Physical examination data significance analysis method based on multiple examination correction

Abstract

The invention belongs to the field of medicine, and particularly relates to a physical examination data significance analysis method based on multiple examination correction. The method comprises the following steps: firstly, carrying out transverse and longitudinal cleaning on physical examination data, then, carrying out classified analysis on the data, calculating the proportion of each disease in different classifications according to a set threshold value, finally, carrying out significance test on a statistical analysis result by utilizing a chi-square test method, and carrying out correction by utilizing a Bonferroni multiple comparison correction method. Saliency analysis is carried out through multiple comparison and correction methods of physical examination data, various physical examination indexes can be synthesized, the physical health condition level of a user can be described objectively, theoretical guidance is provided for health guidance of a doctor on a physical examination person and a physical examination item method for accurately positioning the patient, and meanwhile, the physical examination accuracy is improved. The communication cost between the physical examination user and the doctor is reduced, and the user is urged to pay more attention to the overall health condition level.

Description

Technical Field

[0001] The invention belongs to the medical field, and in particular relates to a significance analysis method of physical examination data based on multiple test correction. Background Art

[0002] With the advent of the big data era, the informatization of the medical industry has also entered its big data era. The main driving force of medical development has always been to solve human health problems. The emergence of artificial intelligence technology has greatly reduced the heavy burden of doctors' work and improved the efficiency of medical services. Using existing outpatient data can help doctors diagnose daily diseases, effectively reduce the workload of medical staff, improve diagnostic accuracy, provide computer information support for the work of modern medical staff, promote the improvement of daily disease management, establish a rapid disease consultation system, and provide automatic medical consultation to some extent. Analyze the logical relationship between test data and disease symptoms, which makes it easy to extract the hidden characteristics of the disease from the data. In addition, the data and conventional indicators of different diseases are compared and calculated and analyzed, which can clearly determine the correspondence between data layers, and then use pathology to test new disease warning indicators.

[0003] In recent years, with the rapid development of health and medical big data platforms, more and more physical examination data have been integrated into big data platforms. How to mine and use massive health and medical data to improve the quality of medical services and enhance the level of communication between doctors and patients is a new challenge. Through physical examination data analysis, physical examination users can intuitively perceive the level of their personal physical health status from a macro perspective, providing a quantitative indicator for reference for long-term health management. However, there are currently several problems: When performing significance analysis of physical examination data, you may encounter the following common problems: (1) Insufficient sample size: If the sample size is insufficient, the results may not be statistically significant enough and an accurate conclusion of significance cannot be drawn; (2) Skewed data distribution: If the analyzed data has skew or outliers, it may affect the results of the significance analysis and require data cleaning or non-parametric analysis; (3) Multiple comparison problems: When comparing multiple groups or multiple indicators, if appropriate multiple comparison corrections are not performed, misjudgment of significance may occur; (4) Use the correct statistical method: Using inappropriate or incorrect statistical methods for analysis may distort the results. Make sure to select statistical methods that are suitable for the data type and research question and apply them correctly; (5) Ignoring the relationship between variables: The significance analysis of physical examination data should consider the relationship between multiple variables, rather than just analyzing each variable independently. Ignoring the relationship between variables may lead to incorrect conclusions, etc. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for analyzing the significance of physical examination data based on multiple test correction. By performing significance analysis on the multiple comparison correction method of physical examination data, various physical examination indicators can be integrated to more objectively describe the user's physical health status, and provide theoretical guidance for the doctor's health guidance to the examinee and the method for accurately locating the patient's physical examination items. At the same time, it reduces the communication cost between the physical examination user and the doctor, and urges the user to pay more attention to the overall health status of the body. The method has high application value, is simple and feasible.

[0005] To achieve the above object, the present invention proposes a method for significance analysis of physical examination data based on multiple test correction, which specifically includes the following steps: S1. Clean the physical examination data horizontally and vertically. First, label the data, with negative as 0, positive as 1, and no test as 3. Further clean the data, which can be divided into horizontal and vertical processing methods. The horizontal processing method is: if a row is all 3, it means that the person has not undergone a physical examination, delete this row and do not participate in the calculation. The vertical processing method is: if a column is all 3, it means that all people have undergone this examination, delete this column to reduce the amount of data. If there are no null values ​​or NaN in the data set, all null values ​​can be converted to 3, and the 3 values ​​can be deleted in the data.

