Key pathogenic factor acquisition method based on rough entropy

Abstract

The invention belongs to the technical field of data mining and knowledge discovery, and particularly relates to a rough entropy-based key pathogenic factor acquisition method, which comprises the following steps of: depicting disease factors by using an intuitionistic fuzzy information system, and establishing an effective granulation mechanism based on a dominant relationship in the intuitionistic fuzzy information system, enabling the information particles under the relation to conform to the granularity monotonicity; uncertainty measurement is carried out on the intuitionistic fuzzy information system based on rough entropy, and then the importance degree of pathogenic factors is described; all key pathogenic factors are obtained through the importance degrees of the pathogenic factors, so that the decision quality of medical diagnosis and the efficiency of knowledge discovery are improved. According to the method, redundant factors and data dimensions are reduced, computing resource consumption is reduced, and decision quality and knowledge discovery efficiency are effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of data mining, and in particular relates to a method for obtaining key pathogenic factors based on rough entropy. Background Art

[0002] With the popularization of sensor technology, the Internet and the Internet of Things, as well as the development of platforms such as social media, data sources have become more diverse and data types have become more complex. Classical rough sets have gradually been unable to handle complex data in daily life, so some scholars have combined intuitionistic fuzzy sets with rough sets and proposed intuitionistic fuzzy rough sets. Intuitionistic fuzzy information systems are widely used due to their excellent ability to process complex data, but due to the extensiveness and fuzziness of their data, they have a large amount of attribute redundancy, so attribute reduction has become a key research issue in intuitionistic fuzzy information systems.

[0003] However, most of the existing methods for reducing the attributes of intuitionistic fuzzy information systems are based on the identification matrix or algebra to define the attribute importance, which leads to low time efficiency. There are relatively few studies on characterizing the attribute importance based on information entropy. As an important method for measuring the attribute importance, information entropy can measure the uncertainty of information systems. Since most similarity measurement methods of intuitionistic fuzzy information systems are not monotonic, it is impossible to directly use information entropy to measure intuitionistic fuzzy information systems. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for obtaining key pathogenic factors based on rough entropy, comprising:

[0005] S1. Characterize disease factors with an intuitionistic fuzzy information system, and establish an effective granulation mechanism based on the dominant relationship in the intuitionistic fuzzy information system, so that the information granules under this relationship conform to the granularity monotonicity and can be applied to the entropy theory;

[0006] S2: Based on the rough entropy, the uncertainty of the intuitive fuzzy information system with granulation mechanism is measured to characterize the importance of pathogenic factors;

[0007] The importance of the pathogenic factors includes: absolute factor importance and relative factor importance;

[0008] S3: Obtain all key pathogenic factors through absolute factor importance and relative factor importance to improve the decision-making quality of medical diagnosis and the efficiency of knowledge discovery.

[0009] Beneficial effects of the present invention:

[0010] The method for obtaining key pathogenic factors based on rough entropy proposed in the present invention characterizes disease factors with an intuitive fuzzy information system, establishes an effective granulation mechanism by utilizing the dominant relationship in the intuitive fuzzy information system, measures the uncertainty of the intuitive fuzzy information system based on rough entropy, and further characterizes the importance of pathogenic factors. The key pathogenic factors are obtained through the absolute factor importance and the relative factor importance, thereby reducing redundant factors and data dimensions and reducing computing resource consumption, and effectively improving the decision-making quality and the efficiency of knowledge discovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a schematic flow chart of a method for obtaining key pathogenic factors based on rough entropy of the present invention;

[0012] Figure 2 This is a schematic diagram of the algorithm for obtaining the key pathogenic factors of the present invention. DETAILED DESCRIPTION

[0013] 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.

[0014] A method for obtaining key pathogenic factors based on rough entropy, such as Figure 1 As shown, the following steps are included:

[0015] S1: Characterize disease factors using intuitionistic fuzzy information system and establish granulation mechanism;

[0016] In intuitionistic fuzzy information systems, due to the change of granularity mechanism, the granulation of data space is no longer strict, but has turned into a kind of coverage, which makes the entropy theory of general information systems unable to be directly applied to intuitionistic fuzzy information systems. Therefore, an effective granulation mechanism is established based on the dominance relationship in intuitionistic fuzzy information systems. In intuitionistic fuzzy information systems, the dominance relationship defines the correlation between different attributes, thereby dividing the data into different information granules. Each information granule represents a specific data pattern, and its uncertainty can be measured by information entropy; the information granules divided by the dominance relationship have granularity monotonicity, that is, as the granulation increases, the uncertainty of the information granules gradually decreases. This monotonicity allows information entropy to be directly applied to intuitionistic fuzzy information systems, making the granulation mechanism closely linked to information entropy, which can more accurately measure the uncertainty of the system and effectively identify key pathogenic factors.

