical metrics were heterogeneous within bootstrap CI bounds. Table 6 displays model coefficients and feature importance values, highlighting systolic blood pressure as the dominant predictor. Table 6 Model interpretability: coefficients and feature importances Feature LR coef (std) LR OR RF importance GB importance Age −0.046 0.955 0.199 0.006 Sex (bin) −0.044 0.957 0.034 0.001 Systolic pressure 3.062 21.368 0.554 0.727 Pulse pressure −1.504 0.222 0.213 0.266 Partial dependence plots in Fig. 17 illustrate the marginal effects of age and SBP on predicted hypertension probability. Figure 17. Partial dependence plots for key predictors in the gradient boosting model Plots showing the marginal effect of SBP and age on predicted hypertension probability. Discussion This pilot study highlights the feasibility of applying machine-learning (ML) methods to predict hypertension and map its distribution using a few input variables, i.e., age and sex, as well as office blood-pressure (BP) readings, in a community-based population. A hypertension prevalence of 25.2% was observed, which is very consistent with the national estimates and that of sub-Saharan Africa, where adult hypertension prevalence of 23–31% has been widely reported ( 3 , 22 ). Prevalence increased sharply with advancing age, as expected from previous epidemiologic studies revealing that vascular stiffness and decreased arterial compliance mediate the age-associated rise in systolic pressure and hypertension risk ( 23 , 24 ). Predictors and model performance SBP was the single most important variable in predicting hypertension in all models among all the variables examined. SBP alone achieved an AUC of 0.865, highlighting considerable discriminatory power for this parameter in isolation. This concurs with the global literature that SBP is the single most reliable predictor of cardiovascular risk, especially in low-resource settings where diastolic and biochemical covariates are not readily obtainable ( 25 – 27 ). The three machine-learning algorithms: penalised logistic regression (PLR), random forest (RF), and gradient boosting (GB), achieved high discrimination (AUC 0.941–1.000) and good calibration (Brier scores ≤ 0.094). Gradient boosting obtained a perfect discrimination on a small hold-out set, but this is likely due to overfitting given the small sample size. Previous large-scale ML studies have consistently suggested that ensemble models, particularly gradient boosting, provide better performance than logistic regression for hypertension and cardiovascular prediction tasks, but the magnitude of their performance advantage is considerably narrowed when the feature sets are small or strongly correlated ( 28 – 30 ). Model interpretability and explainability Explainability analyses, including feature importance and partial dependence plots, consistently identified SBP as the single most important predictor, followed by pulse pressure and age as the next two most important features. These patterns are biologically and clinically plausible, as increased SBP and widened pulse pressure are markers of arterial stiffness, which is a key pathophysiologic determinant of hypertension risk ( 31 , 32 ). The consistency of feature importance across models indicates the robustness of SBP as a primary signal detecting ML-based hypertension detection in community data. SHAP summaries also confirmed these patterns, providing the transparency necessary for responsible ML implementation in health research ( 19 ). Comparison with related studies Recent ML-based hypertension prediction studies from sub-Saharan Africa have demonstrated similar findings, despite methodological heterogeneity. For example, Kimmel et al. (2023) conducted a cross-sectional machine-learning study in Ethiopia and reported an XGBoost model with an AUC of 0.894, an accuracy of 88.81%, a precision of 89.62%, a recall of 97.04%, and an F1-score of 93.18%, with age, weight/BMI, and related features among the top predictors ( 33 ). In China, Ji et al. (2022) employed a tree-based ensemble on adult sample data and identified age and SBP as the best predictors, with an XGBoost AUC of 0.893 ( 34 ). These correspondences emphasize the potential of parsimonious, interpretable ML models to improve hypertension screening in resource-limited settings, particularly when coupled with mobile or community health platforms ( 35 ). Strengths and limitations Strengths of this study are rigorous data cleaning, a reproducible ML workflow with nested cross-validation, and the use of bootstrap confidence intervals to express model uncertainty. The transparent inclusion of calibration and discrimination indices aligns with the emerging reporting standards, such as TRIPOD-AI ( 8 ). A substantial contribution of this work is that it bases the predictive models solely on locally sourced data from Nigeria, which addresses a substantial gap in the global ML health landscape. The majority of hypertension prediction models to date have been developed using high-income country datasets, with very limited representations of African populations in training and validation pipelines ( 36 – 38 ). These geographic and demographic biases limit the generalizability of predictive models and risk exacerbating algorithmic inequities when applied in low- and middle-income countries. This study also reinforces the evidence for context-specific AI development, through training, testing, and validating models on local data, decolonising digital health research through inclusive data practices and equitable model design ( 39 – 41 ). Potential overfitting, especially in the gradient boosting model, cannot be ruled out, and external validation on a larger independent dataset is necessary. Moreover, anthropometric and behavioural covariates (e.g., BMI, smoking, and alcohol use) were unavailable in the dataset, which might have enhanced model generalizability ( 42 ). Hence, given the small sample size and limited number of hypertensive cases, results should be interpreted as exploratory and hypothesis-generating. Implications and future directions This study demonstrates the feasibility of ML-enabled community hypertension surveillance in Nigeria. Despite the small sample size, ML-driven models demonstrated strong predictive accuracy, indicating that properly designed models relying on cost-effective variables can be employed to support large-scale hypertension risk mapping. Future efforts should focus on increasing sample size, considering lifestyle and genetic predictors, and applying external validation. Integration with digital health platforms or mobile screening applications could potentially expand and facilitate early detection and inform community-based interventions ( 43 , 44 ). Conclusion ML methods, especially logistic regression and tree-based ensembles, achieved high discriminatory performance in predicting hypertension using simple demographic and blood-pressure data. Systolic blood pressure (SBP) was the most influential predictor of model outcomes. These results highlight the potential feasibility of data-driven hypertension monitoring in resource-poor environments and provide the basis for broader, external validation studies to guide public health policy-making in Nigeria and throughout sub-Saharan Africa. Abbreviations Abbreviation Full form AUC Area under the receiver-operating characteristic curve AP Average precision (area under the precision–recall curve) BMI Body mass index BP Blood pressure CI Confidence interval CV Cross-validation DBP Diastolic blood pressure EPV Events per variable GB Gradient boosting (ensemble machine-learning method) GBM Gradient boosting machine (synonym of GB) GLM Generalised linear model HBPM Home blood pressure monitoring IQR Interquartile range LOESS Locally estimated scatterplot smoothing LR Logistic regression (penalised logistic regression when specified) MAP Mean arterial pressure (≈ DBP + (SBP − DBP)/3) MBP Mean blood pressure (synonym for MAP in some texts) ML Machine learning NPV Negative predictive value NHREC National Health Research Ethics Committee (Nigeria) NiMSA Nigerian Medical Students’ Association OR Odds ratio PDP Partial dependence plot PP Pulse pressure (SBP − DBP) PPV Positive predictive value PR Precision–recall (curve) RF Random forest (ensemble tree-based model) SBP Systolic blood pressure SD Standard deviation SHAP SHapley Additive exPlanations (model-agnostic explainability method) TRIPOD Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis TRIPOD-AI TRIPOD extension and guidance for prediction models using artificial intelligence / machine learning WHO World Health Organization Declarations Ethics approval and consent to participate Ethical approval for this study was obtained from the Health Research Ethics Committee of the University of Nigeria Teaching Hospital (UNTH-HREC) (Protocol Number: NHREC/05/01/2008B-FW00002458-1RB00002327). All study procedures adhered to the ethical principles outlined in the Declaration of Helsinki (2013 revision). Informed consent to participate was obtained from all individuals before data collection during the community outreach. Participation was voluntary, and all data were anonymized before analysis. Consent for publication Not applicable. All participants provided consent for the use of anonymized data for research and publication purposes. Competing interests All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria, educational grants, participation in speakers’ bureaus, membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript. Availability of data and materials The de-identified data that was used in this study is not available to the public due to community-level privacy considerations and limitations associated with the outreach program. The data may be made available by the corresponding author on reasonable request and with permission from the organizers of the Nkpokiti community outreach. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Author Contributions Godswill U (GU): Conceptualization; Methodology; Data curation; Formal analysis; Software; Visualization; Investigation; Writing – original draft (Methods, Results, Discussion); Writing – review & editing. Treasure O (TO): Conceptualization; Formal analysis; Writing – original draft (Introduction, Abstract, Conclusion); Writing – review & editing. Winnifred A (WNA): Supervision; Project administration; Validation; Writing – review & editing. Acknowledgments The authors sincerely thank the health volunteers, especially the Nigerian Medical Students’ Association (NiMSA) and University of Nigeria Medical Students’ Association (UNMSA) Standing Committees on Policy Implementation (SCOPI), and Research Exchange (SCORE), for their participation and cooperation during data collection. We also appreciate the support of the outreach organizers and data entry team for their assistance in field coordination. References World Health Organization. Guideline for the pharmacological treatment of hypertension in adults. Geneva: WHO. 2021. Available from: https://www.who.int/publications/i/item/9789240033986 Forouzanfar MH, Liu P, Roth GA, Ng M, Biryukov S, Marczak L et al. Global Burden of Hypertension and Systolic Blood Pressure of at Least 110 to 115 mm Hg, 1990–2015. JAMA. 2017;317(2):165–182. 10.1001/jama.2016.19043 . Erratum in: JAMA. 2017;317(6):648. doi: 10.1001/jama.2017.0013. PMID: 28097354. Adeloye D, Basquill C. Estimating the prevalence and awareness rates of hypertension in Africa: a systematic analysis. PLoS ONE. 2014;9(8):e104300. 10.1371/journal.pone.0104300 . PMID: 25090232; PMCID: PMC4121276. Akinlua JT, Meakin R, Umar AM, Freemantle N. Current Prevalence Pattern of Hypertension in Nigeria: A Systematic Review. PLoS ONE. 2015;10(10):e0140021. 10.1371/journal.pone.0140021 . PMID: 26461923; PMCID: PMC4603956. Beaney T, Burrell LM, Castillo RR, Charchar FJ, Cro S, Damasceno A et al. MMM Investigators. May Measurement Month 2018: a pragmatic global screening campaign to raise awareness of blood pressure by the International Society of Hypertension. Eur Heart J. 2019;40(25):2006–2017. 10.1093/eurheartj/ehz300 . Erratum in: Eur Heart J. 2019;40(37):3109. doi: 10.1093/eurheartj/ehz373. PMID: 31041440; PMCID: PMC6600128. Muntner P, Shimbo D, Carey RM, Charleston JB, Gaillard T, Misra S, et al. Measurement of Blood Pressure in Humans: A Scientific Statement From the American Heart Association. Hypertension. 2019;73(5):e35–66. 10.1161/HYP.0000000000000087 . PMID: 30827125; PMCID: PMC11409525. Weng SF, Reps J, Kai J, Garibaldi JM, Qureshi N. Can machine-learning improve cardiovascular risk prediction using routine clinical data? PLoS ONE. 2017;12(4):e0174944. 10.1371/journal.pone.0174944 . PMID: 28376093; PMCID: PMC5380334. Collins GS, Moons KGM, Dhiman P, Riley RD, Beam AL, Van Calster B et al. TRIPOD + AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385:e078378. doi: 10.1136/bmj-2023-078378. Erratum in: BMJ. 2024;385:q902. 10.1136/bmj.q902 . PMID: 38626948; PMCID: PMC11019967. Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves JW, Hill MN, Council on High Blood Pressure Research Professional and Public Education Subcommittee, American Heart Association, et al. Recommendations for blood pressure measurement in humans: an AHA scientific statement from the Council on High Blood Pressure Research Professional and Public Education Subcommittee. J Clin Hypertens (Greenwich). 2005;7(2):102–9. 10.1111/j.1524-6175.2005.04377.x . PMID: 15722655; PMCID: PMC8109470. Guyton AC, Hall JE. Textbook of Medical Physiology. 14th ed. Philadelphia: Elsevier; 2021. Shapiro SS, Wilk MB. An Analysis of Variance Test for Normality (Complete Samples). Biometrika. 1965;52(3/4):591–611. Available from: https://www.jstor.org/stable/2333709 O’Brien RM. A caution regarding rules of thumb for variance inflation factors. Qual Quant. 2007;41(5):673–90. Kavakiotis I, Tsave O, Salifoglou A, Maglaveras N, Vlahavas I, Chouvarda I. Machine Learning and Data Mining Methods in Diabetes Research. Comput Struct Biotechnol J. 2017;15:104–16. 10.1016/j.csbj.2016.12.005 . PMID: 28138367; PMCID: PMC5257026. Varma S, Simon R. Bias in error estimation when using cross-validation for model selection. BMC Bioinformatics. 2006;7:91. 10.1186/1471-2105-7-91 . PMID: 16504092; PMCID: PMC1397873. Cawley GC, Talbot NL. On over-fitting in model selection and subsequent selection bias in performance evaluation. J Mach Learn Res. 2010;11:2079–107. NCD Risk Factor Collaboration (NCD-RisC). Worldwide trends in blood pressure from 1975 to 2015: a pooled analysis of 1479 population-based measurement studies with 19·1 million participants. Lancet. 2017;389(10064):37–55. doi: 10.1016/S0140-6736(16)31919-5. Epub 2016 Nov 16. Erratum in: Lancet. 2020;396(10255):886. 10.1016/S0140-6736(20)31972-3 . PMID: 27863813; PMCID: PMC5220163. Brier GW. Verification of forecasts expressed in terms of probability. Mon Weather Rev. 1950;78(1):1–3. Efron B, Tibshirani RJ. An Introduction to the Bootstrap. Chapman and Hall/CRC; 1994. Lundberg SM, Lee SI. A Unified Approach to Interpreting Model Predictions. Vol. 30, Neural Information Processing Systems. Curran Associates, Inc.; 2017. Available from: https://papers.nips.cc/paper_files/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12(85):2825–30. Available from: https://www.jmlr.org/papers/v12/pedregosa11a.html National Health Research Ethics Committee of Nigeria (NHREC). National Code for Health Research Ethics. Abuja: NHREC; 2007. Ojji DB, Baldridge AS, Orji IA, Shedul GL, Ojo TM, Ye J, Chopra A, et al. Hypertension Treatment in Nigeria Program Investigators. Characteristics, treatment, and control of hypertension in public primary healthcare centers in Nigeria: baseline results from the Hypertension Treatment in Nigeria Program. J Hypertens. 