[0006] S2. Perform data classification analysis by gender, age, and department to preliminarily obtain the data distribution of the physical examination results corresponding to different classifications, and highlight the outliers. Draw statistical graphs based on the classifications. When there are outliers or outliers in the statistical graphs, they usually attract attention, which helps detect anomalies in the data. At the same time, statistical analysis can compare the influence of different variables or factors to observe the relationship and differences between variables.

[0007] S3. Set the threshold of high-proportion diseases to 0.3, calculate the proportion of each disease in different categories of gender, age, and department, and conduct statistical analysis on high-proportion diseases. Divide the number of positive people by the total number of people to get the proportion of positive people. According to the proportion of each disease, the distribution proportion of the disease in the population can be obtained. Multiple permutation and comparison statistics are helpful for multi-dimensional data presentation, and cross-comparison can be used to show the differences and relationships between multiple variables.

[0008] S4. The chi-square test method is used to test the significance of the statistical analysis results. The real meaning of the significance test is to apply probability theory to calculate the probability of obtaining the observed difference in test results if the two test results are actually equally effective. Its purpose is to evaluate how strong the evidence is that one test result is truly better than another. The strength of this evidence is quantified by probability, that is, P value. Therefore, the smaller the P value, the smaller the possibility that the difference in test results occurs by chance. In practice, people often use P < 0.05, P < 0.01, and P < 0.001 to represent the results of significance tests. The selection of these levels is completely arbitrary and there is no mathematical or clinical reason. In order to control the false discovery rate (especially type (false positive)), reduce false discoveries in the study, maintain the level of statistical significance, and obtain more reliable research results, the Bonferroni multiple comparison correction method is used for correction. The Bonferroni multiple comparison correction method can solve the problem of excessively high statistical significance levels that may result from multiple hypothesis tests. Through correction, it is accurately determined whether there is a significant association between the disease and gender, age, and department. Beneficial Effects

[0009] The present invention performs significance analysis on the multiple comparison correction method of physical examination data, which can comprehensively describe the user's physical health status level more objectively, provide theoretical guidance for doctors to provide health guidance to physical examinees and accurately locate physical examination items for patients, and at the same time, reduce the communication cost between physical examination users and doctors, and urge users to pay more attention to the overall physical health status level. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings and appendices required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings and appendices can be obtained based on these drawings without paying creative labor.

[0011] Figure 1 It is a method flow chart (technical route); Figure 2 This is a statistical chart of diseases with high incidence rates classified by gender; Figure 3 This is a statistical chart of diseases with high incidence by age classification; Figure 4 This is a statistical chart of diseases with high proportions classified by department; Table 1 shows the significance test results after correction by Bonferroni multiple comparisons correction method after classification by gender.

[0012] Table 2 shows the significance test results after correction by Bonferroni multiple comparisons method after classification by age.

[0013] Table 3 shows the significance test results after classification by department using the Bonferroni multiple comparison correction method. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. Example 1

[0016] The present invention uses 157480 sets of physical examination data (2540 physical examination users and 62 physical examination items) to propose a physical examination data significance analysis method based on multiple test correction, which specifically includes the following steps: Figure 1 As shown: S1. First, label the data, with negative as 0, positive as 1, and no detection as 3. Further data cleaning is divided into two processing methods: horizontal and vertical. The horizontal processing method is: if a row is all 3, it means that the person has not undergone a physical examination, delete this row and do not participate in the calculation. The vertical processing method is: if a column is all 3, it means that all people have undergone this examination, delete this column to reduce the amount of data. If there are no null values ​​or NaN in the data set, all null values ​​can be converted to 3, and the 3 values ​​can be deleted in the data. After cleaning, invalid data accounts for about 5%, and valid data accounts for about 95%.

