Predicting Hypertension Using Machine Learning
Author Information
Author(s): Li Yike, Higaki Akinori, Jeong Changwon, Hwang Seung Ha, Lee Hayeon, Lee Jun Hyuk, Lee Myeongcheol, Koyanagi Ai, Smith Lee, Rhee Sang Youl, Yon Dong Keon, Lee Jinseok
Primary Institution: Kyung Hee University
Hypothesis
Can machine learning models accurately predict the onset of hypertension based on health checkup data?
Conclusion
The study found that machine learning models, particularly an ensemble of AdaBoost and logistic regression, can effectively predict hypertension onset within five years using health checkup data.
Supporting Evidence
- The ensemble model achieved a balanced accuracy of 81.2% and an AUROC of 0.901.
- Key predictors of hypertension included age, diastolic blood pressure, BMI, systolic blood pressure, and fasting blood glucose.
- The model was validated using an independent dataset from Japan, confirming its predictive capabilities.
Takeaway
This study shows that computers can help doctors figure out who might get high blood pressure in the next few years by looking at their health checkup information.
Methodology
The study used data from two large national cohorts in South Korea and Japan, applying six machine learning models to predict hypertension based on various health indicators.
Potential Biases
Potential biases may arise from the exclusion criteria and the reliance on health checkup data.
Limitations
The study's findings may not be generalizable to all populations due to the specific demographics of the cohorts used.
Participant Demographics
Participants included adults aged 20 and older from South Korea and Japan, with a mix of urban and rural residents.
Statistical Information
P-Value
<0.001
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website