2022;40(5):888–96. Epub 2022 Jan 15. PMID: 35034080; PMCID: PMC9081131. Franklin SS, Gustin W 4th, Wong ND, Larson MG, Weber MA, Kannel WB, Levy D. Hemodynamic patterns of age-related changes in blood pressure. The Framingham Heart Study. Circulation. 1997;96(1):308 – 15. 10.1161/01.cir.96.1.308 . PMID: 9236450. Whelton PK, Carey RM, Aronow WS, Casey DE Jr, Collins KJ, Dennison Himmelfarb C, College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. 2017 ACC/AHA/AAPA/ABC /ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: Executive Summary: A Report of the American. Hypertension. 2018;71(6):1269–1324. doi: 10.1161/HYP.0000000000000066. Epub 2017 Nov 13. Erratum in: Hypertension. 2018;71(6):e136-e139. doi: 10.1161/HYP.0000000000000075. Erratum in: Hypertension. 2018;72(3):e33. 10.1161/HYP.0000000000000080 . PMID: 29133354. NCD Risk Factor Collaboration (NCD-RisC). Long-term and recent trends in hypertension awareness, treatment, and control in 12 high-income countries: an analysis of 123 nationally representative surveys. Lancet. 2019;394(10199):639–51. 10.1016/S0140-6736(19)31145-6 . Epub 2019 Jul 18. PMID: 31327564; PMCID: PMC6717084. Liu T, Krentz A, Lu L, Curcin V. Machine learning based prediction models for cardiovascular disease risk using electronic health records data: systematic review and meta-analysis. Eur Heart J Digit Health. 2024;6(1):7–22. 10.1093/ehjdh/ztae080 . PMID: 39846062; PMCID: PMC11750195. Islam SMS, Talukder A, Awal MA, Siddiqui MMU, Ahamad MM, Ahammed B et al. Machine Learning Approaches for Predicting Hypertension and Its Associated Factors Using Population-Level Data From Three South Asian Countries. Front Cardiovasc Med. 2022;9. Silva GFS, Fagundes TP, Teixeira BC, Chiavegatto Filho ADP. Machine Learning for Hypertension Prediction: a Systematic Review. Curr Hypertens Rep. 2022;24(11):523–33. 10.1007/s11906-022-01212-6 . Epub 2022 Jun 22. PMID: 35731335. Adeleke O, Adebayo S, Halleluyah Aworinde, Adeleke O, Adeniyi AE. Oluwasegun Julius Aroba. Machine learning evaluation of a hypertension screening program in a university workforce over five years. Sci Rep. 2024;14(1). Waburi E, Muriithi D, Eugine Sundays. Integrating Explainable Machine Learning Models for Early Detection of Hypertension: A Transparent Approach to AI-Driven Healthcare. Am J Artif Intell. 2025;9(2):154–66. Safar ME, Plante GE, Mimran A. Arterial stiffness, pulse pressure, and the kidney. Am J Hypertens. 2015;28(5):561–9. 10.1093/ajh/hpu206 . Epub 2014 Dec 4. PMID: 25480804. Laurent S, Boutouyrie P. The structural factor of hypertension: large and small artery alterations. Circ Res. 2015;116(6):1007-21. 10.1161/CIRCRESAHA.116.303596 . PMID: 25767286. Kimeli N, Nkurunziza J, Juma K. Predicting health insurance uptake in Kenya using Random Forest: An analysis of socio-economic and demographic factors. PLoS ONE. 2023;18(11):e0294166–6. Ji W, Zhang Y, Cheng Y, Wang Y, Zhou Y. Development and validation of prediction models for hypertension risks: A cross-sectional study based on 4,287,407 participants. Front Cardiovasc Med. 2022;9:928948. PMID: 36225955; PMCID: PMC9548597. Segun-Omosehin OA, Omiye JA, Elalfy A, Moideen-Sheriff S, Kuti F, Omole O et al. Evaluating the digital health technology landscape in sub-Saharan Africa and its implications for cardiovascular health. npj Cardiovascular Health. 2025;2(1). Available from: https://www.nature.com/articles/s44325-025-00055-9 Wiens J, Saria S, Sendak M, Ghassemi M, Liu VX, Doshi-Velez F et al. Do no harm: a roadmap for responsible machine learning for health care. Nat Med. 2019;25(9):1337–1340. 10.1038/s41591-019-0548-6 . Epub 2019 Aug 19. Erratum in: Nat Med. 2019;25(10):1627. doi: 10.1038/s41591-019-0609-x. PMID: 31427808. Joseph J. Algorithmic bias in public health AI: a silent threat to equity in low-resource settings. Front Public Health. 2025;13:1643180. 10.3389/fpubh.2025.1643180 . PMID: 40771228; PMCID: PMC12325396. Owoyemi A, Owoyemi J, Osiyemi A, Boyd A. Artificial Intelligence for Healthcare in Africa. Front Digit Health. 2020;2:6. 10.3389/fdgth.2020.00006 . PMID: 34713019; PMCID: PMC8521850. Mugalula Kalule Grancia. Decolonizing AI ethics in Africa’s healthcare: An ethical perspective. AI and Ethics. 2024. Nyamawe AS. Is the public sector Africa’s hidden force for AI-driven healthcare transformation? Telematics and Informatics Reports. 2025;20:100258. Available from: https://www.sciencedirect.com/science/article/pii/S2772503025000726 Skalidis I, Maurizi N, Salihu A, Fournier S, Cook S, Iglesias JF, et al. Artificial Intelligence and Advanced Digital Health for Hypertension: Evolving Tools for Precision Cardiovascular Care. Med (Kaunas). 2025;61(9):1597. 10.3390/medicina61091597 . PMID: 41010987; PMCID: PMC12471829. Ren W, Fan K, Liu Z, Wu Y, An H, Liu H. Overcoming Missing Data: Accurately Predicting Cardiovascular Risk in Type 2 Diabetes, A Systematic Review. Journal of Diabetes. 2025 Jan [cited 2025 Jan 26];17(1). Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11753920/ Xiang Y, Li S, Zhang P. An exploration in remote blood pressure management: Application of daily routine pattern based on mobile data in health management. Fundamental Res. 2022;2(1):154–65. Gafane-Matemane LF, Mokwatsi GG, Boateng D. Hypertension management in sub-Saharan Africa: an overview of challenges and opportunities for telemedicine. Connected Health. 2023;2(1):9–22. Available from: https://f.oaes.cc/xmlpdf/6034f523-9888-4243-9985-6067ce2759dd/5499.pdf Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 12 Dec, 2025 Editor assigned by journal 08 Dec, 2025 Editor invited by journal 19 Nov, 2025 Submission checks completed at journal 18 Nov, 2025 First submitted to journal 18 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8088844","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":560537724,"identity":"4e1b01f3-bf69-44df-9f05-66e7e005afdd","order_by":0,"name":"Godswill Uzoechina","email":"data:image/png;base64,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","orcid":"","institution":"University of Nigeria Teaching Hospital","correspondingAuthor":true,"prefix":"","firstName":"Godswill","middleName":"","lastName":"Uzoechina","suffix":""},{"id":560537725,"identity":"97bf753f-7d58-4b91-b453-2984a992a5e6","order_by":1,"name":"Winnifred Njideka Adiri","email":"","orcid":"","institution":"University of Nigeria Teaching Hospital","correspondingAuthor":false,"prefix":"","firstName":"Winnifred","middleName":"Njideka","lastName":"Adiri","suffix":""},{"id":560537726,"identity":"aa07c8e1-5f6b-40db-a56e-017cb90c71ed","order_by":2,"name":"Osajiuba Treasure","email":"","orcid":"","institution":"University of Nigeria Teaching Hospital","correspondingAuthor":false,"prefix":"","firstName":"Osajiuba","middleName":"","lastName":"Treasure","suffix":""}],"badges":[],"createdAt":"2025-11-11 16:32:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8088844/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8088844/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98624483,"identity":"0862fe1d-af12-464d-99b5-3a638b43200e","added_by":"auto","created_at":"2025-12-19 17:08:26","extension":"png","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16099,"visible":true,"origin":"","legend":"","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/86e4a2b6d260fc9791d340e5.png"},{"id":98503984,"identity":"16b6bde8-abe9-4144-a464-921cbb2af303","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40020,"visible":true,"origin":"","legend":"","description":"","filename":"PilotStudyofHypertensionScreeningandMachineLearningPredictionUsingCommunityOutreachDatafromNkpokitiEnuguNigeria.docx","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/e728642642f68df65303c382.docx"},{"id":98625262,"identity":"92f1259c-2053-41ed-af35-736d30fb2140","added_by":"auto","created_at":"2025-12-19 17:09:00","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":67604,"visible":true,"origin":"","legend":"","description":"","filename":"Fig10.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/2b00a19bccb18ab7eee1a449.png"},{"id":98624859,"identity":"654d2c26-a981-4edf-92aa-3571e06d94e2","added_by":"auto","created_at":"2025-12-19 17:08:46","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19956,"visible":true,"origin":"","legend":"","description":"","filename":"Fig11.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/d641d3eb77c2e4a4a697e428.png"},{"id":98624828,"identity":"8be18255-f9dd-4bf9-bcaf-f0cacbd6a10e","added_by":"auto","created_at":"2025-12-19 17:08:44","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19788,"visible":true,"origin":"","legend":"","description":"","filename":"Fig12.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/718d5d92a4c20ea55d8e47ad.png"},{"id":98503992,"identity":"884d056a-af1e-4699-9b00-102e194c3de6","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20107,"visible":true,"origin":"","legend":"","description":"","filename":"Fig13.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/e490a4b679d88916f7b473c7.png"},{"id":98504001,"identity":"6a2954b1-66ef-4711-bf36-8af96e56645e","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21106,"visible":true,"origin":"","legend":"","description":"","filename":"Fig14.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/03ebf42465667b485236b36d.png"},{"id":98624712,"identity":"622db824-e9e1-4a85-bf57-04b0d2088168","added_by":"auto","created_at":"2025-12-19 17:08:40","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20822,"visible":true,"origin":"","legend":"","description":"","filename":"Fig15.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a6a14300786eaae1b5094c51.png"},{"id":98624223,"identity":"04c125f6-1d50-4a22-a8f5-fbe7a4307511","added_by":"auto","created_at":"2025-12-19 17:08:10","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22456,"visible":true,"origin":"","legend":"","description":"","filename":"Fig16.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/589b4cd213bd71d24270263f.png"},{"id":98503998,"identity":"ca701ecf-ea2c-4c93-b165-e6c342b5be55","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27477,"visible":true,"origin":"","legend":"","description":"","filename":"Fig17.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/81a37f7e5dbbaeba5be46c94.png"},{"id":98624145,"identity":"a2848545-6985-4c1a-b166-c72abf248afc","added_by":"auto","created_at":"2025-12-19 17:08:04","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12522,"visible":true,"origin":"","legend":"","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/6f5dd31f7180fb315cc88e0c.png"},{"id":98624469,"identity":"1ea11601-d549-448b-93fa-1ae828f77888","added_by":"auto","created_at":"2025-12-19 17:08:26","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13194,"visible":true,"origin":"","legend":"","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/1b132c416a45787ba6e7c529.png"},{"id":98504004,"identity":"185a5916-c34e-41b0-9d7d-6199218c423a","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27420,"visible":true,"origin":"","legend":"","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/f0dd9b94a215908786b425c1.png"},{"id":98624261,"identity":"7ef615e6-bb55-4ff4-8ff5-c97477ad4d5f","added_by":"auto","created_at":"2025-12-19 17:08:14","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15948,"visible":true,"origin":"","legend":"","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/abcac4c2436183386dc94d7a.png"},{"id":98504009,"identity":"4b8e78cb-1bd0-4e0f-9ab8-f63b99cd9eed","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49782,"visible":true,"origin":"","legend":"","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/b4834ec6d4088ff46583a02a.png"},{"id":98504017,"identity":"d883b74d-90e9-4524-94c5-9627ccb66f39","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":39125,"visible":true,"origin":"","legend":"","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/ae17c297af91961fe7b9b96e.png"},{"id":98623842,"identity":"ed72596a-78f0-44c7-b166-fc0e6461d4af","added_by":"auto","created_at":"2025-12-19 17:07:40","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41320,"visible":true,"origin":"","legend":"","description":"","filename":"Fig8..png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/c89c5ae92e2431298b6947c7.png"},{"id":98624484,"identity":"87fa78bf-16ff-4545-ad48-263d79880613","added_by":"auto","created_at":"2025-12-19 17:08:26","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29649,"visible":true,"origin":"","legend":"","description":"","filename":"Fig9.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/48bc8a3cfb38ee0c8d5739a1.png"},{"id":98624800,"identity":"740755d2-87e3-432a-a99e-12ca7927d405","added_by":"auto","created_at":"2025-12-19 17:08:43","extension":"json","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6771,"visible":true,"origin":"","legend":"","description":"","filename":"9771a0029e154ef295de017954f055c1.json","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/2d83841715298e9081802f32.json"},{"id":98624558,"identity":"82fc1f7a-838d-4e4f-a984-69dbb9c21475","added_by":"auto","created_at":"2025-12-19 17:08:30","extension":"xml","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143632,"visible":true,"origin":"","legend":"","description":"","filename":"9771a0029e154ef295de017954f055c11enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/361f9daa06ca6fc6120e5f73.xml"},{"id":98504015,"identity":"dcf93149-3865-4f56-a617-31e2090bb976","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16099,"visible":true,"origin":"","legend":"","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/077aaed5d9bca44f3d479e42.png"},{"id":98623998,"identity":"008fe353-881b-43c8-bd22-75f83a4688ad","added_by":"auto","created_at":"2025-12-19 17:07:53","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":67604,"visible":true,"origin":"","legend":"","description":"","filename":"Fig10.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/b6598f1b8abd0c2f340ac9b9.png"},{"id":98504037,"identity":"60ead3ec-484e-4215-a3bc-2430e60be39d","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19956,"visible":true,"origin":"","legend":"","description":"","filename":"Fig11.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/f43b07711a558d4d9caf700a.png"},{"id":98625164,"identity":"89693149-bf05-4d8b-87f9-5aece4787e5e","added_by":"auto","created_at":"2025-12-19 17:08:58","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19788,"visible":true,"origin":"","legend":"","description":"","filename":"Fig12.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/6a9107f2dbec4af7de156217.png"},{"id":98624408,"identity":"99a93252-c803-4dac-99f0-68f7a4f74032","added_by":"auto","created_at":"2025-12-19 17:08:23","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20107,"visible":true,"origin":"","legend":"","description":"","filename":"Fig13.