[0017] First, collect the patient's disease-related data from the hospital's electronic medical record system (EMR) and laboratory information system (LIS). These data include but are not limited to the patient's physiological indicators (such as blood pressure, blood sugar, blood lipids, etc.), symptoms (such as fever, cough, fatigue, etc.), lifestyle (such as smoking history, drinking history, exercise habits, etc.), family history, and laboratory test results (such as blood indicators, imaging test results, etc.). These data are usually high-dimensional, highly uncertain, and complex, so they need to be preprocessed.

[0018] The preprocessing steps include data cleaning (removing missing values ​​and outliers), data standardization (converting data of different dimensions to a unified dimension), and data normalization (scaling the data to the [0,1] interval). The preprocessed data can more accurately reflect the health status of patients and provide a reliable data basis for subsequent analysis.

[0019] The preprocessed disease data are characterized by an intuitionistic fuzzy information system, which can more accurately describe the uncertainty and ambiguity of data, and is particularly suitable for the common incompleteness and ambiguity problems in medical data.

[0020] Let the intuitionistic fuzzy information system be IFIS = (U, AT, V a ,f a ), where U and AT are non-empty object sets and attribute sets respectively, V a is the value range of attribute a∈AT, f a (x)=<μ a (x),v a (x)>is the information function, and a∈AT,0≤μ a (x)+v a (x)≤1.

[0021] Among them, U is a non-empty object set, representing the patient group; AT is an attribute set, representing various factors related to the disease (such as physiological indicators, symptoms, lifestyle, etc.); V a is the value range of the attribute; f a It is the information function, which is used to describe the value of each patient on each factor.

[0022] In the intuitionistic fuzzy information system, the dominance relationship is defined to describe the association between different attributes. For example, some physiological indicators may have a stronger explanatory power for certain symptoms, and this relationship can be characterized by the dominance relationship.

[0023] A∈AT, all dominance relations about A are defined as:

[0024]

[0025] From this we can define the dominance class of x with respect to A under the dominance relation:

[0026]

[0027] remember Through dominance relationships, complex disease data can be characterized using a hierarchical information system, laying the foundation for further analysis.

[0028] Among them, A represents the key pathogenic factor; AT represents the set of all pathogenic factors; represents all advantage relations about A; represents the dominant class of x with respect to A; x and y represent different arbitrary objects; U represents a non-empty object set; μ a (x) represents the positive membership of x to the pathogenic factor a; μ a (y) represents the positive membership of y to the pathogenic factor a; ν a (x) represents the negative membership of x to the pathogenic factor a; ν a (y) represents the negative membership of y with respect to the pathogenic factor a.

[0029] S2: Uncertainty measurement is performed through rough entropy to obtain the importance of pathogenic factors;

[0030] Rough entropy can be used to average the uncertainty of different dominant classes in an intuitionistic fuzzy information system. Rough entropy is a tool that can quantify system uncertainty. By calculating the rough entropy of each pathogenic factor, the contribution of the factor to disease diagnosis can be evaluated. The higher the importance of the pathogenic factor, the greater its influence on disease diagnosis.

[0031] A∈AT, the rough entropy of A is defined as:

[0032]

[0033] The relative factor importance of any pathogenic factor a∈A and the absolute factor importance of any factor b∈AT-A are defined as:

[0034] sig in (a,A)=E r (A-{a})-E r (A)

[0035] sig out (b,A)=E r (A)-E r (A∪{b})

[0036] Among them, E r(A) represents the rough entropy of A, A represents the key pathogenic factors; AT represents the set of all pathogenic factors; U represents the set of non-empty objects; Represents x i About A's advantage class; sig in (a, A) represents the relative importance of a to A; {a} represents a set containing only the pathogenic factor a; sig out (b,A) represents the absolute factor importance of b to A; {b} represents a set that only contains the pathogenic factor b.

[0037] Based on the rough entropy, the importance of each pathogenic factor is evaluated. Factors with high importance usually have lower rough entropy values, indicating that they contribute more to disease diagnosis. The relative factor importance indicates the necessity of diagnosis of the pathogenic factor. If sig in (a,A)>0, then a is necessary; the absolute factor importance is used to select the most important factor other than the necessary factors.

[0038] All key pathogenic factors are obtained through absolute factor importance and relative factor importance. The steps are as follows Figure 2 shown.

[0039] Calculate the relative importance of factors to select necessary factors: In order to obtain key pathogenic factors and delete redundant factors that do not contribute much to disease diagnosis, we require that the uncertainty of key pathogenic factors is consistent with the original pathogenic factors. Therefore, we define that key pathogenic factor A must meet two conditions:

[0040] S31: To define a key pathogenic factor A, two conditions must be met:

[0041] (1)E r (AT) = E r (A)

[0042] (2) E r (AT)≠E r (A-{a})

[0043] Among them, E r (AT) represents the rough entropy of AT; E r (A) represents the rough entropy of A; E r (A-{a}) represents the rough entropy of A-{a}; A represents the key pathogenic factor; AT represents the set of all pathogenic factors;

[0044] Thus, we can get S32: Based on conditions (1) and (2), we can get that if A is a key pathogenic factor, it will be the case if and only if it satisfies: E r (AT) = E r (A) sig in(a,A)>0; where sig in (a, A) represents the relative importance of pathogenic factor a to A;

[0045] S33: Let the key pathogenic factors Key(AT) be empty; wherein Key(AT) represents all the key pathogenic factors in AT;

[0046] S34: Calculate the relative importance of all original pathogenic factors to determine whether they are necessary pathogenic factors. If they are greater than 0, they are necessary pathogenic factors, and put all necessary pathogenic factors into the key pathogenic factor Key (AT);

[0047] Calculate the relative importance of all original pathogenic factors, including:

[0048] sig in (a i ,AT)=E r (AT-{a i})-E r (AT)

[0049] Among them, sig in (a i ,AT) represents a i The relative importance of factors for AT; E r (AT-{a i}) means AT-{a i}'s rough entropy; E r (AT) represents the rough entropy of AT; AT represents the set of all pathogenic factors; a i represents the i-th pathogenic factor in AT.

[0050] S35: Compare the rough entropy and use the absolute factor importance to find the final key pathogenic factors.

[0051] Compare the rough entropy and use the absolute factor importance to find the final key pathogenic factors, including:

[0052] S351: If Key(AT) is consistent with the rough entropy of AT, execute S353; wherein Key(AT) represents all key pathogenic factors in AT, and AT represents the set of all pathogenic factors;

[0053] S352: Calculate the absolute importance of all factors in AT-Key (AT), and put the factor with the greatest importance into the key pathogenic factor Key (AT), and execute S351;

[0054] S353: Output the key pathogenic factor Key (AT).

[0055] Calculate the absolute importance of all factors in AT-Key (AT), including:

[0056] sig out (a j ,Key(AT))=E r (Key(AT))-E r (Key(AT)∪{a j})

[0057] Among them, sig out (a j ,Key(AT)) represents the absolute importance of all factors in AT-Key(AT); Key(AT) represents all key pathogenic factors in AT; AT represents the set of all pathogenic factors; a j represents the jth pathogenic factor in AT; E r (Key(AT)) represents the rough entropy of Key(AT); E r (Key(AT)∪{a j}) means Key(AT)∪{a j} is the rough entropy of .

[0058] In this way, the key pathogenic factors are obtained and the remaining redundant data are eliminated, such as eliminating examination indicators that are not strongly related to the diagnosis, providing doctors with more concise and accurate diagnostic data.

[0059] The present invention characterizes disease factors using an intuitive fuzzy information system, establishes an effective granulation mechanism using the dominant relationship in the intuitive fuzzy information system, and measures the uncertainty of the intuitive fuzzy information system based on rough entropy, thereby characterizing the importance of pathogenic factors and obtaining key pathogenic factors through absolute factor importance and relative factor importance.

[0060] The method for acquiring key pathogenic factors based on rough entropy proposed in the present invention effectively avoids the problems of data redundancy and low decision-making efficiency, and can significantly improve the decision-making quality of medical diagnosis and the efficiency of knowledge discovery.

[0061] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for obtaining key pathogenic factors based on rough entropy, characterized in that: include: S1. Characterize disease factors with an intuitionistic fuzzy information system, and establish an effective granulation mechanism based on the dominant relationship in the intuitionistic fuzzy information system, so that the information granules under this relationship conform to the granularity monotonicity and can be applied to the entropy theory; S2: Based on the rough entropy, the uncertainty of the intuitive fuzzy information system with granulation mechanism is measured to characterize the importance of pathogenic factors; The importance of the pathogenic factors includes: absolute factor importance and relative factor importance; S3: Obtain all key pathogenic factors through absolute factor importance and relative factor importance to improve the decision-making quality of medical diagnosis and the efficiency of knowledge discovery.

2. The method for obtaining key pathogenic factors based on rough entropy according to claim 1 is characterized in that: The disease factors are characterized by the intuitionistic fuzzy information system, including: IFIS=(U,AT,V a ,f a ) Where IFIS stands for Intuitive Fuzzy Information System; U and AT represent the non-empty object set and attribute set, respectively, corresponding to the patients and diseases in the disease factor data; V a Indicates the value range of the attribute; f a is the information function.

3. The method for obtaining key pathogenic factors based on rough entropy according to claim 1 is characterized in that: An effective granulation mechanism is established based on the dominant relationship in the intuitionistic fuzzy information system, including: A∈AT, all dominance relations about A are defined as: Based on all dominance relations about A, define the dominance class of attribute x about A: remember Obviously it is monotonic; Among them, A represents the key pathogenic factor; AT represents the set of all pathogenic factors; represents all advantage relations about A; represents the dominant class of x with respect to A; x and y represent different arbitrary objects; U represents a non-empty object set; μ a (x) represents the positive membership of x to the pathogenic factor a; μ a (y) represents the positive membership of y to the pathogenic factor a; v a (x) represents the negative membership of x to the pathogenic factor a; v a (y) represents the negative membership of y with respect to the pathogenic factor a.

4. The method for obtaining key pathogenic factors based on rough entropy according to claim 1 is characterized in that: The uncertainty is measured through rough entropy and the attribute importance is obtained, including: A∈AT, the rough entropy of A is defined as: The relative factor importance of any pathogenic factor a∈A and the absolute factor importance of any factor b∈AT-A are defined as: Mr in (a,A)=And r (A-{a})-E r (TO) Mr out (b,A)=And r (A)-E r (A∪{b}) Among them, E r (A) represents the rough entropy of A, A represents the key pathogenic factors; AT represents the set of all pathogenic factors; U represents the set of non-empty objects; Represents x i About A's advantage class; sig in (a, A) represents the relative importance of a to A; {a} represents a set containing only the pathogenic factor a; sig out (b,A) represents the absolute factor importance of b to A; {b} represents a set that only contains the pathogenic factor b. Based on the rough entropy, the importance of each pathogenic factor is evaluated. Factors with high importance usually have lower rough entropy values, indicating that they contribute more to disease diagnosis; the relative factor importance explains the necessity of diagnosis of the pathogenic factor. in (a,A)>0, then a is necessary; the absolute factor importance is used to select the most important factor other than the necessary factors.

5. The method for obtaining key pathogenic factors based on rough entropy according to claim 1 is characterized in that: All key pathogenic factors are obtained through absolute factor importance and relative factor importance, including: S31: To define a key pathogenic factor A, two conditions must be met: (1)E r (AT)=E r (A) (2) Among them, E r (AT) represents the rough entropy of AT; E r (A) represents the rough entropy of A; E r (A-{a}) represents the rough entropy of A-{a}; A represents the key pathogenic factor; AT represents the set of all pathogenic factors; Condition (1) requires that the uncertainty of the key pathogenic factor is consistent with that of the original pathogenic factor; Condition (2) requires that each of the key pathogenic factors is a necessary factor; S32: Based on conditions (1) and (2), it can be concluded that if A is a key pathogenic factor, it will only be if: sig in (a, A)>0; where sig in (a, A) represents the relative importance of pathogenic factor a to A; S33: Let the key pathogenic factors Key(AT) be empty; wherein Key(AT) represents all the key pathogenic factors in AT; S34: Calculate the relative importance of all original pathogenic factors to determine whether they are necessary pathogenic factors. If they are greater than 0, they are necessary pathogenic factors, and put all necessary pathogenic factors into the key pathogenic factor Key (AT); S35: Compare the rough entropy and use the absolute factor importance to find the final key pathogenic factors.

6. The method for obtaining key pathogenic factors based on rough entropy according to claim 5 is characterized in that: Calculate the relative importance of all original pathogenic factors, including: slg in (a i ,AT)=E r (AT-{a i })-E r (AT) Among them, sig in (a i , AT) represents a i The relative importance of factors for AT; E r (AT-{a i }) means AT-{a i }'s rough entropy; E r (AT) represents the rough entropy of AT; AT represents the set of all pathogenic factors; a i Indicates the first pathogenic factor in AT.

7. The method for obtaining key pathogenic factors based on rough entropy according to claim 5 is characterized in that: Compare the rough entropy and use the absolute factor importance to find the final key pathogenic factors, including: S351: If Key(AT) is consistent with the rough entropy of AT, execute S353; wherein Key(AT) represents all key pathogenic factors in AT, and AT represents the set of all pathogenic factors; S352: Calculate the absolute importance of all factors in AT-Key (AT), and put the factor with the greatest importance into the key pathogenic factor Key (AT), and execute S351; S353: Output the key pathogenic factor Key (AT).

8. The method for obtaining key pathogenic factors based on rough entropy according to claim 7 is characterized in that: Calculate the absolute importance of all factors in AT-Key (AT), including: sig out (a j ,Key(AT))=E r (Key(AT))-E r (Key(AT)∪{a j }) Among them, sig out (a j , Key(AT)) represents the absolute importance of all factors in AT-Key(AT); Key(AT) represents all key pathogenic factors in AT; AT represents the set of all pathogenic factors; a j represents the jth pathogenic factor in AT; E r (Key(AT)) represents the rough entropy of Key(AT); E r (Key(AT)∪{a j }) means Key(AT)∪{a j } is the rough entropy of .

Images