A machine learning-based tool for estimating individual hypertension risk, developed using data from the WHO Rwanda STEP Survey
# Hypertension Risk Predictor — Rwanda
A machine learning-based tool for estimating individual hypertension
risk, developed using data from the WHO Rwanda STEP Survey.
## Live Application
hypertension-risk-predictor…
## Overview
Users provide demographic, lifestyle, and health-history information,
with the option to add body measurements for a refined estimate. The
tool returns a personalized hypertension risk score accompanied by
tailored, evidence-based recommendations.
## Methodology
The underlying model employs monotonic constraints on clinically
directional variables to ensure predictions align with established
medical understanding (e.g., risk increases with age, elevated BMI,
and other known risk factors).
Where clinical measurements are not provided by the user, nine
variables (including BMI, waist circumference, heart rate, cholesterol,
glucose, HDL, sodium, creatinine, and waist-hip ratio) are estimated
using individually evidence-tested, demographically-restricted models.
This approach was developed specifically to prevent lifestyle or
behavioural responses from inadvertently influencing clinical estimates
in ways inconsistent with sound epidemiological practice. The pipeline
has been independently verified through systematic diagnostic testing
to confirm the absence of such effects across all nine variables.
## Disclaimer
This tool provides a risk estimate and associated recommendations. It
does not constitute a clinical diagnosis. Blood pressure measurement
by a qualified health professional is required for diagnostic purposes.