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/5c3e48f627d41768db2b6714.png"},{"id":98775078,"identity":"da3018ba-af2b-4fc7-9e7d-d7a3f891edb6","added_by":"auto","created_at":"2025-12-22 12:18:22","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21106,"visible":true,"origin":"","legend":"","description":"","filename":"Fig14.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/5cd54b5a0ee08c089b5618e8.png"},{"id":98624816,"identity":"82dd4319-749a-48ae-9e9c-fd168d7725d7","added_by":"auto","created_at":"2025-12-19 17:08:43","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20822,"visible":true,"origin":"","legend":"","description":"","filename":"Fig15.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/f0e350489e8e0d803ccbf020.png"},{"id":98624452,"identity":"acace90f-3f39-4820-9338-6636ae1b8285","added_by":"auto","created_at":"2025-12-19 17:08:25","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22456,"visible":true,"origin":"","legend":"","description":"","filename":"Fig16.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a606fc41d526a07fb79e4fdd.png"},{"id":98624427,"identity":"76060168-36fd-47a3-b62f-d4f21e1f5c25","added_by":"auto","created_at":"2025-12-19 17:08:24","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27477,"visible":true,"origin":"","legend":"","description":"","filename":"Fig17.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/d999db8d01b6bf1ec1e5eb04.png"},{"id":98624467,"identity":"c5c66a65-a047-42d1-a79b-af95a9002422","added_by":"auto","created_at":"2025-12-19 17:08:26","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12522,"visible":true,"origin":"","legend":"","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/23e6b105c4a2dfd035ed7704.png"},{"id":98504026,"identity":"ec4eebf6-4c52-4a1e-affd-13d679eda283","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13194,"visible":true,"origin":"","legend":"","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/cab2f8d1e50c64daa4015c5f.png"},{"id":98625419,"identity":"0a79fba8-d4ad-4d2b-a209-0ca4aa058585","added_by":"auto","created_at":"2025-12-19 17:09:06","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27420,"visible":true,"origin":"","legend":"","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/e42fa570a0216119fc75efe3.png"},{"id":98624897,"identity":"8a4af5ff-a9bb-4dfc-8b15-66d330ecd547","added_by":"auto","created_at":"2025-12-19 17:08:48","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15948,"visible":true,"origin":"","legend":"","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a7f8e4c725f9659f773e0f83.png"},{"id":98504033,"identity":"0289140a-f9e2-42ae-bb43-be7edf987c76","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49782,"visible":true,"origin":"","legend":"","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/87b78fce49f8339b962716fe.png"},{"id":98504029,"identity":"6497b16e-bf34-4196-b67d-7ffbdd3b25d5","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":39125,"visible":true,"origin":"","legend":"","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/42cdf39b8311ce04f606d037.png"},{"id":98504042,"identity":"a989051c-075c-4c95-be90-c809841aeb0e","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"png","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41320,"visible":true,"origin":"","legend":"","description":"","filename":"Fig8..png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a48c0dd7666d60e48587e069.png"},{"id":98624639,"identity":"b081e737-f185-437d-9d3c-1b1b4fc03ae5","added_by":"auto","created_at":"2025-12-19 17:08:36","extension":"png","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29649,"visible":true,"origin":"","legend":"","description":"","filename":"Fig9.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/89bf88df06dae99ff7ee9dec.png"},{"id":98504048,"identity":"83f84b23-4d80-4bca-9816-3de50dfb3228","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"png","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4184,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/0925d7c3b2de76d5bfbdb30d.png"},{"id":98504019,"identity":"16defc74-e03d-4c3f-98a1-e07b6ff28355","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":38,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20075,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig10.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/01ba06d070d77dfa3b27ef17.png"},{"id":98504041,"identity":"e855071f-2019-4285-a9ff-731d57393afe","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"png","order_by":39,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5424,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig11.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/cd4866847a900e77ca24b303.png"},{"id":98625028,"identity":"3da45aec-b2e8-41a9-a44c-964f62afcf98","added_by":"auto","created_at":"2025-12-19 17:08:54","extension":"png","order_by":40,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5406,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig12.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/0644f1b8d4954710d1fd0d16.png"},{"id":98504028,"identity":"59e9265a-dcc2-477f-b269-266e038c01d3","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":41,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5486,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig13.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/3a3886282d0817f07a01331f.png"},{"id":98504021,"identity":"89209013-d995-4b99-a33a-70c2c6bc5462","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":42,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5539,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig14.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/c4b6f3a09a5299bad5b4ffb3.png"},{"id":98624725,"identity":"825fb56e-b055-4fea-b7c9-25259e762535","added_by":"auto","created_at":"2025-12-19 17:08:40","extension":"png","order_by":43,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5664,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig15.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/caf84eb6d879aaf23e18c941.png"},{"id":98624267,"identity":"edcf5f2f-53e0-443c-b3cd-d623a2299c7c","added_by":"auto","created_at":"2025-12-19 17:08:14","extension":"png","order_by":44,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5775,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig16.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/95eaa7c219513db49bc9d50a.png"},{"id":98504030,"identity":"e12d5968-b16f-4a2d-80d9-1a2ba954afe5","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":45,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10031,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig17.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/1e1cf4705610abd70698c78a.png"},{"id":98624775,"identity":"a22f4d21-27a4-4972-99bf-99c4358f1808","added_by":"auto","created_at":"2025-12-19 17:08:42","extension":"png","order_by":46,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6223,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/e80d5790a56ef12ec52348d1.png"},{"id":98624533,"identity":"b6befce6-bc45-45ad-ba62-622866a300d9","added_by":"auto","created_at":"2025-12-19 17:08:29","extension":"png","order_by":47,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4499,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/4d96d810c2a324acb6fd446f.png"},{"id":98625383,"identity":"161e37e7-e3da-4b25-8200-f4c5e9170a8c","added_by":"auto","created_at":"2025-12-19 17:09:04","extension":"png","order_by":48,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7626,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/7eeb977ff2908d257e851226.png"},{"id":98625425,"identity":"1df01609-ea90-4f81-af6d-f197fa2b4af7","added_by":"auto","created_at":"2025-12-19 17:09:06","extension":"png","order_by":49,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4924,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/645ca1083adb2b49390e96fb.png"},{"id":98624727,"identity":"d67f8f05-a21c-463d-809e-b56cd0477aaf","added_by":"auto","created_at":"2025-12-19 17:08:41","extension":"png","order_by":50,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12887,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a9066d1f1661864310bc8dc2.png"},{"id":98504044,"identity":"e2d1cf57-aa02-47a8-9b5a-9b02f8d903e7","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"png","order_by":51,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11535,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig7.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/aa49f1d4889e109d84b63897.png"},{"id":98504040,"identity":"4968dfed-6a30-4e3a-8e16-69562fe7040e","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"png","order_by":52,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13724,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig8..png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/974e72bc09aa82dd93b04c28.png"},{"id":98504034,"identity":"0bff4e21-d30a-4803-adcb-ed1154731bd7","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":53,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9534,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig9.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/075dce13d0b14c2628f894b8.png"},{"id":98504049,"identity":"5c471c96-5916-4047-8631-1346f222cad5","added_by":"auto","created_at":"2025-12-18 10:22:37","extension":"xml","order_by":54,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":138763,"visible":true,"origin":"","legend":"","description":"","filename":"9771a0029e154ef295de017954f055c11structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/67d77c1bf9085d04b25d1692.xml"},{"id":98624734,"identity":"e08fe949-7578-49b2-9859-42b280c32ef7","added_by":"auto","created_at":"2025-12-19 17:08:41","extension":"html","order_by":55,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":159118,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/41c3ea4266db43fdb09c61d9.html"},{"id":98624401,"identity":"70ecae04-c413-4ca0-988f-61e3ed9a655e","added_by":"auto","created_at":"2025-12-19 17:08:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":16099,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAge distribution of participants: \u003c/strong\u003eHistogram showing the distribution of participant ages (n = 115).\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/81951bc8050a1143920b240e.png"},{"id":98503979,"identity":"d796fe2e-1ba7-4942-befe-11eb6ea8afbf","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":12522,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDistribution of systolic blood pressure (SBP): \u003c/strong\u003eHistogram of systolic blood pressure measurements (mmHg) among all participants.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/03242004267ac9d824910929.png"},{"id":98503982,"identity":"faf1f9b7-d1c1-4a72-874a-9647036feb7e","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":13194,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDistribution of diastolic blood pressure (DBP): \u003c/strong\u003eHistogram of diastolic blood pressure measurements (mmHg) in the study sample.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/97736a79d2018251ef6acb0f.png"},{"id":98623892,"identity":"d8850c89-9907-4718-bf3e-26558bb793b2","added_by":"auto","created_at":"2025-12-19 17:07:46","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":27420,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRelationship between age and systolic blood pressure: \u003c/strong\u003eScatterplot showing a positive trend between participant age and SBP with linear fit.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/8cc8a29dc00d1969b8adcd8c.png"},{"id":98503985,"identity":"e658eb4f-ea96-450f-8b89-3f0b58232697","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":15948,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSystolic blood pressure by sex: \u003c/strong\u003eBoxplots comparing SBP distributions between male and female participants; no significant difference observed (p = 0.8606).\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/3d8577710105590b8b137585.png"},{"id":98624433,"identity":"cd78726a-79da-48d4-840a-3947bb64356e","added_by":"auto","created_at":"2025-12-19 17:08:24","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":49782,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation matrix for age and blood pressure variables: \u003c/strong\u003eAnnotated heatmap illustration for Age, SBP, DBP, Pulse Pressure (PP), and Mean Arterial Pressure (MAP).\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/ec9be91d186d43641309693c.png"},{"id":98625349,"identity":"d308112b-c7ab-44cb-bf5c-f38d6c70451e","added_by":"auto","created_at":"2025-12-19 17:09:03","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":39125,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC curves for logistic regression, random forest, and gradient boosting models: \u003c/strong\u003eReceiver-operating characteristic curves comparing model discrimination on the hold-out test set.\u003c/p\u003e","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/e2e8b3004ebb943a0cdb378d.png"},{"id":98625364,"identity":"7d16caa7-bb9c-4fbc-92a3-abee83804fec","added_by":"auto","created_at":"2025-12-19 17:09:03","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":41320,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBootstrap ROC for logistic regression with 95% confidence interval: \u003c/strong\u003eMean ROC curve from 2000 bootstrap resamples showing logistic regression performance variability.\u003c/p\u003e","description":"","filename":"Fig8..png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/2e0affe9cb80aefb7beecb02.png"},{"id":98503990,"identity":"e3310312-596b-437a-85e6-3ff1eaf0f9a5","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":29649,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePrecision–recall curves for all models: \u003c/strong\u003ePrecision–recall plots illustrating model performance under class imbalance conditions.\u003c/p\u003e","description":"","filename":"Fig9.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/a9fd6ee71015ca7386425aa7.png"},{"id":98624857,"identity":"3f26e47f-e132-4bb2-b763-dd458e9b62e5","added_by":"auto","created_at":"2025-12-19 17:08:46","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":67604,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCalibration plots for logistic regression, random forest, and gradient boosting: \u003c/strong\u003eReliability diagrams comparing predicted probabilities with observed outcomes for each model.\u003c/p\u003e","description":"","filename":"Fig10.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/8d96b0b6dd1920b61a793f05.png"},{"id":98503994,"identity":"2f0aab99-8729-4764-8a3c-0c18fd2daa09","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":19956,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConfusion-matrix heatmaps at threshold 0.5: \u003c/strong\u003eHeatmaps showing counts of true/false positives and negatives for each model on the test set.\u003c/p\u003e","description":"","filename":"Fig11.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/f09187c4aded39a286182c18.png"},{"id":98503996,"identity":"f1008140-9f0f-4ea6-9dd2-0b1a4596ad45","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":19788,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConfusion-matrix heatmaps at threshold 0.5: \u003c/strong\u003eHeatmaps showing counts of true/false positives and negatives for each model on the test set.\u003c/p\u003e","description":"","filename":"Fig12.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/051e91faf10c1d2abc5cf6c4.png"},{"id":98624249,"identity":"621f8880-fb06-441c-82de-7229dec79f62","added_by":"auto","created_at":"2025-12-19 17:08:12","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":20107,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConfusion-matrix heatmaps at threshold 0.5: \u003c/strong\u003eHeatmaps showing counts of true/false positives and negatives for each model on the test set.\u003c/p\u003e","description":"","filename":"Fig13.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/0f1404938eaad4a739cd02f9.png"},{"id":98624498,"identity":"e8fccb19-fdea-492d-8680-87348f5b8fb7","added_by":"auto","created_at":"2025-12-19 17:08:27","extension":"png","order_by":14,"title":"Figure 14","display":"","copyAsset":false,"role":"figure","size":21106,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature importance and logistic regression coefficients: \u003c/strong\u003eBar charts displaying feature importances (RF/GB) and corresponding odds ratios (LR), highlighting SBP as the top predictor.\u003c/p\u003e","description":"","filename":"Fig14.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/6e7c94e8cfa8657e7b0e9dbe.png"},{"id":98504006,"identity":"131d8293-2e30-44ce-a03c-48f66267984c","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":15,"title":"Figure 15","display":"","copyAsset":false,"role":"figure","size":20822,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature importance and logistic regression coefficients: \u003c/strong\u003eBar charts displaying feature importances (RF/GB) and corresponding odds ratios (LR), highlighting SBP as the top predictor.\u003c/p\u003e","description":"","filename":"Fig15.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/14c5f9359f521cc8a69e54b7.png"},{"id":98504005,"identity":"193170fe-906c-47dc-8cce-8ad83bf97e98","added_by":"auto","created_at":"2025-12-18 10:22:35","extension":"png","order_by":16,"title":"Figure 16","display":"","copyAsset":false,"role":"figure","size":22456,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature importance and logistic regression coefficients: \u003c/strong\u003eBar charts displaying feature importances (RF/GB) and corresponding odds ratios (LR), highlighting SBP as the top predictor.\u003c/p\u003e","description":"","filename":"Fig16.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/60c079f8a8fb40c869f8ca00.png"},{"id":98504011,"identity":"4513a37f-af6e-424c-a2f7-1f7b5e08cf2e","added_by":"auto","created_at":"2025-12-18 10:22:36","extension":"png","order_by":17,"title":"Figure 17","display":"","copyAsset":false,"role":"figure","size":27477,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePartial dependence plots for key predictors in the gradient boosting model: \u003c/strong\u003ePlots showing the marginal effect of SBP and age on predicted hypertension probability.\u003c/p\u003e","description":"","filename":"Fig17.png","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/3c8df5250763fa16879e9ae3.png"},{"id":98782620,"identity":"b92fdd93-9e4b-486b-92cd-bfd1acb2adb4","added_by":"auto","created_at":"2025-12-22 12:40:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2479280,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8088844/v1/ba5bf172-49e7-4a07-aed4-1ecf19322455.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Pilot Study of Hypertension Screening and Machine-Learning Prediction Using Community Outreach Data from Nkpokiti, Enugu, Nigeria Short Title: Machine Learning Prediction of Hypertension using Community Blood Pressure Data in Nigeria","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHigh blood pressure is the single leading risk factor for cardiovascular and kidney disease worldwide and premature death, and its effective and early detection is a cornerstone of prevention (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). Globally, more than one billion adults have hypertension, and control rates remain low, especially in low- and middle-income countries with weak and fragile health systems (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). In Nigeria, the burden is significant and rising: national reviews report prevalence figures often within 20\u0026ndash;40% with significant regional disparities (with rates higher in the southern geopolitical zones compared to the north) and consistent deficits in awareness, treatment, and control (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCommunity-based screening and outreach activities are a cost-effective approach to increase detection and awareness, as demonstrated by large campaigns such as May Measurement Month that have screened millions and highlighted the scalability and public-health utility of opportunistic blood pressure (BP) screening (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Yet outreach venues are frequently subjected to practical limitations, such as restricted time, scarcity of trained personnel, and dependence on single BP measurements, which heighten the risk of misclassification due to measurement variability, the white-coat phenomenon, or temporary influences (e.g., exercise, caffeine) (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). As a result, consensus guidelines and measurement sciences literature advocate for repeat measurements and computation of averaged values in order to enhance diagnostic accuracy, but repeat measurements are often not feasible during rapid, mass screening efforts (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMachine-learning (ML) methodologies offer an alternative and complementary option for triage in resource-constrained screening by leveraging predictive signals from basic, routinely collected variables (age, sex, first BP reading) to direct individuals for confirmatory testing or referral. Systematic and empirical research demonstrates that ML can enhance the prediction of cardiovascular risk compared to conventional models; however, there is significant heterogeneity among studies in terms of outcome definitions, predictors utilized, and the reliability of validation methods. Calibration and transparency in reporting are often insufficient; constraints that should be overcome before clinical deployment and application (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). The recent TRIPOD\u0026thinsp;+\u0026thinsp;AI\u0026ensp;guidance stresses strong internal and external validation, transparent reporting of model development and calibration, along with sharing of code and model specification for reproducibility (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThere is a significant gap in evidence for feasibility studies assessing the performance of ML-based hypertension prediction, specifically in real-world outreach screening settings across sub-Saharan Africa, utilizing minimal or low-cost predictors. Such pilots in these settings are needed: (a) to estimate local prevalence under outreach conditions, (b) to measure misclassification induced by use of single as opposed to mean BP recordings, and (c) to determine whether ML techniques achieve useful and adequate discrimination and calibration under field conditions. Actual\u0026ensp;feasibility evidence will guide the development of larger, multicenter validation studies and the possible adoption of ML-enabled decision tools in the community screening workflow.\u003c/p\u003e \u003cp\u003eTherefore, this pilot study is designed to (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) determine the prevalence of hypertension in a community-based outreach population in Nkpokiti, Enugu; (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) measure concordance and misclassification between single and mean BP measurements; and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) construct and internally validate ML models (regularised logistic regression as primary model, with tree-based ensemble models: random forest and gradient boosting as comparators) for predicting hypertension using age, sex and mean BP reading through nested cross-validation, reporting discrimination, calibration, and decision-analytic metrics in accordance with\u0026ensp;TRIPOD\u0026thinsp;+\u0026thinsp;AI guidelines (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e"},{"header":"Materials and Methodology","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design and setting\u003c/h2\u003e \u003cp\u003e This was a cross-sectional pilot study conducted using de-identified data obtained from a community hypertension screening\u0026ensp;in Nkpokiti, Enugu State, Nigeria, by the Nigerian Medical Students\u0026rsquo; Association (NiMSA) and University of Nigeria Medical Students\u0026rsquo; Association (UNMSA) Standing Committees on Policy Implementation (SCOPI), and Research Exchange (SCORE). Data collection took place\u0026ensp;on 17th May 2025. The data consisted of demographic information (age, sex) and paired systolic and diastolic blood pressure (BP) readings for each subject. No new data collection was conducted for this secondary\u0026ensp;analysis.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eParticipants and eligibility\u003c/h3\u003e\n\u003cp\u003eAll outreach attendees with a documented age, sex, and at least one paired systolic (SBP)\u0026ensp;and diastolic (DBP) BP measurement were included. Records were excluded from the study if age or sex were missing or if BP values were implausible (SBP\u0026thinsp;\u0026lt;\u0026thinsp;70\u0026ensp;mmHg or \u0026gt;\u0026thinsp;250 mmHg, DBP\u0026thinsp;\u0026lt;\u0026thinsp;40mmHg or \u0026gt;\u0026thinsp;150 mmHg) (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Following\u0026ensp;quality control, the sample size was 115 participants.\u003c/p\u003e\n\u003ch3\u003eOutcome definition\u003c/h3\u003e\n\u003cp\u003eThe outcome of interest was hypertension (systolic blood pressure [SBP]\u0026thinsp;\u0026ge;\u0026thinsp;140 mmHg and/or diastolic blood\u0026ensp;pressure [DBP]\u0026thinsp;\u0026ge;\u0026thinsp;90 mmHg) according to World Health Organization (WHO) cutoff points in the setting of community-based screening (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). When available, the mean of\u0026ensp;at least two readings was used for classification; otherwise, one valid reading was used.\u003c/p\u003e\n\u003ch3\u003eData preparation and derived variables\u003c/h3\u003e\n\u003cp\u003eData cleaning and\u0026ensp;transformation were performed with reproducible scripts in Python 3.9. Calculated variables were\u0026ensp;pulse pressure (PP\u0026thinsp;=\u0026thinsp;SBP \u0026ndash; DBP, mmHg) and mean arterial pressure (MAP\u0026thinsp;=\u0026thinsp;DBP + [SBP \u0026ndash; DBP]/3, mmHg) (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). For descriptive purposes, age groups were stratified as \u0026lt;\u0026thinsp;40, 40\u0026ndash;59, and \u0026ge;\u0026thinsp;60 years. Erroneous, duplicate, and implausible submissions were excluded. Analytic variables had no\u0026ensp;missingness.\u003c/p\u003e\n\u003ch3\u003eDescriptive statistical analysis\u003c/h3\u003e\n\u003cp\u003eContinuous variables were presented using mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD) or medians (interquartile range [IQR]), and categorical variables as frequencies and percentages. Normality was tested using the Shapiro\u0026ndash;Wilk test and visual inspection (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Between-group comparisons were made with\u0026ensp;Student\u0026rsquo;s t test or Mann\u0026ndash;Whitney U test as appropriate. Univariable logistic regression models calculated the unadjusted odds ratios (ORs) and 95%\u0026ensp;confidence intervals (CIs) for each covariate (age, sex, SBP, DBP, PP, MAP). The variance inflation factors (VIFs) were used to assess and detect multicollinearity among BP measures, and highly collinear predictors were removed from the multivariable models (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eMachine-learning framework\u003c/h2\u003e \u003cp\u003eThe machine-\u0026ensp;learning (ML) analysis was exploratory to assess feasibility and internal validity, and not for definitive model development. The predictors were age, sex, SBP, and PP; these had been selected for parsimony and a low level of collinearity (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). The following three model classes were trained:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePenalized logistic regression (L2 regularisation; inverse strength C\u0026thinsp;=\u0026thinsp;0.01 \u0026ndash;\u0026ensp;10).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eRandom forest classifier (n_estimators =\u0026ensp;200; max_depth\u0026thinsp;=\u0026thinsp;3 \u0026ndash; None).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eGradient boosting classifier (n_estimators\u0026thinsp;=\u0026thinsp;100\u0026ndash;200; learning_rate\u0026thinsp;=\u0026thinsp;0.05\u0026ndash;0.1; max_depth\u0026thinsp;=\u0026thinsp;2\u0026ndash;3).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eEach of the models was used with class_weight\u0026ensp;= \u0026lsquo;balanced\u0026rsquo; to address a relatively moderate class imbalance (prevalence\u0026thinsp;\u0026asymp;\u0026thinsp;25%). No standalone single tree classifier was used.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCross-validation and hyperparameter tuning\u003c/h3\u003e\n\u003cp\u003eA nested 5-fold stratified cross-validation (CV) was implemented to avoid overfitting (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). Each outer fold estimated generalisation performance, while inner folds tuned hyperparameters by grid search using the area under the receiver operating characteristic curve (AUC) as the selection metric. To avoid data leakage, all preprocessing procedures (scaling, encoding, and imputation) were performed\u0026ensp;within the inner-CV loops (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Continuous variables were standardized (zero mean and unit variance) using the training fold statistics, and categorical variables were binary encoded.\u003c/p\u003e\n\u003ch3\u003eModel evaluation\u003c/h3\u003e\n\u003cp\u003ePerformance was evaluated using discrimination, calibration, and classification indices. Discrimination was reported in terms of AUC and average precision (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). Calibration was analysed with the Brier score and reliability plots\u0026ensp;(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Accuracy, sensitivity, specificity, precision, F1-score, and balanced accuracy were averaged across the outer folds. A bootstrap resampling with 2,000\u0026ensp;iterations was then used to compute 95% CIs for the AUC and Brier scores (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eModel interpretability\u003c/h2\u003e \u003cp\u003eTransparency of the model was prioritised in concordance with TRIPOD-AI guidance (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). For logistic regression, the coefficients were presented as odds ratios (ORs) over 10 mmHg or per standard-deviation increase with bootstrap 95% CIs. For the ensemble models, feature importance scores were averaged across outer folds, and the top predictors (age, SBP) were plotted using partial dependence plots (PDP) for visualising marginal effects. SHAP (SHapley Additive exPlanations) values were calculated to assess local and global feature contributions for the tree-based models (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity analyses\u003c/h2\u003e \u003cp\u003eWe repeated analyses (a) replacing SBP with MAP; (b) excluding derived variables (PP, MAP) to assess the sensitivity of collinearity; (c) employing a different hypertension threshold (\u0026ge;\u0026thinsp;130/80 mmHg); and (d) using different random seeds and CV folds (5-fold vs 10-fold) to confirm the stability of the metrics.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eStatistical software and reproducibility\u003c/h2\u003e \u003cp\u003eAll analyses were conducted using Python 3.9 with pandas, numpy, scikit-learn, scipy, statsmodels, matplotlib, and shap packages (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Random seeds were set to ensure reproducibility. The entire analysis pipeline, including code and environment files, is deposited in a public GitHub repository. Anonymized data may be made available on reasonable request after approval from an ethics committee.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eEthical approval and consent\u003c/strong\u003e \u003cp\u003eThe consents for participation in the outreach and for the use of data in research were verbal and written. Permission for secondary analysis was obtained from the Ethics Committee of the University of Nigeria Teaching Hospital (NHREC/05/01/2008B-FW00002458-1RB00002327). All approaches conformed to the Declaration of Helsinki and the Nigerian National Code of Health Research Ethics (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eReporting standards\u003c/h2\u003e \u003cp\u003eThe study was conducted in accordance with the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis with Artificial Intelligence (TRIPOD-AI) guidance (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e) to facilitate transparent methods, reproducibility, and explainability.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eSample and data quality\u003c/h2\u003e \u003cp\u003eOne hundred and fifteen (115) participant records met the inclusion\u0026ensp;criteria and were included in the analysis. Age, sex, systolic blood pressure (SBP), and diastolic blood pressure (DBP) were\u0026ensp;available for all records; there were no missing values across the analytic variables, and no duplicate records were found. There were no implausible BP values detected after pre-specified quality control procedures (no SBP\u0026thinsp;\u0026lt;\u0026thinsp;70 or \u0026ge;\u0026thinsp;250\u0026ensp;mmHg; no DBP\u0026thinsp;\u0026lt;\u0026thinsp;40 or \u0026ge;\u0026thinsp;150 mmHg).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eParticipant characteristics\u003c/h2\u003e \u003cp\u003eThe mean age of\u0026ensp;participants was 36.4 years (SD 14.4; median 33; range 18\u0026ndash;84). N\u0026thinsp;=\u0026thinsp;33 were male\u0026ensp;(28.7%) and n\u0026thinsp;=\u0026thinsp;82 (71.3%) female. The summary BP measures were: mean SBP 124.5 mmHg (SD\u0026ensp;11.7; median 126; range 100\u0026ndash;163), mean DBP 80.7 mmHg (SD 9.9; median 80; range 60\u0026ndash;116), mean pulse pressure 43.8 mmHg (SD 9.8; median 44; range 20\u0026ndash;80), and mean arterial pressure 95.3 mmHg (SD 9.5). Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents the baseline demographic and blood pressure characteristics of the study participants.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eParticipant characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall (N\u0026thinsp;=\u0026thinsp;115)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e36.38 (14.39)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, median (IQR)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e33 (\u0026mdash;)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, range\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18\u0026ndash;84\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex \u0026mdash; female, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e82 (71.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex \u0026mdash; male, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e33 (28.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSBP, mean (SD) (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e124.50 (11.72)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSBP, median (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e126\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSBP, range (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u0026ndash;163\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBP, mean (SD) (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e80.73 (9.92)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBP, median (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e80\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBP, range (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60\u0026ndash;116\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePulse pressure, mean (SD) (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e43.77 (9.80)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMAP, mean (SD) (mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e95.32 (9.49)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension (WHO definition), n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29 (25.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eParticipant age distribution is shown in Fig.\u0026nbsp;1.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;1. Age distribution of participants\u003c/strong\u003e \u003cp\u003e Histogram showing the distribution of participant ages (n\u0026thinsp;=\u0026thinsp;115).\u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe distribution of systolic blood pressure (SBP) measurements is presented in Fig.\u0026nbsp;2, while diastolic blood pressure (DBP) is shown in Fig.\u0026nbsp;3.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;2. Distribution of systolic blood pressure (SBP)\u003c/strong\u003e \u003cp\u003eHistogram of systolic blood pressure measurements (mmHg) among all participants.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;3. Distribution of diastolic blood pressure (DBP)\u003c/strong\u003e \u003cp\u003eHistogram of diastolic blood pressure measurements (mmHg) in the study sample.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eHypertension prevalence\u003c/h2\u003e \u003cp\u003eApplication of the predefined criteria for SBP\u0026thinsp;\u0026ge;\u0026thinsp;140 mmHg and/or DBP\u0026thinsp;\u0026ge;\u0026thinsp;90 mmHg (WHO screening definition) resulted in 29 of 115 participants being classified as having high blood pressure, with a total prevalence of 25.2% (29/115) (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). By sex, prevalence was 21/82\u0026thinsp;=\u0026thinsp;25.6% for females and 8/33\u0026thinsp;=\u0026thinsp;24.2% for males. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the distribution of hypertension across age and sex groups, including normality and group comparison tests.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eHypertension prevalence by age and sex\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStratum\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHypertensive n (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u0026thinsp;\u0026lt;\u0026thinsp;40 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10 (12.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge 40\u0026ndash;59 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e13 (52.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u0026thinsp;\u0026ge;\u0026thinsp;60 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6 (54.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale (overall)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e21 (25.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale (overall)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8 (24.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOverall\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e115\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e29 (25.2%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eA positive relationship between age and SBP is illustrated in Fig.\u0026nbsp;4.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;4. Relationship between age and systolic blood pressure\u003c/strong\u003e \u003cp\u003eScatterplot showing a positive trend between participant age and SBP with linear fit.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eFigure 5 compares SBP levels between male and female participants, showing no significant difference (p\u0026thinsp;=\u0026thinsp;0.8606).\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;5. Systolic blood pressure by sex\u003c/strong\u003e \u003cp\u003eBoxplots comparing SBP distributions between male and female participants; no significant difference observed (p\u0026thinsp;=\u0026thinsp;0.8606).\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eDistributional checks and group comparisons\u003c/h2\u003e \u003cp\u003eShapiro\u0026ndash;Wilk tests highlighted a digression from normality in some variables\u0026ensp;(Age: W\u0026thinsp;=\u0026thinsp;0.871, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; SBP: W\u0026thinsp;=\u0026thinsp;0.977, p\u0026thinsp;=\u0026thinsp;0.044; DBP: W\u0026thinsp;=\u0026thinsp;0.950, p\u0026thinsp;=\u0026thinsp;0.0003; MAP: W\u0026thinsp;=\u0026thinsp;0.976, p\u0026thinsp;=\u0026thinsp;0.0374), but there was no strong evidence against normality in pulse pressure (W\u0026thinsp;=\u0026thinsp;0.983, p\u0026thinsp;=\u0026thinsp;0.142).\u003c/p\u003e \u003cp\u003eThere was no evidence of a sex-based difference in SBP across the sample (two-sample\u0026ensp;test: t\u0026thinsp;=\u0026thinsp;0.176, p\u0026thinsp;=\u0026thinsp;0.861), and there was no significant association between sex and hypertensive status (χ\u0026sup2; = 0.000, p\u0026thinsp;=\u0026thinsp;1.000) via contingency testing, as SBP normality by sex: male SBP Shapiro p\u0026thinsp;=\u0026thinsp;0.371; female SBP Shapiro p\u0026thinsp;=\u0026thinsp;0.073. A Kruskal\u0026ndash;Wallis test showed that there is a reliable difference in SBP across the predetermined\u0026ensp;age groups (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), in accordance with a stepwise increase in prevalence with age.\u003c/p\u003e \u003cp\u003eCorrelations among age, SBP, DBP, pulse pressure (PP), and mean arterial pressure (MAP) are visualized in Fig.\u0026nbsp;6.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;6. Correlation matrix for age and blood pressure variables\u003c/strong\u003e \u003cp\u003eAnnotated heatmap illustration for Age, SBP, DBP, Pulse Pressure (PP), and Mean Arterial Pressure (MAP).\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eUnivariable associations with hypertension\u003c/h2\u003e \u003cp\u003eSBP, DBP, and MAP demonstrated strong univariable associations with the hypertension label; sex did not. Because multiple BP formulations are deterministically related, these univariable findings were interpreted cautiously and steered collinearity checks. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e summarizes the univariable associations between candidate predictors and hypertension status.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eUnivariable logistic regression result\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePredictor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (per year)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.024\u0026ndash;1.090\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.0005\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (male vs female)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.930\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.364\u0026ndash;2.375\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.879\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSBP (per mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.208\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.116\u0026ndash;1.307\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.00001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBP (per mmHg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.681\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.314\u0026ndash;2.151\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000035\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePulse pressure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.988\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.946\u0026ndash;1.032\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.595\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMAP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.351\u0026ndash;2.223\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000015\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eMulticollinearity diagnostics\u003c/h2\u003e \u003cp\u003eThe Variance inflation factor (VIF) analysis on Age, SBP, DBP, PP, and MAP showed very large or infinite VIFs for SBP/DBP/PP/MAP (highlighting deterministic collinearity) and a small VIF for Age (~\u0026thinsp;1.30). Furthermore, multivariable models were limited to\u0026ensp;a few parsimonious variables (SBP kept as the principal BP predictor; PP investigated separately).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eSingle-predictor discrimination (AUC with bootstrap CIs)\u003c/h2\u003e \u003cp\u003eBootstrap AUCs\u0026ensp;(2000 resamples) for individual predictors were:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eSBP: AUC\u0026ensp;= 0.865; 95% CI\u0026ensp;= (0.777, 0.941).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAge:\u0026ensp;AUC\u0026thinsp;=\u0026thinsp;0.738; 95% CI = (0.615, 0.852).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePulse pressure:\u0026ensp;AUC\u0026thinsp;=\u0026thinsp;0.455; 95% CI = (0.325, 0.588).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eSBP alone showed significant discrimination for the binary\u0026ensp;hypertension label, age showed modest discrimination, and pulse pressure alone performed weakly.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eClass balance and train/test split\u003c/h2\u003e \u003cp\u003eThe overall class distribution (total) was [86 29] (negative, positive). We adopted a stratified 80/20 split for training and hold-out testing, which yielded:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTraining set: n\u0026thinsp;=\u0026thinsp;92 samples with 23 hypertensives (shape reported as (92, 4) where 4 is the number of predictors/features used).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTest set (hold-out): n\u0026thinsp;=\u0026thinsp;23 samples with 6 hypertensives (shape (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e)).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThese numbers were used for the illustrative hold-out performance reported below; AUC optimization and nested CV (inner 5-fold) were used for hyperparameter tuning during development.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eMachine-learning model training, best hyperparameters, and hold-out performance\u003c/h2\u003e \u003cp\u003eModels evaluated were penalised logistic regression (primary), random forest (RF), and gradient boosting (GB). The perfect AUC for the GB model on the small hold-out (n\u0026ensp;= 23; 6 hypertensives) possibly demonstrates optimistic estimation bias due to the limited sample; bootstrap CIs of exactly 1.000 are an\u0026ensp;artefact of this small resampling distribution and should be interpreted cautiously.\u003c/p\u003e \u003cp\u003eModel coefficients and feature importances were consistent: SBP had the greatest predictive importance in all models. Example LR coefficients (standardised\u0026ensp;predictors): SBP coef\u0026thinsp;=\u0026thinsp;3.062 (OR 21.37 per SD), Age coef\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.046 (OR 0.955 per SD), Sex coef\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.044 (OR 0.957 per SD), PP coef\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;1.504 (OR 0.222 per SD). RF and GB feature-importance rankings assigned the biggest weight to systolic pressure (RF importance 0.554; GB\u0026ensp;importance 0.727). Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e reports the predictive performance metrics of the logistic regression, random forest, and gradient boosting models.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performance summary\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBest params (selected)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTest AUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC_boot_mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95% CI (AUC)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eBrier\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLogistic regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC\u0026thinsp;=\u0026thinsp;1, penalty\u0026thinsp;=\u0026thinsp;L2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.800\u0026ndash;1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.667\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.824\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom forest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003en_estimators\u0026thinsp;=\u0026thinsp;200, max_depth\u0026thinsp;=\u0026thinsp;None, min_samples_leaf\u0026thinsp;=\u0026thinsp;1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.961\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.960\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.857\u0026ndash;1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.088\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGradient boosting\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003en_estimators\u0026thinsp;=\u0026thinsp;200, lr\u0026thinsp;=\u0026thinsp;0.1, max_depth\u0026thinsp;=\u0026thinsp;3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u0026ndash;1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.038\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e presents the comparative performance of the three machine learning models on the hold-out test set, showing that all achieved high discrimination, with gradient boosting performing best overall.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel comparison summary (hold-out test set)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003emodel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC_boot_mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC_CI_low\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAUC_CI_high\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLogisticRegression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.667\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.824\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandomForest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.961\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.960\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGradientBoosting\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe discriminative performance of the logistic regression, random forest, and gradient boosting models is shown by the ROC curves in Fig.\u0026nbsp;7.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;7. ROC curves for logistic regression, random forest, and gradient boosting models\u003c/strong\u003e \u003cp\u003eReceiver-operating characteristic curves comparing model discrimination on the hold-out test set.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe robustness of the logistic regression model is further highlighted by the bootstrap ROC curve with a 95% confidence interval in Fig.\u0026nbsp;8.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;8. Bootstrap ROC for logistic regression with 95% confidence interval\u003c/strong\u003e \u003cp\u003eMean ROC curve from 2000 bootstrap resamples showing logistic regression performance variability.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eFigure 9 displays the precision\u0026ndash;recall curves, highlighting model performance under class-imbalance conditions.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;9. Precision\u0026ndash;recall curves for all models\u003c/strong\u003e \u003cp\u003ePrecision\u0026ndash;recall plots illustrating model performance under class imbalance conditions.\u003c/p\u003e \u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eCalibration and probabilistic accuracy\u003c/h2\u003e \u003cp\u003eCalibration was evaluated using reliability plots and the Brier scores. Brier scores on the test set were LR\u0026thinsp;=\u0026thinsp;0.094, RF\u0026thinsp;=\u0026thinsp;0.088, GB\u0026thinsp;=\u0026thinsp;0.038. Reliability diagrams indicated that GB demonstrated the best apparent calibration on the small test set, while LR and RF showed slight miscalibration in some probability bins. Due to the sample size, these assessments of calibration are descriptive and necessitate confirmation and definitiveness in larger samples.\u003c/p\u003e \u003cp\u003eCalibration plots comparing predicted probabilities with observed outcomes for each model are provided in Fig.\u0026nbsp;10.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;10. Calibration plots for logistic regression, random forest, and gradient boosting\u003c/strong\u003e \u003cp\u003eReliability diagrams comparing predicted probabilities with observed outcomes for each model.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eModel confusion matrices at the 0.5 probability threshold are shown in Figs.\u0026nbsp;11\u0026ndash;13.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;11\u0026ndash;13. Confusion-matrix heatmaps at threshold 0.5\u003c/strong\u003e \u003cp\u003eHeatmaps showing counts of true/false positives and negatives for each model on the test set.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eExplainability and diagnostics\u003c/h2\u003e \u003cp\u003eROC curves for all models also demonstrated good discrimination on the hold-out dataset, with higher discrimination for GB and RF when compared to LR. Partial dependence plots (GB) demonstrated a monotonic increase in predicted\u0026ensp;probability with rising SBP and a smaller positive effect for age. SHAP summaries for tree ensembles also confirmed SBP as the primary determinant for predictive analysis.\u003c/p\u003e \u003cp\u003eFeature importance scores for the random forest and gradient boosting models, along with logistic regression coefficients (odds ratios), are summarized in Figs.\u0026nbsp;14\u0026ndash;16.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;14\u0026ndash;16. Feature importance and logistic regression coefficients\u003c/strong\u003e \u003cp\u003eBar charts displaying feature importances (RF/GB) and corresponding odds ratios (LR), highlighting SBP as the top predictor.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003eSensitivity analyses\u003c/h2\u003e \u003cp\u003eResults were robust to alternate choices of predictors (MAP in place of SBP) and to the removal of derived variables (PP, MAP); SBP remained the strongest predictor, and performance estimates were directionally consistent across sensitivity checks. Variation of CV seeds and number of folds yielded stable qualitative results, but the numerical metrics were heterogeneous within bootstrap CI bounds. Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e displays model coefficients and feature importance values, highlighting systolic blood pressure as the dominant predictor.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel interpretability: coefficients and feature importances\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFeature\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLR coef (std)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLR OR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRF importance\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eGB importance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u0026minus;0.046\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.955\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.199\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (bin)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u0026minus;0.044\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.034\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSystolic pressure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.062\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e21.368\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.554\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.727\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePulse pressure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u0026minus;1.504\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.222\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.213\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.266\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003ePartial dependence plots in Fig.\u0026nbsp;17 illustrate the marginal effects of age and SBP on predicted hypertension probability.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFigure\u0026nbsp;17. Partial dependence plots for key predictors in the gradient boosting model\u003c/strong\u003e \u003cp\u003ePlots showing the marginal effect of SBP and age on predicted hypertension probability.\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis pilot study highlights the feasibility of applying machine-learning (ML) methods to predict hypertension and map its distribution using a few input variables, i.e., age and sex, as well as office blood-pressure (BP) readings, in a community-based population. A hypertension\u0026ensp;prevalence of 25.2% was observed, which is very consistent with the national estimates and that of sub-Saharan Africa, where adult hypertension prevalence of 23\u0026ndash;31% has been widely reported (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Prevalence increased sharply with advancing age, as expected from previous epidemiologic studies revealing that vascular stiffness and decreased arterial\u0026ensp;compliance mediate the age-associated rise in systolic pressure and hypertension risk (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003ePredictors and model performance\u003c/h2\u003e \u003cp\u003eSBP was the single most important variable in predicting hypertension in all models among all the variables examined. SBP alone achieved an AUC of 0.865, highlighting considerable discriminatory power for this parameter in isolation. This concurs with the global literature that SBP is the single most reliable predictor of cardiovascular risk, especially in low-resource settings where diastolic and biochemical covariates are not readily obtainable (\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe three machine-learning algorithms: penalised logistic regression (PLR), random forest (RF), and gradient boosting (GB), achieved high discrimination (AUC 0.941\u0026ndash;1.000) and good calibration (Brier scores\u0026thinsp;\u0026le;\u0026thinsp;0.094). Gradient\u0026ensp;boosting obtained a perfect discrimination on a small hold-out set, but this is likely due to overfitting given the small sample size. Previous large-scale ML studies have consistently suggested that ensemble models, particularly gradient boosting, provide better performance than logistic regression for hypertension and cardiovascular prediction tasks, but the magnitude of their performance advantage is considerably narrowed when the feature sets are small or strongly correlated (\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eModel interpretability and explainability\u003c/h3\u003e\n\u003cp\u003eExplainability analyses, including feature importance and partial dependence plots, consistently identified SBP as the single\u0026ensp;most important predictor, followed by pulse pressure and age as the next two most important features. These patterns are biologically and clinically plausible, as increased SBP and widened pulse pressure are markers of arterial stiffness, which is a key pathophysiologic determinant of hypertension risk (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). The consistency of feature importance across models indicates the robustness of SBP as a primary signal detecting ML-based hypertension detection in community data. SHAP summaries also confirmed these patterns, providing the transparency necessary for responsible ML implementation in health\u0026ensp;research (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003eComparison with related studies\u003c/h2\u003e \u003cp\u003eRecent ML-based hypertension prediction studies from sub-Saharan Africa have demonstrated similar findings, despite methodological heterogeneity. For example, Kimmel et al. (2023) conducted a cross-sectional machine-learning study in Ethiopia and reported an XGBoost model with an AUC of 0.894, an accuracy of 88.81%, a precision of 89.62%, a recall of 97.04%, and an F1-score of 93.18%, with age, weight/BMI, and related features among the top predictors (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). In China, Ji et al. (2022) employed a tree-based ensemble on adult sample data and identified age and SBP as the best predictors, with an XGBoost AUC of 0.893 (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). These correspondences emphasize the potential of parsimonious, interpretable ML models to improve\u0026ensp;hypertension screening in resource-limited settings, particularly when coupled with mobile or community health platforms (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003eStrengths and limitations\u003c/h2\u003e \u003cp\u003eStrengths of this study are rigorous data cleaning, a reproducible ML workflow with nested cross-validation, and the use of bootstrap confidence intervals to express model uncertainty. The transparent inclusion of calibration and discrimination indices aligns with\u0026ensp;the emerging reporting standards, such as TRIPOD-AI (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eA substantial contribution of this work is that it bases the predictive models solely on locally sourced data from Nigeria, which addresses a substantial gap in the global ML health landscape. The majority of hypertension prediction models to date have been developed using high-income country datasets, with very limited representations of African populations in training and validation pipelines (\u003cspan additionalcitationids=\"CR37\" citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). These geographic and demographic biases limit the generalizability of predictive models and risk exacerbating algorithmic inequities when applied in low- and middle-income countries. This study also reinforces the evidence for context-specific AI development, through training, testing, and validating models on local data, decolonising digital health research through inclusive data practices and equitable model design (\u003cspan additionalcitationids=\"CR40\" citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePotential overfitting, especially in the gradient\u0026ensp;boosting model, cannot be ruled out, and external validation on a larger independent dataset is necessary. Moreover, anthropometric\u0026ensp;and behavioural covariates (e.g., BMI, smoking, and alcohol use) were unavailable in the dataset, which might have enhanced model generalizability (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). Hence, given the small sample size and limited number of hypertensive cases, results should be interpreted as exploratory and hypothesis-generating.\u003c/p\u003e \u003cdiv id=\"Sec33\" class=\"Section3\"\u003e \u003ch2\u003eImplications and future directions\u003c/h2\u003e \u003cp\u003eThis study demonstrates the feasibility of ML-enabled community hypertension surveillance in Nigeria. Despite the small sample size, ML-driven models demonstrated strong predictive accuracy, indicating that properly designed models relying on cost-effective variables can be employed to support large-scale hypertension risk mapping. Future efforts should focus on increasing sample size, considering lifestyle and genetic predictors, and applying external validation. Integration with digital health platforms or mobile screening applications could potentially expand and facilitate early detection and inform community-based interventions (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eML methods, especially logistic regression and tree-based ensembles, achieved high discriminatory performance in predicting hypertension using simple demographic and blood-pressure data. Systolic blood pressure (SBP) was\u0026ensp;the most influential predictor of model outcomes. These results highlight the potential feasibility of data-driven hypertension monitoring in resource-poor environments and provide the basis for broader, external validation studies to guide public health policy-making in Nigeria and throughout sub-Saharan Africa.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e\u003cb\u003eAbbreviation\u003c/b\u003e\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e \u003cb\u003eFull form\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eArea under the receiver-operating characteristic curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAverage precision (area under the precision\u0026ndash;recall curve)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBody mass index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBlood pressure\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eConfidence interval\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCross-validation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDBP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDiastolic blood pressure\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEPV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eEvents per variable\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGB\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGradient boosting (ensemble machine-learning method)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGBM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGradient boosting machine (synonym of GB)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGLM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGeneralised linear model\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHBPM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eHome blood pressure monitoring\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIQR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eInterquartile range\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLOESS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLocally estimated scatterplot smoothing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLogistic regression (penalised logistic regression when specified)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMAP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMean arterial pressure (\u0026asymp;\u0026thinsp;DBP + (SBP\u0026thinsp;\u0026minus;\u0026thinsp;DBP)/3)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMBP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMean blood pressure (synonym for MAP in some texts)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eML\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMachine learning\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNPV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNegative predictive value\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNHREC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e National Health Research Ethics Committee (Nigeria)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNiMSA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNigerian Medical Students\u0026rsquo; Association\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eOR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eOdds ratio\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePDP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePartial dependence plot\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePulse pressure (SBP\u0026thinsp;\u0026minus;\u0026thinsp;DBP)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePPV\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePositive predictive value\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePrecision\u0026ndash;recall (curve)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRandom forest (ensemble tree-based model)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSBP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSystolic blood pressure\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eStandard deviation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSHAP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSHapley Additive exPlanations (model-agnostic explainability method)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTRIPOD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTransparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTRIPOD-AI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTRIPOD extension and guidance for prediction models using artificial intelligence / machine learning\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eWHO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eWorld Health Organization\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch3\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eEthical approval for this study was obtained from the Health Research Ethics Committee of the University of Nigeria Teaching Hospital (UNTH-HREC) (Protocol Number: NHREC/05/01/2008B-FW00002458-1RB00002327). All study procedures adhered to the ethical principles outlined in the Declaration of Helsinki (2013 revision).\u003c/p\u003e\n\u003cp\u003eInformed consent to participate was obtained from all individuals before data collection during the community outreach. Participation was voluntary, and all data were anonymized before analysis.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eNot applicable. All participants provided consent for the use of anonymized data for research and publication purposes.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eAll authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria, educational grants, participation in speakers\u0026rsquo; bureaus, membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe de-identified data that was used in\u0026ensp;this study is not available to the public due to community-level privacy considerations and limitations associated with the outreach program. The data may be made available by the corresponding author on reasonable request and with permission from the organizers of the Nkpokiti community outreach.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003eGodswill U (GU):\u003c/strong\u003e Conceptualization; Methodology; Data curation; Formal analysis; Software; Visualization; Investigation; Writing \u0026ndash; original draft (Methods, Results, Discussion); Writing \u0026ndash; review \u0026amp; editing.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eTreasure O (TO):\u003c/strong\u003e Conceptualization; Formal analysis; Writing \u0026ndash; original draft (Introduction, Abstract, Conclusion); Writing \u0026ndash; review \u0026amp; editing.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eWinnifred A (WNA):\u003c/strong\u003e Supervision; Project administration; Validation; Writing \u0026ndash; review \u0026amp; editing.\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe authors sincerely thank the health volunteers, especially the Nigerian Medical Students\u0026rsquo; Association (NiMSA) and University of Nigeria Medical Students\u0026rsquo; Association (UNMSA) Standing Committees on Policy Implementation (SCOPI), and Research Exchange (SCORE), for their participation and cooperation during data collection. We also appreciate the support of the outreach organizers and data entry team for their assistance in field coordination.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization. Guideline for the pharmacological treatment of hypertension in adults. Geneva: WHO. 2021. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.who.int/publications/i/item/9789240033986\u003c/span\u003e\u003cspan address=\"https://www.who.int/publications/i/item/9789240033986\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForouzanfar MH, Liu P, Roth GA, Ng M, Biryukov S, Marczak L et al. Global Burden of Hypertension and Systolic Blood Pressure of at Least 110 to 115 mm Hg, 1990\u0026ndash;2015. JAMA. 2017;317(2):165\u0026ndash;182. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2016.19043\u003c/span\u003e\u003cspan address=\"10.1001/jama.2016.19043\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Erratum in: JAMA. 2017;317(6):648. doi: 10.1001/jama.2017.0013. PMID: 28097354.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdeloye D, Basquill C. Estimating the prevalence and awareness rates of hypertension in Africa: a systematic analysis. PLoS ONE. 2014;9(8):e104300. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0104300\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0104300\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 25090232; PMCID: PMC4121276.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAkinlua JT, Meakin R, Umar AM, Freemantle N. Current Prevalence Pattern of Hypertension in Nigeria: A Systematic Review. PLoS ONE. 2015;10(10):e0140021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0140021\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0140021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 26461923; PMCID: PMC4603956.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeaney T, Burrell LM, Castillo RR, Charchar FJ, Cro S, Damasceno A et al. MMM Investigators. May Measurement Month 2018: a pragmatic global screening campaign to raise awareness of blood pressure by the International Society of Hypertension. Eur Heart J. 2019;40(25):2006\u0026ndash;2017. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/eurheartj/ehz300\u003c/span\u003e\u003cspan address=\"10.1093/eurheartj/ehz300\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Erratum in: Eur Heart J. 2019;40(37):3109. doi: 10.1093/eurheartj/ehz373. PMID: 31041440; PMCID: PMC6600128.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuntner P, Shimbo D, Carey RM, Charleston JB, Gaillard T, Misra S, et al. Measurement of Blood Pressure in Humans: A Scientific Statement From the American Heart Association. Hypertension. 2019;73(5):e35\u0026ndash;66. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/HYP.0000000000000087\u003c/span\u003e\u003cspan address=\"10.1161/HYP.0000000000000087\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 30827125; PMCID: PMC11409525.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeng SF, Reps J, Kai J, Garibaldi JM, Qureshi N. Can machine-learning improve cardiovascular risk prediction using routine clinical data? PLoS ONE. 2017;12(4):e0174944. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0174944\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0174944\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 28376093; PMCID: PMC5380334.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins GS, Moons KGM, Dhiman P, Riley RD, Beam AL, Van Calster B et al. TRIPOD\u0026thinsp;+\u0026thinsp;AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385:e078378. doi: 10.1136/bmj-2023-078378. Erratum in: BMJ. 2024;385:q902. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.q902\u003c/span\u003e\u003cspan address=\"10.1136/bmj.q902\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 38626948; PMCID: PMC11019967.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePickering TG, Hall JE, Appel LJ, Falkner BE, Graves JW, Hill MN, Council on High Blood Pressure Research Professional and Public Education Subcommittee, American Heart Association, et al. Recommendations for blood pressure measurement in humans: an AHA scientific statement from the Council on High Blood Pressure Research Professional and Public Education Subcommittee. J Clin Hypertens (Greenwich). 2005;7(2):102\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1524-6175.2005.04377.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1524-6175.2005.04377.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 15722655; PMCID: PMC8109470.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuyton AC, Hall JE. Textbook of Medical Physiology. 14th ed. Philadelphia: Elsevier; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShapiro SS, Wilk MB. An Analysis of Variance Test for Normality (Complete Samples). Biometrika. 1965;52(3/4):591\u0026ndash;611. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.jstor.org/stable/2333709\u003c/span\u003e\u003cspan address=\"https://www.jstor.org/stable/2333709\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eO\u0026rsquo;Brien RM. A caution regarding rules of thumb for variance inflation factors. Qual Quant. 2007;41(5):673\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKavakiotis I, Tsave O, Salifoglou A, Maglaveras N, Vlahavas I, Chouvarda I. Machine Learning and Data Mining Methods in Diabetes Research. Comput Struct Biotechnol J. 2017;15:104\u0026ndash;16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.csbj.2016.12.005\u003c/span\u003e\u003cspan address=\"10.1016/j.csbj.2016.12.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 28138367; PMCID: PMC5257026.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVarma S, Simon R. Bias in error estimation when using cross-validation for model selection. BMC Bioinformatics. 