Application of Cloud Computing in the Prediction of Exercise Improvement of Cardiovascular and Digestive Systems in Obese Patients.

Based on the cardiovascular and digestive problems of obese patients, this paper adopted the cloud computing method and selected 100 subjects with big data (23 normal weight subjects, 3740 overweight patients, and 40 obese patients) as the research objects, studying the heart configuration and their digestive system of obese people. Results show that BMI = L (24 ≥ BMI > 27.9) and BMI = XL (BMI > 27.9) were identified as target correlation projects in this experiment, associated with each cardiac structural parameters, respectively. Cloud computing facilitates early detection, early prevention, and early intervention in heart configuration changes in overweight and obese patients.

1. Introduction

With the improvement of living standard and the change of lifestyle, overweight and obese patients are common in the whole country and even the world. Statistics show that in most European and American countries, more than half of the population is overweight or obese. In China, the number of obese patients has also shown an obvious rising trend in the past 20 years. Overweight and obesity are caused by excessive accumulation of body fat, especially triglycerides [1]. In overweight and obese patients, lipids can accumulate in the skin, viscera, tissues, and even blood. Excessive lipids will increase the oxygen consumption of tissues and the burden of viscera, damage organ function, and affect organism metabolism. It will lead to obstacles in coordination between external environment, organs, and tissues, breaks the stability of the organism's internal environment, and leads to various concurrent diseases. Foreign studies have shown that overweight and obesity are associated with hypertension, coronary heart disease, T2DM, dyslipidemia, stroke, and other diseases [2]. Mild obesity may cause mild changes in the structure and function of liver and cardiovascular system, but there are no obvious clinical symptoms. Patients with moderate and severe obesity have changes in the cardiovascular system, digestive system, and even joint system and neuropsychiatric system and also have obvious clinical symptoms, such as palpitation, shortness of breath, discomfort in the liver area, joint pain, fatigue, and anxiety [3].

As an important organ of human body, the heart plays a central role in the process of continuing life. With the gradual accumulation of fat in the body and the increasing weight, the cardiac blood output and circulating blood volume increased, and ventricular contraction intensified. A series of adaptive changes such as increased wall tension and decreased systemic vascular resistance will happen. For the changes of visceral configuration and function, we should detect and prevent the cardiovascular diseases as early as possible. In evaluating the influence of overweight and obesity on cardiac structure and function, echocardiography has the advantages of economy and noninvasiveness. It is one of the main detection technologies at present [4]. It is found that overweight and obese patients have larger left atrioventricular cavity diameter, thicker wall, and increased left ventricular mass index in different degrees compared with healthy people with normal weight.

Zheng et al. studied the relationship between obesity and left ventricular hypertrophy and found that the former was an independent influencing factor of the latter, and the heavier the obesity, the higher the incidence of cardiovascular disease [5]. In addition, the study found that overweight fat mainly changes with the magnitude of the right room chamber, thickening of the right room, and the degree of change and BMI. Weight is getting more and more important. Most studies have shown that the E-peak (2-point diastolic blood flow velocity) of overweight or obesity is reduced. A-peak increased (mitral diastolic flow velocity), while E/A ratio decreased. In addition, there was an extension of IVRT, and an extension of the invasive trend of obesity was observed. Kang, SH, and other studies found that with the increase of body mass index, RV scholar peak velocity EM (early motion velocity of three-point loop stroke), Sm (peak motion velocity of three-point loop retreat), and strain rate decreased significantly, which proved the severity of right ventricular dysfunction in patients with BMI and overweight and obesity [6]. For adults, as people grow older, the basal metabolic rate will decline. If people's diet does not decrease and consumption is less than intake, it will lead to gaining weight [7]. Proper exercise can increase consumption, thus inhibiting the increase of fat. American researchers have found through experiments that the cause of obesity is related to ghrelin and gastrointestinal kinin in human body. Experiments have found that ghrelin in normal people has regular ups and downs, while ghrelin in obese people has a horizontal trend. Obese people have a level of ghrelin for a day, and their hunger is not obvious. When obese people eat, ghrelin will also be at a certain level. However, the gastrointestinal kinin (stomach satiety) of obese people showed a downward trend, which means that obese people did not feel satiety, which explained why obese people often complained that they were not full [8, 9]. According to the investigation, the earliest semi-automatic segmentation method based on MR image is threshold segmentation algorithm. This algorithm distinguishes fat tissue pixels from non-fat tissue pixels by artificially determined threshold but cannot distinguish visceral fat from subcutaneous fat. The algorithm is greatly affected by noise because the thresholds of fat signal and non-fat signal in MR image are not global. Therefore, there is sometimes a big error in segmenting fat and non-fat regions according to threshold. It is found that the performance of abdominal fat, especially visceral fat, on MR images is very close to that of non-fat tissues. Typical automatic segmentation methods include fuzzy c-means clustering algorithm, snake algorithm, algorithm-based on fuzzy clustering and level set, and algorithm based on traditional K-means [10, 11].

