Prediction of Masked Hypertension and Masked Uncontrolled Hypertension Using Machine Learning

Application of artificial intelligence in hypertension

Hypertension is an important modifiable risk factor for morbidity and mortality associated with cardiovascular disease. The incidence of hypertension is increasing not only in Korea but also in many Western countries due to the aging of the population and the increase in unhealthy lifestyles. However, hypertension control rates remain low due to poor adherence to antihypertensive medications, low awareness of hypertension, and numerous factors that contribute to hypertension, including diet, environment, lifestyle, obesity, and genetics. Because artificial intelligence (AI) involves data-driven algorithms, AI is an asset to understanding chronic diseases that are influenced by multiple factors, such as hypertension. Although several hypertension studies using AI have been published recently, most are exploratory descriptive studies that are often difficult for clinicians to understand and have little clinical relevance. This review aims to provide a clinician-centered perspective on AI by showing recent studies on the relevance of AI for patients with hypertension. The review is organized into sections on blood pressure measurement and hypertension diagnosis, prognosis, and management.

Predicting hypertension control using machine learning

Hypertension is a widely prevalent disease and uncontrolled hypertension predisposes affected individuals to severe adverse effects. Though the importance of controlling hypertension is clear, the multitude of therapeutic regimens and patient factors that affect the success of blood pressure control makes it difficult to predict the likelihood to predict whether a patient's blood pressure will be controlled. This project endeavors to investigate whether machine learning can accurately predict the control of a patient's hypertension within 12 months of a clinical encounter. To build the machine learning model, a retrospective review of the electronic medical records of 350,008 patients 18 years of age and older between January 1, 2015 and June 1, 2022 was performed to form model training and testing cohorts. The data included in the model included medication combinations, patient laboratory values, vital sign measurements, comorbidities, healthcare encounters, and demographic information. The mean age of the patient population was 65.6 years with 161,283 (46.1%) men and 275,001 (78.6%) white. A sliding time window of data was used to both prohibit data leakage from training sets to test sets and to maximize model performance. This sliding window resulted in using the study data to create 287 predictive models each using 2 years of training data and one week of testing data for a total study duration of five and a half years. Model performance was combined across all models. The primary outcome, prediction of blood pressure control within 12 months demonstrated an area under the curve of 0.76 (95% confidence interval; 0.75-0.76), sensitivity of 61.52% (61.0-62.03%), specificity of 75.69% (75.25-76.13%), positive predictive value of 67.75% (67.51-67.99%), and negative predictive value of 70.49% (70.32-70.66%). An AUC of 0.756 is considered to be moderately good for machine learning models. While the accuracy of this model is promising, it is impossible to state with certainty the clinical relevancy of any clinical support ML model without deploying it in a clinical setting and studying its impact on health outcomes. By also incorporating uncertainty analysis for every prediction, the authors believe that this approach offers the best-known solution to predicting hypertension control and that machine learning may be able to improve the accuracy of hypertension control predictions using patient information already available in the electronic health record. This method can serve as a foundation with further research to strengthen the model accuracy and to help determine clinical relevance.

Recent developments in machine learning modeling methods for hypertension treatment

Hypertension is the leading cause of cardiovascular complications. This review focuses on the advancements in medical artificial intelligence (AI) models aimed at individualized treatment for hypertension, with particular emphasis on the approach to time-series big data on blood pressure and the development of interpretable medical AI models. The digitalization of daily blood pressure records and the downsizing of measurement devices enable the accumulation and utilization of time-series data. As mainstream blood pressure data shift from snapshots to time series, the clinical significance of blood pressure variability will be clarified. The time-series blood pressure prediction model demonstrated the capability to forecast blood pressure variabilities with a reasonable degree of accuracy for up to four weeks in advance. In recent years, various explainable AI techniques have been proposed for different purposes of model interpretation. It is essential to select the appropriate technique based on the clinical aspects; for example, actionable path-planning techniques can present individualized intervention plans to efficiently improve outcomes such as hypertension. Despite considerable progress in this field, challenges remain, such as the need for the prospective validation of AI-driven interventions and the development of comprehensive systems that integrate multiple AI methods. Future research should focus on addressing these challenges and refining the AI models to ensure their practical applicability in real-world clinical settings. Furthermore, the implementation of interdisciplinary collaborations among AI experts, clinicians, and healthcare providers are crucial to further optimizing and validate AI-driven solutions for hypertension management.

Building a predictive model for hypertension related to environmental chemicals using machine learning

Hypertension is a chronic cardiovascular disease characterized by elevated blood pressure that can lead to a number of complications. There is evidence that the numerous environmental substances to which humans are exposed facilitate the emergence of diseases. In this work, we sought to investigate the relationship between exposure to environmental contaminants and hypertension as well as the predictive value of such exposures. The National Health and Nutrition Survey (NHANES) provided us with the information we needed (2005-2012). A total of 4492 participants were included in our study, and we incorporated more common environmental chemicals and covariates by feature selection followed by regularized network analysis. Then, we applied various machine learning (ML) methods, such as extreme gradient boosting (XGBoost), random forest classifier (RF), logistic regression (LR), multilayer perceptron (MLP), and support vector machine (SVM), to predict hypertension by chemical exposure. Finally, SHapley Additive exPlanations (SHAP) were further applied to interpret the features. After the initial feature screening, we included a total of 29 variables (including 21 chemicals) for ML. The areas under the curve (AUCs) of the five ML models XGBoost, RF, LR, MLP, and SVM were 0.729, 0.723, 0.721, 0.730, and 0.731, respectively. Butylparaben (BUP), propylparaben (PPB), and 9-hydroxyfluorene (P17) were the three factors in the prediction model with the highest SHAP values. Comparing five ML models, we found that environmental exposure may play an important role in hypertension. The assessment of important chemical exposure parameters lays the groundwork for more targeted therapies, and the optimized ML models are likely to predict hypertension.

Correlation analysis of hypertension, traditional Chinese medicine constitution, and LPL gene polymorphism in the elderly in communities in Shanghai

BACKGROUND

Research on the genetic mechanisms of hypertension has been a hot topic in the cardiovascular field.

OBJECTIVE

To study the correlation between senile hypertension and traditional Chinese medicine (TCM) constitution and lipoprotein lipase (LPL) gene polymorphism and to provide the theoretical basis for TCM prevention and treatment of hypertension.

METHODS

The elderly population in communities in Shanghai (hypertensive: 264 cases; non-hypertensive: 159 cases) was taken as the research object. Essential data and information on TCM constitution were collected. The LPL gene mutation was detected using the second-generation sequencing method. Statistical analysis was performed to clarify the relationship between hypertension and senile hypertension. The correlation of TCM constitution with risk factors and LPL gene polymorphisms was studied.

RESULTS

The primary TCM constitutions in the hypertension group were phlegm-dampness constitution (51.52%), yin-deficiency constitution (17.42%), balanced constitution (15.53%), and yin-deficiency (9.43%). Logistic regression analysis showed that the phlegm-dampness constitution (P< 0.05, OR = 2.587) and yin-deficiency constitution (P< 0.01, OR = 2.693) were the risk constitutions of hypertension in the elderly. A total of 37 LPL gene mutation loci (SNP: 22; new discovery: 15) were detected in the LPL gene, and the mutation rates of rs254, rs255, rs3208305, rs316, rs11570891, rs328, rs11570893, and rs13702 were relatively high, which were 26.24%, 26.24%, 16.08%, 14.66%, 13.24%, 12.06%, and 10.64%. In the phlegm-dampness group, the proportion of rs254 CC type, rs255 TT type, and rs13702 TT type in the hypertensive group (77.21%, 77.21%, and 93.38%) was higher than that in the non-hypertensive group (56.41%, 56.41%, and 82.05%), The difference was statistically significant (P< 0.05).

CONCLUSION

The phlegm-dampness constitution and yin-deficiency constitution are the risk factors of hypertension in the elderly; in the phlegm-dampness population, rs254 CC type, rs255 TT type, and rs13702 TT type are the risk factors for elderly hypertension.

Shapely additive values can effectively visualize pertinent covariates in machine learning when predicting hypertension

Machine learning methods are widely used within the medical field to enhance prediction. However, little is known about the reliability and efficacy of these models to predict long-term medical outcomes such as blood pressure using lifestyle factors, such as diet. The authors assessed whether machine-learning techniques could accurately predict hypertension risk using nutritional information. A cross-sectional study using data from the National Health and Nutrition Examination Survey (NHANES) between January 2017 and March 2020. XGBoost was used as the machine-learning model of choice in this study due to its increased performance relative to other common methods within medical studies. Model prediction metrics (e.g., AUROC, Balanced Accuracy) were used to measure overall model efficacy, covariate Gain statistics (percentage each covariate contributes to the overall prediction) and SHapely Additive exPlanations (SHAP, method to visualize each covariate) were used to provide explanations to machine-learning output and increase the transparency of this otherwise cryptic method. Of a total of 9650 eligible patients, the mean age was 41.02 (SD = 22.16), 4792 (50%) males, 4858 (50%) female, 3407 (35%) White patients, 2567 (27%) Black patients, 2108 (22%) Hispanic patients, and 981 (10%) Asian patients. From evaluation of model gain statistics, age was found to be the single strongest predictor of hypertension, with a gain of 53.1%. Additionally, demographic factors such as poverty and Black race were also strong predictors of hypertension, with gain of 4.33% and 4.18%, respectively. Nutritional Covariates contributed 37% to the overall prediction: Sodium, Caffeine, Potassium, and Alcohol intake being significantly represented within the model. Machine Learning can be used to predict hypertension.

Development of a convenient and effective hypertension risk prediction model and exploration of the relationship between Serum Ferritin and Hypertension Risk: a study based on NHANES 2017-March 2020

Background

Hypertension is a major public health problem, and its resulting other cardiovascular diseases are the leading cause of death worldwide. In this study, we constructed a convenient and high-performance hypertension risk prediction model to assist in clinical diagnosis and explore other important influencing factors.

Methods

We included 8,073 people from NHANES (2017-March 2020), using their 120 features to form the original dataset. After data pre-processing, we removed several redundant features through LASSO regression and correlation analysis. Thirteen commonly used machine learning methods were used to construct prediction models, and then, the methods with better performance were coupled with recursive feature elimination to determine the optimal feature subset. After data balancing through SMOTE, we integrated these better-performing learners to construct a fusion model based for predicting hypertension risk on stacking strategy. In addition, to explore the relationship between serum ferritin and the risk of hypertension, we performed a univariate analysis and divided it into four level groups (Q1 to Q4) by quartiles, with the lowest level group (Q1) as the reference, and performed multiple logistic regression analysis and trend analysis.

Results

The optimal feature subsets were: age, BMI, waist, SBP, DBP, Cre, UACR, serum ferritin, HbA1C, and doctors recommend reducing salt intake. Compared to other machine learning models, the constructed fusion model showed better predictive performance with precision, accuracy, recall, F1 value and AUC of 0.871, 0.873, 0.871, 0.869 and 0.966, respectively. For the analysis of the relationship between serum ferritin and hypertension, after controlling for all co-variates, OR and 95% CI from Q2 to Q4, compared to Q1, were 1.396 (1.176-1.658), 1.499 (1.254-1.791), and 1.645 (1.360-1.989), respectively, with P < 0.01 and P for trend <0.001.

Conclusion

The hypertension risk prediction model developed in this study is efficient in predicting hypertension with only 10 low-cost and easily accessible features, which is cost-effective in assisting clinical diagnosis. We also found a trend correlation between serum ferritin levels and the risk of hypertension.

Masked Hypertension in Healthy Children and Adolescents: Who Should Be Screened?

PURPOSE OF REVIEW

The goal is to review masked hypertension (MH) as a relatively new phenomenon when patients have normal office BP but elevated out-of-office BP. Firstly, it was described in children in 2004. It has received increased attention in the past decade.

RECENT FINDINGS

The prevalence of MH in different pediatric populations differs widely between 0 and 60% based on the population studied, definition of MH, or method of out-of-office BP measurement. The highest prevalence of MH has been demonstrated in children with chronic kidney disease (CKD), obesity, diabetes, and after heart transplantation. In healthy children but with risk factors for hypertension such as prematurity, overweight/obesity, diabetes, chronic kidney disease, or positive family history of hypertension, the prevalence of MH is 9%. In healthy children without risk factors for hypertension, the prevalence of MH is very low ranging 0-3%. In healthy children, only patients with the following clinical conditions should be screened for MH: high-normal/elevated office BP, positive family history of hypertension, and those referred for suspected hypertension who have normal office BP in the secondary/tertiary center.

Accuracy of screening strategies for masked hypertension: a large-scale nationwide study based on home blood pressure monitoring

This study compared the ability of guideline-proposed office blood pressure (OBP) screening thresholds [European Society of Hypertension (ESH) guidelines: 130/85 mmHg for individuals with an OBP < 140/90 mmHg; American College of Cardiology/American Heart Association (ACC/AHA) guidelines: 120/75 mmHg for individuals with an OBP < 130/80 mmHg] and novel screening scores to identify normotensive individuals at high risk of having masked hypertension (MH) in an office setting. We cross-sectionally evaluated untreated participants with an OBP < 140/90 mmHg (n = 22,266) and an OBP < 130/80 mmHg (n = 10,005) who underwent home blood pressure monitoring (HBPM) (derivation cohort) from 686 Brazilian sites. MH was defined according to criteria suggested by the ESH (OBP < 140/90 mmHg; HBPM ≥ 135/85 mmHg), Brazilian Society of Cardiology (BSC) (OBP < 140/90 mmHg; HBPM ≥ 130/80 mmHg) and ACC/AHA (OBP < 130/80 mmHg; HBPM ≥ 130/80 mmHg). Scores were generated from multivariable logistic regression coefficients between MH and clinical variables (OBP, age, sex, and BMI). Considering the ESH, BSC, and ACC/AHA criteria, 17.2%, 38.5%, and 21.2% of the participants had MH, respectively. Guideline-proposed OBP screening thresholds yielded area under curve (AUC) values of 0.640 (for ESH criteria), 0.641 (for BSC criteria), and 0.619 (for ACC/AHA criteria) for predicting MH, while scores presented as continuous variables or quartiles yielded AUC values of 0.700 and 0.688 (for ESH criteria), 0.720 and 0.709 (for BSC criteria), and 0.671 and 0.661 (for ACC/AHA criteria), respectively. Further analyses performed with alternative untreated participants (validation cohort; n = 2807 with an OBP < 140/90 mmHg; n = 1269 with an OBP < 130/80 mmHg) yielded similar AUC values. In conclusion, the accuracy of guideline-proposed OBP screening thresholds in identifying individuals at high risk of having MH in an office setting is limited and is inferior to that yielded by scores derived from simple clinical variables.

Prediction of Masked Uncontrolled Hypertension Detected by Ambulatory Blood Pressure Monitoring

The aim of this study was to provide prediction models for masked uncontrolled hypertension (MUCH) detected by ambulatory blood pressure (BP) monitoring in an Italian population. We studied 738 treated hypertensive patients with normal clinic BPs classified as having controlled hypertension (CH) or MUCH if their daytime BP was < or ≥135/85 mmHg regardless of nighttime BP, respectively, or CH or MUCH if their 24-h BP was < or ≥130/80 mmHg regardless of daytime or nighttime BP, respectively. We detected 215 (29%) and 275 (37%) patients with MUCH using daytime and 24-h BP thresholds, respectively. Multivariate logistic regression analysis showed that males, those with a smoking habit, left ventricular hypertrophy (LVH), and a clinic systolic BP between 130−139 mmHg and/or clinic diastolic BP between 85−89 mmHg were associated with MUCH. The area under the receiver operating characteristic curve showed good accuracy at 0.78 (95% CI 0.75−0.81, p < 0.0001) and 0.77 (95% CI 0.73−0.80, p < 0.0001) for MUCH defined by daytime and 24 h BP, respectively. Internal validation suggested a good predictive performance of the models. Males, those with a smoking habit, LVH, and high-normal clinic BP are indicators of MUCH and models including these factors provide good diagnostic accuracy in identifying this ambulatory BP phenotype.

Nomogram based on clinical features at a single outpatient visit to predict masked hypertension and masked uncontrolled hypertension: A study of diagnostic accuracy

Patients with masked hypertension (MH) and masked uncontrolled hypertension (MUCH) are easily overlooked, and both cause target organ damage. We propose a prediction model for MH and MUCH patients based on clinical features at a single outpatient visit. Data collection was planned before the index test and reference standard were after. Thus, we retrospectively collect analyzed 804 subjects who underwent ambulatory blood pressure monitoring (ABPM) at Renmin Hospital of Wuhan University. These patients were divided into normotension/controlled hypertension group (n = 121), MH/MUCH (n = 347), and sustained hypertension (SH)/sustained uncontrolled hypertension group (SUCH) (n = 302) for baseline characteristic analysis. Models were constructed by logistic regression, a nomogram was visualized, and internal validation by bootstrapping. All groups were performed according to the definition proposed by the Chinese Hypertension Association. Compared with normotension/controlled hypertension, patients with MH/MUCH had higher office blood pressure (BP) and were more likely to have poor liver and kidney function, metabolic disorder and myocardial damage. By analysis, [office systolic blood pressure (OSBP)] (P = .004) and [office diastolic blood pressure (ODBP)] (P = .007) were independent predictors of MH and MUCH. By logistic regression backward stepping method, office BP, body mass index (BMI), total cholesterol (Tch), high-density lipoprotein cholesterol (HDL-C), and left ventricular mass index are contained in this model [area under curve (AUC) = 0.755] and its mean absolute error is 0.015. Therefore, the prediction model established by the clinical characteristics or relevant data obtained from a single outpatient clinic can accurately predict MH and MUCH.

Blood pressure response to standing is a strong determinant of masked hypertension in young to middle-age individuals

OBJECTIVE

The pathophysiologic mechanisms of masked hypertension are still debated. The aim of this study was to investigate whether the blood pressure response to standing is a determinant of masked hypertension in young individuals.

DESIGN AND METHODS

We studied 1078 individuals (mean age 33.2 ± 8.5 years) with stage-1 untreated hypertension at baseline. Orthostatic response was defined as the difference between six SBP measurements in the orthostatic and supine postures. People with a response more than 6.5 mmHg (upper decile) were defined as hyperreactors. After 3 months of follow-up, 24-h ambulatory BP was measured and the participants were classified as normotensives (N = 120), white-coat hypertensive individuals (N = 168), masked hypertensive individuals (N = 166) and sustained hypertensive individuals (N = 624). In 591 participants, 24-h urinary epinephrine was also measured.

RESULTS

Orthostatic response was an independent predictor of masked hypertension after 3 months (P = 0.001). In the whole group, the odds ratio for the Hyperreactors was 2.5 [95% confidence interval (95% CI) 1.5-4.0, P < 0.001]. In the participants stratified by orthostatic response and urinary epinephrine, the odds ratio for masked hypertension was 4.2 (95% CI, 1.8-9.9, P = 0.001) in the hyperreactors with epinephrine above the median and was 2.6 (95% CI, 0.9-7.3, P = 0.069) in those with epinephrine below the median. The association between orthostatic response and masked hypertension was confirmed in the cross-sectional analysis after 3 months (P < 0.001).

CONCLUSION

The present findings indicate that hyperreactivity to standing is a significant determinant of masked hypertension. The odds ratio for masked hypertension was even quadrupled in people with an orthostatic response more than 6.5 mmHg and high urinary epinephrine suggesting a role of sympathoadrenergic activity in the pathogenesis of masked hypertension.

Identification of nine signature proteins involved in periodontitis by integrated analysis of TMT proteomics and transcriptomics

Recently, there are many researches on signature molecules of periodontitis derived from different periodontal tissues to determine the disease occurrence and development, and deepen the understanding of this complex disease. Among them, a variety of omics techniques have been utilized to analyze periodontitis pathology and progression. However, few accurate signature molecules are known and available. Herein, we aimed to screened and identified signature molecules suitable for distinguishing periodontitis patients using machine learning models by integrated analysis of TMT proteomics and transcriptomics with the purpose of finding novel prediction or diagnosis targets. Differential protein profiles, functional enrichment analysis, and protein–protein interaction network analysis were conducted based on TMT proteomics of 15 gingival tissues from healthy and periodontitis patients. DEPs correlating with periodontitis were screened using LASSO regression. We constructed a new diagnostic model using an artificial neural network (ANN) and verified its efficacy based on periodontitis transcriptomics datasets (GSE10334 and GSE16134). Western blotting validated expression levels of hub DEPs. TMT proteomics revealed 5658 proteins and 115 DEPs, and the 115 DEPs are closely related to inflammation and immune activity. Nine hub DEPs were screened by LASSO, and the ANN model distinguished healthy from periodontitis patients. The model showed satisfactory classification ability for both training (AUC=0.972) and validation (AUC=0.881) cohorts by ROC analysis. Expression levels of the 9 hub DEPs were validated and consistent with TMT proteomics quantitation. Our work reveals that nine hub DEPs in gingival tissues are closely related to the occurrence and progression of periodontitis and are potential signature molecules involved in periodontitis.

Advancing Prediction of Risk of Intraoperative Massive Blood Transfusion in Liver Transplantation With Machine Learning Models. A Multicenter Retrospective Study

Background

Liver transplantation surgery is often accompanied by massive blood loss and massive transfusion (MT), while MT can cause many serious complications related to high mortality. Therefore, there is an urgent need for a model that can predict the demand for MT to reduce the waste of blood resources and improve the prognosis of patients.

Objective

To develop a model for predicting intraoperative massive blood transfusion in liver transplantation surgery based on machine learning algorithms.

Methods

A total of 1,239 patients who underwent liver transplantation surgery in three large grade lll-A general hospitals of China from March 2014 to November 2021 were included and analyzed. A total of 1193 cases were randomly divided into the training set (70%) and test set (30%), and 46 cases were prospectively collected as a validation set. The outcome of this study was an intraoperative massive blood transfusion. A total of 27 candidate risk factors were collected, and recursive feature elimination (RFE) was used to select key features based on the Categorical Boosting (CatBoost) model. A total of ten machine learning models were built, among which the three best performing models and the traditional logistic regression (LR) method were prospectively verified in the validation set. The Area Under the Receiver Operating Characteristic Curve (AUROC) was used for model performance evaluation. The Shapley additive explanation value was applied to explain the complex ensemble learning models.

Results

Fifteen key variables were screened out, including age, weight, hemoglobin, platelets, white blood cells count, activated partial thromboplastin time, prothrombin time, thrombin time, direct bilirubin, aspartate aminotransferase, total protein, albumin, globulin, creatinine, urea. Among all algorithms, the predictive performance of the CatBoost model (AUROC: 0.810) was the best. In the prospective validation cohort, LR performed far less well than other algorithms.

Conclusion

A prediction model for massive blood transfusion in liver transplantation surgery was successfully established based on the CatBoost algorithm, and a certain degree of generalization verification is carried out in the validation set. The model may be superior to the traditional LR model and other algorithms, and it can more accurately predict the risk of massive blood transfusions and guide clinical decision-making.