2006;7:91. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1471-2105-7-91\u003c/span\u003e\u003cspan address=\"10.1186/1471-2105-7-91\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 16504092; PMCID: PMC1397873.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCawley GC, Talbot NL. On over-fitting in model selection and subsequent selection bias in performance evaluation. J Mach Learn Res. 2010;11:2079\u0026ndash;107.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNCD Risk Factor Collaboration (NCD-RisC). Worldwide trends in blood pressure from 1975 to 2015: a pooled analysis of 1479 population-based measurement studies with 19\u0026middot;1 million participants. Lancet. 2017;389(10064):37\u0026ndash;55. doi: 10.1016/S0140-6736(16)31919-5. Epub 2016 Nov 16. Erratum in: Lancet. 2020;396(10255):886. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(20)31972-3\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(20)31972-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 27863813; PMCID: PMC5220163.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrier GW. Verification of forecasts expressed in terms of probability. Mon Weather Rev. 1950;78(1):1\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEfron B, Tibshirani RJ. An Introduction to the Bootstrap. Chapman and Hall/CRC; 1994.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundberg SM, Lee SI. A Unified Approach to Interpreting Model Predictions. Vol. 30, Neural Information Processing Systems. Curran Associates, Inc.; 2017. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://papers.nips.cc/paper_files/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html\u003c/span\u003e\u003cspan address=\"https://papers.nips.cc/paper_files/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12(85):2825\u0026ndash;30. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.jmlr.org/papers/v12/pedregosa11a.html\u003c/span\u003e\u003cspan address=\"https://www.jmlr.org/papers/v12/pedregosa11a.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNational Health Research Ethics Committee of Nigeria (NHREC). National Code for Health Research Ethics. Abuja: NHREC; 2007.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOjji DB, Baldridge AS, Orji IA, Shedul GL, Ojo TM, Ye J, Chopra A, et al. Hypertension Treatment in Nigeria Program Investigators. Characteristics, treatment, and control of hypertension in public primary healthcare centers in Nigeria: baseline results from the Hypertension Treatment in Nigeria Program. J Hypertens. 2022;40(5):888\u0026ndash;96. Epub 2022 Jan 15. PMID: 35034080; PMCID: PMC9081131.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFranklin SS, Gustin W 4th, Wong ND, Larson MG, Weber MA, Kannel WB, Levy D. Hemodynamic patterns of age-related changes in blood pressure. The Framingham Heart Study. Circulation. 1997;96(1):308\u0026thinsp;\u0026ndash;\u0026thinsp;15. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/01.cir.96.1.308\u003c/span\u003e\u003cspan address=\"10.1161/01.cir.96.1.308\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 9236450.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWhelton PK, Carey RM, Aronow WS, Casey DE Jr, Collins KJ, Dennison Himmelfarb C, College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. 2017 ACC/AHA/AAPA/ABC\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA\u003c/span\u003e\u003cspan address=\"http:///ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: Executive Summary: A Report of the American. Hypertension. 2018;71(6):1269\u0026ndash;1324. doi: 10.1161/HYP.0000000000000066. Epub 2017 Nov 13. Erratum in: Hypertension. 2018;71(6):e136-e139. doi: 10.1161/HYP.0000000000000075. Erratum in: Hypertension. 2018;72(3):e33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/HYP.0000000000000080\u003c/span\u003e\u003cspan address=\"10.1161/HYP.0000000000000080\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 29133354.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNCD Risk Factor Collaboration (NCD-RisC). Long-term and recent trends in hypertension awareness, treatment, and control in 12 high-income countries: an analysis of 123 nationally representative surveys. Lancet. 2019;394(10199):639\u0026ndash;51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(19)31145-6\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(19)31145-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2019 Jul 18. PMID: 31327564; PMCID: PMC6717084.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu T, Krentz A, Lu L, Curcin V. Machine learning based prediction models for cardiovascular disease risk using electronic health records data: systematic review and meta-analysis. Eur Heart J Digit Health. 2024;6(1):7\u0026ndash;22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ehjdh/ztae080\u003c/span\u003e\u003cspan address=\"10.1093/ehjdh/ztae080\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 39846062; PMCID: PMC11750195.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIslam SMS, Talukder A, Awal MA, Siddiqui MMU, Ahamad MM, Ahammed B et al. Machine Learning Approaches for Predicting Hypertension and Its Associated Factors Using Population-Level Data From Three South Asian Countries. Front Cardiovasc Med. 2022;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSilva GFS, Fagundes TP, Teixeira BC, Chiavegatto Filho ADP. Machine Learning for Hypertension Prediction: a Systematic Review. Curr Hypertens Rep. 2022;24(11):523\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11906-022-01212-6\u003c/span\u003e\u003cspan address=\"10.1007/s11906-022-01212-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2022 Jun 22. PMID: 35731335.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdeleke O, Adebayo S, Halleluyah Aworinde, Adeleke O, Adeniyi AE. Oluwasegun Julius Aroba. Machine learning evaluation of a hypertension screening program in a university workforce over five years. Sci Rep. 2024;14(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWaburi E, Muriithi D, Eugine Sundays. Integrating Explainable Machine Learning Models for Early Detection of Hypertension: A Transparent Approach to AI-Driven Healthcare. Am J Artif Intell. 2025;9(2):154\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSafar ME, Plante GE, Mimran A. Arterial stiffness, pulse pressure, and the kidney. Am J Hypertens. 2015;28(5):561\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ajh/hpu206\u003c/span\u003e\u003cspan address=\"10.1093/ajh/hpu206\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2014 Dec 4. PMID: 25480804.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLaurent S, Boutouyrie P. The structural factor of hypertension: large and small artery alterations. Circ Res. 2015;116(6):1007-21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/CIRCRESAHA.116.303596\u003c/span\u003e\u003cspan address=\"10.1161/CIRCRESAHA.116.303596\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 25767286.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKimeli N, Nkurunziza J, Juma K. Predicting health insurance uptake in Kenya using Random Forest: An analysis of socio-economic and demographic factors. PLoS ONE. 2023;18(11):e0294166\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJi W, Zhang Y, Cheng Y, Wang Y, Zhou Y. Development and validation of prediction models for hypertension risks: A cross-sectional study based on 4,287,407 participants. Front Cardiovasc Med. 2022;9:928948. PMID: 36225955; PMCID: PMC9548597.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSegun-Omosehin OA, Omiye JA, Elalfy A, Moideen-Sheriff S, Kuti F, Omole O et al. Evaluating the digital health technology landscape in sub-Saharan Africa and its implications for cardiovascular health. npj Cardiovascular Health. 2025;2(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nature.com/articles/s44325-025-00055-9\u003c/span\u003e\u003cspan address=\"https://www.nature.com/articles/s44325-025-00055-9\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWiens J, Saria S, Sendak M, Ghassemi M, Liu VX, Doshi-Velez F et al. Do no harm: a roadmap for responsible machine learning for health care. Nat Med. 2019;25(9):1337\u0026ndash;1340. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-019-0548-6\u003c/span\u003e\u003cspan address=\"10.1038/s41591-019-0548-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Epub 2019 Aug 19. Erratum in: Nat Med. 2019;25(10):1627. doi: 10.1038/s41591-019-0609-x. PMID: 31427808.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoseph J. Algorithmic bias in public health AI: a silent threat to equity in low-resource settings. Front Public Health. 2025;13:1643180. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fpubh.2025.1643180\u003c/span\u003e\u003cspan address=\"10.3389/fpubh.2025.1643180\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 40771228; PMCID: PMC12325396.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOwoyemi A, Owoyemi J, Osiyemi A, Boyd A. Artificial Intelligence for Healthcare in Africa. Front Digit Health. 2020;2:6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fdgth.2020.00006\u003c/span\u003e\u003cspan address=\"10.3389/fdgth.2020.00006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 34713019; PMCID: PMC8521850.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMugalula Kalule Grancia. Decolonizing AI ethics in Africa\u0026rsquo;s healthcare: An ethical perspective. AI and Ethics. 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNyamawe AS. Is the public sector Africa\u0026rsquo;s hidden force for AI-driven healthcare transformation? Telematics and Informatics Reports. 2025;20:100258. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.sciencedirect.com/science/article/pii/S2772503025000726\u003c/span\u003e\u003cspan address=\"https://www.sciencedirect.com/science/article/pii/S2772503025000726\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSkalidis I, Maurizi N, Salihu A, Fournier S, Cook S, Iglesias JF, et al. Artificial Intelligence and Advanced Digital Health for Hypertension: Evolving Tools for Precision Cardiovascular Care. Med (Kaunas). 2025;61(9):1597. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/medicina61091597\u003c/span\u003e\u003cspan address=\"10.3390/medicina61091597\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 41010987; PMCID: PMC12471829.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRen W, Fan K, Liu Z, Wu Y, An H, Liu H. Overcoming Missing Data: Accurately Predicting Cardiovascular Risk in Type 2 Diabetes, A Systematic Review. Journal of Diabetes. 2025 Jan [cited 2025 Jan 26];17(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC11753920/\u003c/span\u003e\u003cspan address=\"https://pmc.ncbi.nlm.nih.gov/articles/PMC11753920/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiang Y, Li S, Zhang P. An exploration in remote blood pressure management: Application of daily routine pattern based on mobile data in health management. Fundamental Res. 2022;2(1):154\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGafane-Matemane LF, Mokwatsi GG, Boateng D. Hypertension management in sub-Saharan Africa: an overview of challenges and opportunities for telemedicine. Connected Health. 2023;2(1):9\u0026ndash;22. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://f.oaes.cc/xmlpdf/6034f523-9888-4243-9985-6067ce2759dd/5499.pdf\u003c/span\u003e\u003cspan address=\"https://f.oaes.cc/xmlpdf/6034f523-9888-4243-9985-6067ce2759dd/5499.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8088844/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8088844/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eHypertension is often undiagnosed in\u0026ensp;many low-income countries. While machine learning (ML) may enhance triage during community screening, there is, however, limited evidence from outreach programs across African settings. This pilot\u0026ensp;investigated the prevalence and feasibility of ML prediction using minimal data collected locally in a Nigerian outreach.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eCross-sectional analysis of anonymized data from a community outreach in Nkpokiti, Enugu, was performed. Eligible records (n\u0026thinsp;=\u0026thinsp;115) included age, sex, and at\u0026ensp;least one paired systolic/diastolic blood-pressure (BP) measurement. Hypertension was also defined as mean\u0026ensp;SBP\u0026thinsp;\u0026ge;\u0026thinsp;140 mmHg and/or DBP\u0026thinsp;\u0026ge;\u0026thinsp;90 mmHg. Predictors were age, sex, first\u0026ensp;SBP, and pulse pressure. We trained penalized logistic regression (primary), random forest, and gradient-boosting models using nested 5-fold cross-validation for hyperparameter tuning; final illustrative results are reported on a stratified 80/20 hold-out. Discrimination (AUC), calibration (Brier score), and classification metrics were calculated with bootstrap confidence intervals.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eMedian age was 33 years\u0026ensp;(mean 36.4, SD 14.4); 71.3% were female. The prevalence of hypertension\u0026ensp;was 25.2% (29/115), increasing with age from 12.7% (\u0026lt;\u0026thinsp;40y) to 52.0% (40\u0026ndash;59y) and 54.5% (\u0026ge;\u0026thinsp;60y). SBP alone yielded an AUC of 0.865 (95%\u0026ensp;CI 0.777\u0026ndash;0.941). On the hold-out set (n\u0026thinsp;=\u0026thinsp;23; 6 positives), penalized logistic regression achieved an AUC of 0.941 (bootstrap mean 0.939; 95% CI 0.800\u0026ndash;1.000), accuracy of 0.783, and Brier score of 0.094. Random forest: AUC 0.961, accuracy 0.826, Brier 0.088. A Gradient boosting method showed perfect discrimination on this small hold-out set (AUC 1.000) with a Brier score of 0.038, probably reflecting optimistic estimation due to small sample size.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eIn this pilot study, ML models comprising age, sex, and simple BP measures demonstrated excellent discrimination for hypertension, supporting the feasibility of context-specific, low-cost triage tools. These findings are exploratory; external validation with larger, representative African samples is needed before deployment.\u003c/p\u003e","manuscriptTitle":"Pilot Study of Hypertension Screening and Machine-Learning Prediction Using Community Outreach Data from Nkpokiti, Enugu, Nigeria Short Title: Machine Learning Prediction of Hypertension using Community Blood Pressure Data in Nigeria","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-18 10:22:30","doi":"10.21203/rs.3.rs-8088844/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-12-12T18:17:29+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-12-08T15:54:39+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-11-19T14:23:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-18T16:39:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2025-11-18T16:36:05+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5cc7e9dd-5dbc-4549-acaf-875adb3f2e67","owner":[],"postedDate":"December 18th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-12-18T10:22:30+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-18 10:22:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8088844","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8088844","identity":"rs-8088844","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}