In this study, association rules (Apriori algorithm) were used to analyze the cardiac configuration parameters and clinical information of overweight and obesity in DM patients and find out some characteristic indicators to reflect the actual changes of cardiac configuration and related influencing factors, so as to provide guidance for early detection and early prevention of cardiac allocation in overweight and obese patients. Exercise can not only prevent obesity and reduce fat but also prevent some complications caused by obesity.

2. Research Methods

Characteristics of cardiac structural changes in overweight and obese patients: the mechanism of cardiac structural changes in overweight and obese patients is the decrease of systemic vascular resistance and the adaptive increase of cardiac output, circulating blood volume, and ventricular wall tension in order to adapt to the state of lipid accumulation and weight rise; this can lead to increased effort in the left atrial retraction, and sustained heavy volume load can eventually lead to enlargement of the left atrium.

2.1. Objects and Methods

According to the diagnostic criteria proposed in the recommendations of the Chinese Working Group on Obesity, the subjects had a BMI of 18.5–24 kg/m2 as normal weight, a BMI of 24–27.9 kg/m2 as overweight, and a BMI of 27.9 kg/m2 as obese. A total of 100 overweight and obese patients (80 cases) and subjects with normal weight were selected as the objects of this study; among them, there were 50 males, 50 females, 23 normal weight subjects, 37 overweight subjects, and 40 obese subjects. Patients with congenital heart disease, valvular heart disease, cardiomyopathy, and other heart diseases that cause serious changes in heart configuration were excluded. Patients with chronic diseases such as diabetes and hypertension that have been proved to have an impact on heart configuration were also excluded [12, 13].

The instruments and analysis tools are Philips E33 and EPIC 7C color Doppler echocardiography (Philips products in the Netherlands), s5-1 probe, and the probe frequency is 3.5–5 MHz. The ultrasonic workstation is equipped with QLAB quantitative analysis software and height and weight meter. The user data analysis tool is Microsoft Excel 2007, and the data mining tool is WEKA Explorer.

2.2. Research Methods

Data collection and establishment of cardiac configuration database: the name, gender, blood pressure, heart rate, blood glucose, blood lipid, and other clinical information of subjects are from the hospital information and laboratory system. Each subject was asked about the course of obesity, and their height and weight were strictly measured and recorded [14]. The ultrasonic data of heart configuration were obtained according to the standardized mapping standard, such as the left atrial diameter (LAD), left ventricular diameter (LVD), left atrial area (LAA), left atrial volume (LAV), and left ventricle myocardial weight (LVM), at the end of systolic period. The obtained clinical information and cardiac configuration data of the subjects were constructed into a cardiac configuration database.

Data preprocessing includes data cleaning, integration, and conversion. As one of the most important steps in preprocessing, the discretization process is to convert the continuous values into a form suitable for association mining. The range of normal values is the reference range that we currently diagnose in normal people, the discrete value of age is divided into Y (<30 years old), M (30–60 years old), and O (>60 years old), and the discrete value of LAD is divided into S (<23 mm), M (23–38 mm), and L (>38 mm) (which can be modified according to the needs of mining targets in the process of data mining) (see Table 1 for specific discrete processing).

Table 1

Discretization of clinical and cardiac routine measurements.

Discrete values	The original attribute values	Discrete values	The original attribute values	Discrete values	The original attribute values
Age-Y	<30 old	Age-M	30–60 old	Age-O	>60 old
Fat year-S	<10 years	Fat year-N	10–20 years	Fat year-L	>20 years
BMI-N	<24	BMI-L	24–27.9 kg/m2	BMI-XL	>27.9 kg/m2
LAD-S	<23 mm	LAD-N	23–38 mm	LAD-L	>38 mm
LAA-S	<8 m2	LAA-N	8–17 m2	LAA-L	>17 m2
LAV-S	<18 ml	LAV-N	18–56 ml	LAV-L	>56 ml
LVM-S	<60 g	LAA-M	60–158 g	LAA-L	>158 g
...

Association rules (Apriori algorithm): association rules can be expressed in the form of M = =>M, where the former is the comparison item and the latter is the result. The strength of association rules is measured by confidence and support. Confidence indicates the frequency of M occurring in transactions containing N. Support indicates how often M and N occur together in the entire transaction library [15, 16]. Confidence represents the credibility of the association rule, while support represents the importance of the rule. Association rule process: (1) set the minimum support and confidence (in this experiment, the support is tentatively set as 0.1 and the confidence is set as 0.5); (2) find frequent item sets (scan and find the item sets greater than the minimum support); and (3) extract effective association rules (find the item set greater than the minimum confidence in frequent item set).

The core of the algorithm is as follows.

Perform a uniform sampling on the image and divide the image into several rectangular areas according to distance, that is, as the initial superpixels; the side length of each initial superpixel is expressed as follows:in which n is the total pixel quantity of the image and k is the expected average pixel quantity of each superpixel.

To prevent the center of superpixels from falling on the edge pixels of the image, the point with the lowest gradient near the center of gravity of each superpixel (such as 3 ∗3 neighborhood) is selected as the initial center of each superpixel [17-19]. The center of each superpixel is expressed as C=[v, x, y], where v corresponds to the gray value of the average pixel of the superpixel and x, y correspond to the spatial coordinate position information of the center of the superpixel. Their calculation formulas are as follows:in which v corresponds to the gray value of a single pixel; x, y represent spatial target position information of the corresponding pixel; and J is the number of pixels contained in the current superpixel [20].

3. Research Results

3.1. Data Preprocessing Results

After the preprocessing the original data, such as cleaning, integration, and discretization, the data recording table can be directly mined after being converted into a transaction database, as shown in Table 2.

Table 2

Cardiac configuration transaction of overweight and obese patients.

The project ID	Project
1	Sex = female, age = M, fat year = L, BMI = L, ..., LAD = L, LAA = L,
2	LVD = N, LVM = N, ...
3	Sex = male, age = O, fat year = L, BMI = XL, ...
...	...

3.2. Establishment and Extraction Results of Association Rules

In this study, overweight and obese patients were mainly targeted. Firstly, BMI = L and BMI = XL were set as the target association items, and the association items were established with each index association item of heart architecture one by one, and the mining results of association rules were obtained, as shown in Table 3.

Table 3

Effective association rules for cardiac architecture indicators in overweight and obese patients.

Serial number	Association rules	Rules of interpretation
1	BMI = L = =>LAD = L < conf:(0.60)>	24 ≥ BMI>27.9 = =>anteroposterior diameter of left atrium at end of contraction>38 mm has a confidence of 60%, support ≥10%
2	BMI = XL = =>LAD = L < conf:(0.73)>	BMI > 27.9 = =>anteroposterior diameter of left atrium at end of contraction>38 mm has a confidence of 73%, support ≥10%
3	BMI = L = =>LAA = L < conf:(0.69)>	24 ≥ BMI > 27.9 = =>left atrial area at end of contraction > the confidence level of 17 mm2 is 69%, support ≥10%
4	BMI = XL = =>LAA = L < conf:(0.80)>	BMI > 27.9 = =>left atrial area at end of contraction > the confidence level of 17 mm2 is 80%, support ≥10%
5	BMI = XL = =>LAV = L < conf:(0.70)>	BMI > 27.9 = =>Left atrial volume at end of contraction>56 ML has a confidence of 70%, support ≥10%
6	BMI = XL = =>LVM = L < conf:(0.63)	BMI > 27.9 = =>Left ventricular myocardial weight at end of contraction>158G has a confidence rating of 63%, support ≥10%
...		...

