Why the study?
Effective hypertension management is critical to decrease uncontrolled hypertension, and early identification of patients at risk would enable targeted, proactive treatments.
Can machine learning models using electronic health records data better predict the risk of uncontrolled hypertension compared to using the last blood pressure measure alone?
Can machine learning models using electronic health records data better predict the risk of uncontrolled hypertension compared to using the last blood pressure measure alone?
Machine learning models, including simple logistic regression, can outperform single blood pressure measurements in predicting near-term risk of uncontrolled hypertension using electronic health records data.
Answers are grounded in this paper and cite their sources.
Machine learning models may aid early risk stratification for uncontrolled hypertension; leaves open prospective validation before clinical adoption.
Hypertension is a major risk factor for stroke, cardiovascular disease, and-stage renal disease, and its prevalence is expected to rise dramatically. hypertension management is thus critical. A particular priority is the incidence of uncontrolled hypertension. Early identification of at risk for uncontrolled hypertension would allow targeted use of, proactive treatments. We develop machine learning models(logistic regression and recurrent neural networks) to stratify patients with to the risk of exhibiting uncontrolled hypertension within the coming-month period. We trained and tested models using EHR data from 14,407 and3,009 patients, respectively. The best model achieved an AUROC of 0.719, the simple, competitive baseline of relying prediction based on last BP measure alone (0.634). Perhaps surprisingly, recurrent neural did not outperform a simple logistic regression for this task, that linear models should be included as strong baselines for tasks using EHR
No takes yet. Share an insight, caveat, or question.
Mohammadi et al. (2019) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: