Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 18, 2026PLoS ONEOpen Access

Machine learning and deep learning–based prediction of hypertension and analysis of its major risk factors in Bangladesh

View Full Paper
Ask AI
Bookmark
Share

Key result

Random forest models achieve ~69% recall for identifying hypertension while weighted logistic regression maximizes accuracy.

  • 95% CI 17.2-18.9
  • n=14,283

Why the study?

Hypertension is a leading cause of cardiovascular morbidity and mortality in Bangladesh, but its prevalence, risk factors, and predictive modeling using ML and DL approaches were not fully characterized.

Can machine learning and deep learning models accurately predict hypertension risk in adults in Bangladesh?

Population

14,283 adults (≥18 years) from the 2022 Bangladesh Demographic and Health Survey

Design

Cross-sectional analysis

Authors

SCShawrab ChandraMMM. M. Imran MollaSISamiul Islam

Discussion

Loading...

Member takes

Overview

ML models may aid hypertension identification in Bangladesh; hypothesis-generating and requires prospective validation before clinical use.

Key Points

  • To determine the prevalence and sociodemographic risk factors of hypertension in Bangladesh and evaluate the predictive performance of machine learning and deep learning models.
  • Analyzed cross-sectional survey data from the 2022 Bangladesh Demographic and Health Survey comprising N=14,283 adults (≥18 years).
  • Trained and evaluated four machine learning algorithms (weighted logistic regression, random forest, extreme gradient boosting, light gradient boosting machine) and two deep learning architectures (TabNet and multi-layer perceptron).
  • Assessed predictive accuracy using precision, recall, specificity, F1 score, AUC-ROC, and AUC-PR.
  • Overall hypertension prevalence was 18.04% (95% CI: 17.2%–18.9%), with higher rates observed in women (18.87%) than men (16.97%).
  • Weighted logistic regression achieved the highest accuracy (0.817), specificity (0.981), precision (0.444), and AUC-ROC (0.751), but demonstrated low recall (0.070).
  • Random forest demonstrated the highest recall (0.687) and F1-score (0.460), identifying age, body mass index, sex, family size, and education level as the most important predictors.

Study Design

Type

Cross-Sectional (n=14,283)

Structured PICO

Can machine learning and deep learning models accurately predict hypertension risk in adults in Bangladesh?

P
Population
14,283 adults (≥18 years) from the 2022 Bangladesh Demographic and Health Survey analyzed to estimate hypertension prevalence and evaluate machine learning prediction models.
E
Exposure
Machine learning (weighted logistic regression, random forest, extreme gradient boosting, light gradient boosting machine) and deep learning (TabNet, multi-layer perceptron) models
O
Outcome
Prediction of hypertension risk (evaluated by accuracy, precision, recall, specificity, F1 score, AUC-ROC, and AUC-PR)

Machine learning models, particularly random forest, show promise in identifying individuals at risk for hypertension in Bangladesh, though external validation is needed.

Limitations

  • Weighted logistic regression exhibited low recall, limiting its utility for identifying individuals with hypertension
  • Further external validation and assessment of clinical utility are required before implementation
  • Requires further external validation
  • Requires assessment of clinical utility before implementation

Cite This Study

Chandra et al. (2026) conducted a cross-sectional in Hypertension (n=14,283). Socio-demographic risk factors was evaluated on Hypertension prevalence (95% CI 17.2-18.9). Hypertension prevalence in Bangladesh was 18.04%, and while weighted logistic regression had the highest accuracy, random forest models achieved the highest recall (0.687) for identification.

synapsesocial.com/papers/6aacf5ed0c46fbdff987dd3chttps://doi.org/10.1371/journal.pone.0358471
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Machine‐Learning–Based Prediction of Hypertension and Its Risk Factors Among Adults in the Northern Region of Bangladesh2026 · 2 citations
  2. 2Application of machine learning approaches to develop predictive models for diabetes and hypertension among Bangladeshi Adults2026 · 1 citations
  3. 3Machine Learning Approaches for Predicting Hypertension and Its Associated Factors Using Population-Level Data From Three South Asian Countries2022 · 104 citations
  4. 4Machine learning approaches to identify influential factors associated with hypertension and prehypertension among rural adults in Bangladesh2026
  5. 5Machine learning approaches in identifying factors associated with hypertension and undiagnosed hypertension in adults in rural areas of Bangladesh2026 · 1 citations