As can be seen from Table 3, BMI = L and BMI = XL of overweight and obese patients form strong association rules with LAD = L, LAA = L, LAV = L, and LVM = L; therefore, the characteristic indexes of cardiac structural changes in overweight and obese patients were LAD, LAA, LAV, and LVM.

The parameters are set as 5 × 20 × 20 initial superpixels and 18 fan-shaped regions. The test results are shown in Figure 1, in which 10 datasets from 100 are numbered 1–10, respectively, datasets 1–5 are from overweight people, and datasets 6–10 are from normal people.

Figure 1

Segmentation results of different test datasets.

3.3. Discussion

Association rule is one of the classical algorithms of data mining, which has been applied in many fields such as medicine. Its functions include the following: (1) to examine the long-established data knowledge model in the DM industry and (2) discover hidden new associations between data. Association rules are applied to establish association in medical big data, extract effective rules, and find the hidden association information clock in medical data, which is conducive to disease diagnosis. The application of association rules in medical field has been explored and achieved by experts at home and abroad.

Characteristics of cardiac structural changes in overweight and obese patients: the mechanism of cardiac structural changes in overweight and obese patients is the decrease of systemic vascular resistance and the adaptive increase of cardiac output, circulating blood volume, and ventricular wall tension in order to adapt to the state of lipid accumulation and weight rise; this can lead to increased effort in the left atrial gyrus, and sustained heavy volume load can eventually lead to enlargement of the left atrium.

BMI = L (24 ≥ BMI > 27.9) and BMI = XL (BMI > 27.9) were identified as target association items in this study and were associated with cardiac structural parameters, respectively. From Table 3, effective association rule numbers 1 and 2 were extracted, and it can be seen that BMI = L, BMI = XL, and LAD = L established strong association rules; the rule indicated that the LAD of overweight and obese patients tended to increase at the end of systolic period with high reliability (No. 1 is 60% and No. 2 is 71%); the correlation results obtained by the same number (3, 4, 5, 6, and so on) showed that the LAA, LAV, and LVM of overweight and obese patients tended to increase, which also had a high credibility. Based on the correlation results in Table 3, LAD, LAA, LAV, and LVM at the end of systolic period can establish a strong correlation with BMI = L and BMI = XL target related items. It can be seen that in overweight and obese patients, the early changes of heart structure are mainly the size of left atrium and LVM. In this study, the changes of left atrial size and LVM in overweight and obese patients were consistent with previous studies. However, as the study was at the preliminary stage of association rule study with a small number of samples, no strong association rule was found between BMI and changes in ventricular wall thickness and left ventricular diameter; this is not consistent with the results of some previous studies, but it may be that association rules are more advantageous and sensitive than other statistical methods in finding the characteristic indicators of cardiac structural changes. The influencing factors of cardiac structural changes in overweight and obese patients are the early prevention and intervention of cardiac structural changes in overweight and obese patients. It is necessary to further use association rules to find the influencing factors of cardiac structural changes. Therefore, in this experiment, the characteristic indicators identified in Table 3, such as left atrial diameter and LAA, were correlated with the basic clinical information of subjects one by one, and effective association rules were extracted. When fat year <10 years, LAA tends to be normal and has high credibility (79%), and when fat year >20 years, LAA tends to increase credibility (70%); therefore, fat year is another important factor affecting cardiac structural changes.

4. Conclusions

Sports not only prevent obesity but also prevent some complications of complications in obesity by reducing fat. Exercise can not only reduce fat but also prevent the occurrence of chronic diseases and improve physical quality and social adaptability. In addition, exercise can relieve stress to make people feel happy, improve people's respiratory system, and improve cardiopulmonary function. For obese adults, some low-intensity exercises can achieve a good effect. Most of the exercise projects are slow, and the strength can be controlled around 40%, so that the duration is relatively long, and it can achieve a good effect. It is possible to formulate a sports prescription to obese people. It is based on the child's age and physical condition.