Why the study?
The 2017 AHA/ACC guidelines lowered blood pressure cut-offs for hypertension, but the effect of these changes on supervised learning prediction of cardiovascular disease was unknown.
Does using continuous systolic blood pressure improve the prediction of cardiovascular disease compared to using binned hypertension guidelines (AHA/ACC or JNC7) in machine learning models?
Does using continuous systolic blood pressure improve the prediction of cardiovascular disease compared to using binned hypertension guidelines (AHA/ACC or JNC7) in machine learning models?
Dichotomizing blood pressure using AHA/ACC or JNC7 guidelines reduces the accuracy of machine learning models in predicting cardiovascular disease compared to using continuous systolic blood pressure.
Answers are grounded in this paper and cite their sources.
Continuous SBP may enhance ML CVD prediction over guideline bins; leaves open prospective validation before any practice change.
The criteria used to categorize patients as either hypertensive or normotensive were changed in 2017 by the American Heart Association and the American College of Cardiology (AHA/ACC). The updated guidelines lowered the criteria by which individuals are classified as hypertensive; systolic blood pressure (SBP) cut-off from ≥140 mmHg to ≥130 mmHg and diastolic blood pressure from ≥90 mmHg to ≥80 mmHg. The purpose of this study was to investigate what effect these changes in diagnostic criteria had on the ability of supervised learning to predict cardiovascular disease. Three models were developed and tested. Two models using a binned hypertension measure based on either the AHA/ACC new released guidelines or the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC7) original guidelines. The third model used SBP as a continuous variable. Data from 68,657 patients was processed through decision tree algorithm to determine which model offered the best accuracy. For both female and male subjects, the model with SBP returned the best area under the receiver operator characteristic curve and overall better sensitivity and specificity values. Our results showed that changing the criteria by which individuals are classified as hypertensive or normotensive negatively impacted the ability of decision tree to predict cardiovascular disease in both females and males.
No takes yet. Share an insight, caveat, or question.
Bopp et al. (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: