Research On Early Warning Of Hypertension Risk And Zero-load Exercise Intervention Methods

Hypertension is a major public health problem worldwide.It is the most important risk factor for cardiovascular and cerebrovascular diseases and it causes premature death and heavy economic burden.High prevalence and low control rate are the two main characteristics of hypertension in China.Exercise is a basic way to prevent and control hypertension,and its effectiveness has been confirmed by numerous studies.There are two issues in carrying out exercise intervention for hypertension at the grassroots level in China.Firstly,how to accurately identify high-risk groups of hypertension and achieve "pre-intervention and prevention first";Secondly,how to achieve effective control of exercise intensity in a large number of people who are not accustomed to wearable devices.Researchers have proposed a series of early warning models and explored a variety of exercise intensity control methods.However,there are still some difficulties in current research,such as low sensitivity,complex collection of predictors,and long warning cycles are existed in current early warning models.Meanwhile,there is a lack of simple and effective methods for accurate control of exercise intensity.Aiming at the problem of low sensitivity of the early warning model of hypertension,an improved algorithm framework is proposed to improve the accuracy and adaptability of the hypertension early warning model.At the same time,a machine learning algorithm that integrates influencing factors exploration and model construction is put forward.The algorithm is utilized to construct a zero-load exercise intensity control method that is based on individual health indicators,and then the effectiveness of the new method is verified through field tests.Finally,a certain number of individuals at high risk of hypertension are screened through the proposed early warning model of hypertension,and then zero-load exercise intervention is carried out on them for several months to study the dose-effect relationship between zero-load exercise intervention and the reduction of hypertension risk.The main work and innovations of this dissertation are summarized as follows.1.Research on hypertension risk assessment method based on easy-to-collect risk factorsA hypertension risk assessment method combining univariate logistic regression analysis,optimized random forest,is proposed.Univariate logistic regression analysis is utilized to accurately identify the key risk factors for hypertension.A random forest algorithm with adaptive hyperparameters that can automatically match the optimal hyperparameters and can improve the performance of the model is developed.By calculating the contribution of each risk factor to the performance of the model,a method for analyzing the importance of features is proposed,to make the assessment model with more interpretability.The innovation of this method can effectively integrate and utilize easy-to-collect risk factors containing lifestyle information,and can automatically extract features that significant to hypertension,so as to achieve an accurate and interpretable assessment of the current risk of hypertension.The experimental results on clinical data show that the proposed risk assessment method can effectively improve the prediction accuracy and adaptability,and is suitable for the secondary prevention of hypertension in the grassroots community.2.Research on the framework for predicting the risk of hypertension in the next yearA framework for predicting the risk of hypertension in the next year based on XGBoost algorithm is proposed.By calculating and analyzing the Euclidean distance between different labeled samples,a noise data removal algorithm for labeled data is proposed to improve the quality of the original data.A feature selection method combining univariate logistic regression and variable importance analysis is developed to accurately identify the risk factors of onset of hypertension in the next year.Furthermore,the hyperparameters of the XGBoost classifier are automatically adjusted in combination with Bayesian theory,and finally the optimized XGBoost classifier is used to build a prediction model for the risk of onset of hypertension in the coming year.Comparative experiments on clinical data sets verify the effectiveness and robustness of the proposed algorithm framework.The research provides methods for screening high-risk groups in the primary prevention of hypertension in grassroots communities.3.Research on the method of zero-load exercise intensity control based on health indicatorsA machine learning algorithm is proposed to explore the influencing factors of the relationship between heart rate(HR)and rating of perceived exertion(RPE),and to construct a conversion model between HR and RPE.Taking into account the possible influence of demographics,anthropometrics,body composition,cardiovascular function,basic exercise ability and other health indicators on the relationship between RPE and HR,forward search method is utilized to explore influencing factors.Furthermore,Gaussian Process Regression(GPR)is used to construct a conversion model between HR and RPE.Through the collection and research of a certain number of cycle ergometer exercise data,indicators of age,RHR,CAP,BFR and BMI are found to be the influencing factors of the relationship between HR and RPE.Comparative experiments showed that the proposed method improves the accuracy of using RPE instead of HR to control exercise intensity.Based on the model,the method of zero-load exercise intensity control is realized using RPE instead of HR.Meanwhile,this is the first study of the effectiveness of the zero-load exercise intensity control method in the free exercise of large-scale populations.Based on this,a tracking study of exercise intervention has been carried out in the general healthy population.The above research provides methods and empirical support for the implementation of exercise intervention for high-risk groups of hypertension.4.Early identification and exercise intervention for people at high risk of hypertensionIn order to study the dose-effect relationship of exercise intervention on the risk reduction of hypertension and to verify the effectiveness of the constructed early warning model and zero-load intervention method,63 high-risk groups(age:24-69 years old,average age:41 years)are screened out of the routine physical examination population in Sanxiaokou Community of Hefei City,this used the constructed risk prediction model for the onset of hypertension in the next year.Then,they undergo a three-month zero-load exercise intervention.The results show that after the exercise intervention,the health indicators of men and women have improved in different degrees.The corresponding blood pressure and hypertension risk factors have been significantly improved,and the risk of hypertension has been significantly reduced(P<0.001).The results of the study revealed that the proposed hypertension early warning method can effectively screen out the high-risk groups of hypertension among the grassroots residents,and the zero-load exercise intervention can improve the risk factors of hypertension and effectively reduce the risk of hypertension.