[0017] S2. Classify and analyze the data by gender, age, and department, and preliminarily obtain the data distribution of the physical examination results corresponding to different categories, and highlight the abnormal values. Draw a statistical chart based on the classification, such as Figure 2-4 When there are outliers or outliers in a statistical graph, they usually attract attention, which helps detect abnormalities in the data. At the same time, statistical analysis can compare the influence of multiple different variables or factors to observe the relationship and differences between variables.

[0018] S3. Set the threshold of high-proportion diseases to 0.3, calculate the proportion of each disease in different categories of gender, age, and department, and conduct statistical analysis on high-proportion diseases. Divide the number of positive people by the total number of people to get the proportion of positive people. According to the proportion of each disease, the distribution proportion of the disease in the population can be obtained. Multiple permutation and comparison statistics are helpful for multi-dimensional data presentation, and cross-comparison can be used to show the differences and relationships between multiple variables. According to the statistical analysis of the data, the following conclusions are drawn: Classification by gender: (1) The main diseases of men are concentrated in "refractive error, obesity (mild), hyperlipidemia, fatty liver, hyperuricemia, abnormal liver function indexes, abnormal kidney function indexes", among which abnormal liver function indexes accounted for the highest proportion. (2) The main diseases of women are concentrated in "refractive error, abnormal electrocardiogram, hyperlipidemia, abnormal liver function indexes, thyroid nodules, breast hyperplasia", among which abnormal liver function indexes and breast hyperplasia accounted for the highest proportion.

[0019] By age: (1) The main diseases of people aged 20-30 years old are concentrated in "refractive error, hyperlipidemia, hyperuricemia, abnormal liver function indicators, breast hyperplasia, abnormal kidney function indicators", among which abnormal liver function accounts for the highest proportion. (2) The main diseases of people aged 30-40 years old are concentrated in "refractive error, obesity, hyperlipidemia, fatty liver, hyperuricemia, abnormal liver function indicators, breast hyperplasia, abnormal kidney function indicators", among which abnormal liver function accounts for the highest proportion. (3) The main diseases of people aged 40-50 years old are concentrated in "refractive error, abnormal electrocardiogram, obesity, hyperlipidemia, fatty liver, hyperuricemia, abnormal liver function indicators, thyroid nodules, abnormal kidney function indicators", among which abnormal liver function and hyperlipidemia account for the highest proportion. (4) The main diseases of people aged 50-60 years old are concentrated in "refractive error, obesity (mild), hyperlipidemia, fatty liver, hyperuricemia, abnormal liver function indicators, thyroid nodules, abnormal kidney function indicators", among which abnormal liver function and hyperlipidemia account for the highest proportion.

[0020] Classification by department: Data analysis was conducted on 20 departments, and the diseases with high proportion in each department were obtained. Among them, abnormal liver function accounted for the highest proportion in the general physical examination items of the 20 departments.

[0021] S4. In order to verify the accuracy of the above statistical results, the chi-square test method was first used to perform a significance test on the statistical analysis results. The P value represents the probability of the chi-square statistic or more extreme values ​​observed in the hypothesis test. The smaller the P value, the more significant the observed difference. Conversely, the larger the P value, the less significant the observed difference. In order to control the false discovery rate (especially the type (false positive)), reduce false discoveries in the study, maintain the level of statistical significance, and obtain more reliable research results, the Bonferroni multiple comparison correction method was used for correction. The Bonferroni multiple comparison correction method can solve the problem of excessively high statistical significance levels that may result from multiple hypothesis tests. Correction can be used to accurately determine whether there is a significant association between the disease and gender, age, and department. The specific results and analysis are as follows: (1) According to gender classification, after Bonferroni multiple comparison correction method, the results are shown in Table 1: there is a significant association between the disease "refractive error" and gender (corrected P-value=0.031); there is a significant association between the disease "obesity" and gender (corrected P-value=0.021); there is a significant association between the disease "hyperlipidemia" and gender (corrected P-value=3.07e-05); there is a significant association between the disease "fatty liver" and gender (corrected P-value=0.040); there is no significant association between the disease "hyperuricemia" and gender (corrected P-value=1); there is no significant association between the disease "abnormal liver function indexes" and gender (corrected P-value=1); there is no significant association between the disease "abnormal renal function indexes" and gender (corrected P-value=1); there is no significant association between the disease "mild obesity" and gender (corrected P-value=0.075).

