Prediction of hypertension using traditional regression and machine learning models: A systematic review and meta-analysis

Abstract

Objective We aimed to identify existing hypertension risk prediction models developed using traditional regression-based or machine learning approaches and compare their predictive performance. Methods We systematically searched MEDLINE, EMBASE, Web of Science, Scopus, and the grey literature for studies predicting the risk of hypertension among the general adult population. Summary statistics from the individual studies were the C-statistic, and a random-effects meta-analysis was used to obtain pooled estimates. The predictive performance of pooled estimates was compared between traditional regression-based models and machine learning-based models. The potential sources of heterogeneity were assessed using meta-regression, and study quality was assessed using the PROBAST (Prediction model Risk Of Bias ASsessment Tool) checklist. Results Of 14,778 articles, 52 articles were selected for systematic review and 32 for meta-analysis. The overall pooled C-statistics was 0.75 [0.73–0.77] for the traditional regression-based models and 0.76 [0.72–0.79] for the machine learning-based models. High heterogeneity in C-statistic was observed. The age (p = 0.011), and sex (p = 0.044) of the participants and the number of risk factors considered in the model (p = 0.001) were identified as a source of heterogeneity in traditional regression-based models. Conclusion We attempted to provide a comprehensive evaluation of hypertension risk prediction models. Many models with acceptable-to-good predictive performance were identified. Only a few models were externally validated, and the risk of bias and applicability was a concern in many studies. Overall discrimination was similar between models derived from traditional regression analysis and machine learning methods. More external validation and impact studies to implement the hypertension risk prediction model in clinical practice are required.

Can machine learning improve risk prediction of incident hypertension? An internal method comparison and external validation of the Framingham risk model using HUNT Study data

“…While the XGBoost model achieved the best mean AUC and Brier Score, the difference versus the much simpler elastic regression model was small, and far smaller than the variation induced by bootstrapping on the testing set. Comparing the AUC scores we achieved in this study versus a meta-analysis of the literature, we find that it is close to the expected mean AUC for a hypertension risk model (5)(6)(7). Relative to other studies, the small differences in discrimination between the elastic regression and the best performing machine learning model, is similar to that found in other studies (11,21,23).…”

“…In total, 23 722 individuals were found eligible for this study. The features available for our study are well-established risk factors of hypertension and CVD and commonly used in risk modelling of incident hypertension (7,39). We estimated physical activity by a novel physical activity metric, Personal Activity Intelligence (PAI).…”

“…When not restricting to a specific disease, a systematic review by Christodoulou et al found that logistic regression performed just as well as machine learning for clinical prediction models when limiting themselves to studies with low risk of bias (14). Considering the topic of hypertension risk models specifically, a recent systematic review identified and summarized the previous work (7). In that review, it was found that a large variation reported in discrimination performance was unrelated to the type of modelling algorithms being used.…”

Development and validation of a hypertension risk prediction model and construction of a risk score in a Canadian population

“…In this case, we won’t be able to compare the validated model’s predictive performance in our dataset. Considering the aforementioned factors and information from our recent systematic review 38 , we selected the models by Parikh et al 15 , Kivimӓki et al 39 , Lim et al 10 , Chien et al 14 , and Wang et al 12 for validation in our dataset. The model by Parikh et al 15 , also known as the Framingham Risk Score (FRS), was developed in the United States in a predominantly White population, with age, sex, systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, parental hypertension, cigarette smoking, and age by DBP as final variables.…”

Body composition predicts hypertension using machine learning methods: A Cohort Study

“…In the past few years, growing number of researchers have used machine learning and data mining algorithms to diagnose and treat health conditions such as heart (11) and brain (12) diseases. Their non-invasive nature and accuracy have enabled health professionals to quickly identify at-risk individuals and use more e cient preventive and managing strategies (13).…”