A Predictive Model For Public Behavior Data From Hypertensive Disorders Patient

Hypertension is one of the most common chronic diseases in the country and is often accompanied by a variety of complications that seriously affect the health of the nation.The factors involved in the development of hypertension are complex and are closely related to genetics,lifestyle and human activity.There have been many genomics,metabolomics and polytomies-based analyses examining the effects of genes and food intake on hypertension,but for human activity,current research has only focused on the quantitative analysis of specific exercise behaviors,and there is a lack of research using machine learning methods to learn the relationship between hypertension and exercise from accelerometer data.This is largely due to the fact that the dimensionality,length and data characteristics of the raw accelerometer data are not conducive to model training.the sequence mapping algorithm proposed by Paw and Wang effectively addresses this issue through a data encoding scheme.However,the activity intensity of the method is based on expert experience in partitioning a single dataset and cannot be adapted to different acceleration datasets with varying characteristics.Further research is therefore needed to develop methods that apply to different datasets and to seek more precise suggestions for activity patterns.In response to the above problems,the following work was carried out in this paper:1.The Hidden Semi Markov Model(HSMM)was used to optimize the sequence mapping method to automatically determine the best activity threshold,solving the problem that the activity threshold needed to be selected manually in the original method.At the same time,the accelerometer data were downscaled and coded,compressing 72,788,489 3D data into 25,639,159 1D codes,shortening to 35.23% of the original length.2.The activity combination module was proposed to extract the time series features and apply them to the hypertension prediction model.Compared with a single motion module,the activity combination module is more effective and can characterize complex activity intensity distribution patterns.And accordingly,the importance of the activity combination module with a width of 2 to 5 is found in the prediction of hypertension disease.Based on the above work,this study found that modules with lower activity levels were more likely to be risk factors for hypertension,while modules containing vigorous-intensity activity(VPA)were negatively associated with the risk of developing hypertension.Furthermore,sedentary interruptions alone may not prevent hypertension and only become protective factors against hypertensive disease when they contain VPA.Therefore,personal activity habits can be changed through tailored interventions to increase high levels of activity in the daily routine,which can have a positive effect on the prevention and control of hypertensive disease and reduce the risk of developing the disease.