[0022]

[0023] (2) After the Bonferroni multiple comparison correction method was used according to age classification, the following results were obtained as shown in Figure 2: There was a significant correlation between the disease "refractive error" and age (corrected P-value=0.024); there was a significant correlation between the disease "hyperlipidemia" and age (corrected P-value=2.38e-05); there was a significant correlation between the disease "breast hyperplasia" and age (corrected P-value=0.00011); there was no significant correlation between the disease "abnormal renal function indicators" and age (corrected P-value=1); there was no significant correlation between the disease "hyperuricemia" and age (corrected P-value=1); there was no significant correlation between the disease "abnormal liver function indicators" and age (corrected P-value=1).

[0024] (3) According to the department classification, the Bonferroni multiple comparison correction method was used to obtain the following results, as shown in Table 3: There is a significant association between the disease "refractive error" and the department (adjusted P-value=1.42e-06); there is a significant association between the disease "obesity" and the department (adjusted P-value=2.98e-06); there is a significant association between the disease "hyperlipidemia" and the department (adjusted P-value=1.48e-05); there is a significant association between the disease "fatty liver" and the department (adjusted P-value=0.00057); there is a significant association between the disease "abnormal liver function index" and the department (adjusted P-value=0.00017); there is a significant association between the disease "thyroid nodules" and the department (adjusted P-value=0.00017). There is a significant correlation between the disease "breast hyperplasia" and the department (adjusted P-value=0.00034); there is a significant correlation between the disease "abnormal renal function indicators" and the department (adjusted P-value=1.86e-29); there is a significant correlation between the disease "abnormal electrocardiogram" and the department (adjusted P-value=2.75e-05); there is no significant correlation between the disease "abnormal electrocardiogram" and the department (adjusted P-value=0.34).

[0025]

[0026] S5. By performing significance analysis on the multiple comparison correction method of physical examination data, various physical examination indicators can be integrated to more objectively describe the user's physical health status, which provides theoretical guidance for doctors to provide health guidance to the examinees and accurately locate the physical examination items of patients. At the same time, it reduces the communication cost between physical examination users and doctors, and urges users to pay more attention to the overall physical health status.

[0027] The embodiments described above are only descriptions of the preferred modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for significance analysis of physical examination data based on multiple test correction, specifically comprising the following steps: S1. Clean the physical examination data horizontally and vertically; S2. Classify and analyze the data by gender, age, and department, and preliminarily obtain the data distribution of the physical examination results corresponding to different categories, and highlight the abnormal values; S3. Set the threshold of high-proportion diseases, calculate the proportion of each disease in different categories, and perform statistical analysis on high-proportion diseases; S4. The chi-square test method was used to test the significance of the statistical analysis results, and the Bonferroni multiple comparison correction method was used for correction.

2. A method for significance analysis of physical examination data based on multiple testing correction according to claim 1, characterized in that: S1 specifically includes the following steps: First, the data is labeled, with negative as 0, positive as 1, and no test as 3. Then the data is cleaned horizontally and vertically. The horizontal processing method is: if a row is all 3, it means that the person has not undergone a physical examination. Delete this row and do not participate in the calculation of l; the vertical processing method is: if a column is all 3, it means that all people have undergone this examination. Delete this column to reduce the amount of data; if there are no null values ​​or NaN in the data set, convert all null values ​​to 3 and delete the 3 value in the data.

3. The method for significance analysis of physical examination data based on multiple testing correction according to claim 1, characterized in that: S2 specifically includes the following steps: Classify the data by gender, age, and department, draw statistical graphs based on the classification, and highlight abnormal values ​​or outliers when they exist in the statistical graph.

4. The method for significance analysis of physical examination data based on multiple testing correction according to claim 1, characterized in that: S3 specifically includes the following steps: Calculate the proportion of each disease in different categories of gender, age, and department, and conduct statistical analysis on diseases with high proportions. Divide the number of positive people by the total number of people to get the proportion of the positive people. Based on the proportion of each disease, the distribution proportion of the disease in the population can be obtained.

Images