Patient condition assessment method and system based on multi-source information fusion mechanism

Abstract

The invention discloses a patient condition assessment method and system based on a multi-source information fusion mechanism. The invention relates to the technical field of medical service management. The basic disease index data set D is constructed into a data matrix dm with one row representing different historical patients and one column representing different indexes; performing principal component analysis (PCA) based on the data matrix dm to obtain an evaluation factor A; calculating a death speed curve Dvc of a historical patient, and clustering the patient by using an ordered clustering algorithm according to the historical medical record data D'and the death speed curve Dvc; a data analysis technology, including principal component analysis, ordered clustering analysis, a D-S evidence theory and the like, is fully utilized, and quantitative data with different sources and different properties are organically combined to form a comprehensive consideration factor. The data-driven method not only improves the accuracy and efficiency of evaluation, but also provides a powerful decision support tool for hospital managers.

Description

Technical Field

[0001] The present invention relates to the field of medical service management technology, specifically to the technical direction of incorporating patients' underlying diseases and mortality rates into medical reputation considerations, and especially to a patient condition assessment method and system based on a multi-source information fusion mechanism. Background Art

[0002] The reputation of a hospital is similar to the reputation of a company. It is an important intangible asset and plays an important role in attracting patients and improving the competitiveness of the hospital. The evaluation of reputation is not only a measure of the hospital's external image, but also a comprehensive reflection of its internal management, service quality, technical level and other factors.

[0003] Hospital reputation management faces many challenges, such as asymmetric medical information and diverse patient needs. For this problem of quantifying hospital reputation, many existing technologies attempt to solve and further improve the objectivity and accuracy of reputation evaluation, such as: (1) CN202110489756.7 discloses a reputation risk management capability assessment method (publication date 2024-08-20). By combining the business value with the weight, the scoring information of each evaluation indicator can be obtained. The scoring information is sorted to obtain the corresponding evaluation information of each evaluation dimension. At the same time, CN201310343622.X discloses a service trust assessment method based on service quality and reputation (publication date 2016-07-06). With the credibility of service selection as the background, it integrates service trust information from multiple sources to realize a reputation evaluation mechanism; However, the above-mentioned existing technologies cannot reflect the seasonal changes in physiological functions or disease spectra, cannot quantify the data of various characteristics, cannot provide a more comprehensive evaluation and reflect the reputation of the hospital, and combine business values ​​with weights, often relying on the subjective judgment of experts or managers to set weights, which lacks objectivity.

[0004] (2) CN202110185476.7 discloses a reputation-based V2X node message credibility assessment method (publication date 2022-08-26), which comprehensively determines the credibility of the response message based on the time difference between the time information related to the request content and the time information when responding to the request message, the distance between the location information related to the request content and the location information when responding to the request message, and the reputation value of the message responder; at the same time, CN202310883674.X discloses an assessment method based on big data analysis (publication date 2023-08-29), which collects information data related to the target group based on big data, extracts features of positive data, negative data and neutral data, extracts feature keywords and several topics, and mines and then realizes a reputation evaluation mechanism; However, the above-mentioned existing technologies mainly rely on the time difference between the time information related to the request content and the time of response, the distance between the location information, and the reputation value of the message responder to determine the credibility of the response message. This evaluation method ignores the performance of the hospital in many aspects such as patient treatment, service quality, medical technology, and the factors of death caused by patients themselves. The evaluation dimension and accuracy are relatively limited. In addition, although the traditional method collects information data related to the target based on big data, it does not make full use of the data. Because it mainly focuses on the feature extraction of positive data, negative data, and neutral data.

[0005] (5) CN202210141263.9 discloses a reputation risk monitoring and quantitative assessment method (publication date 2022-06-24), which performs feature matching on the negative feature information to be processed to obtain the matched negative feature information to be processed; performs risk monitoring on the matched negative feature information to be processed to obtain corresponding early warning information; However, the above existing technologies are relatively static and it is difficult to adjust the evaluation model as the hospital environment changes. When comprehensively determining the credibility of the response message, there is a lack of a comprehensive evaluation mechanism to evaluate the combination of different factor weights and quantify the data of each feature, and thus it is impossible to obtain a more comprehensive reputation evaluation result.

[0006] To this end, the present invention proposes a patient condition assessment method and system based on a multi-source information fusion mechanism. Summary of the invention

[0007] In view of this, the present invention hopes to provide a patient condition assessment method and system based on a multi-source information fusion mechanism to solve or alleviate the technical problems existing in the prior art, namely: (1) How to reflect seasonal changes in physiological functions or disease spectrum and quantify data of various characteristics; (2) How to avoid the subjective problem of relying on the subjective judgment of experts or managers to set weights; (3) How to comprehensively combine different quantitative data based on a data-driven approach to better adapt to changes in the hospital environment and quantify the data of each feature to obtain a more comprehensive reputation evaluation result.

[0008] The technical solution of the present invention is achieved in this way: First, the patient condition assessment method based on multi-source information fusion mechanism: 1. Overview: This invention aims to build a comprehensive hospital reputation evaluation system by comprehensively evaluating multiple factors such as the severity of the patient's underlying disease, mortality risk, and the reputation of the hospital. The principal component analysis (PCA) is used to evaluate the patient's underlying disease, and the ordered cluster analysis of the patient's mortality risk is used. At the same time, multiple characteristic data of the hospital are collected and quantified. The DS evidence theory is used to combine the treatment factor and the comprehensive reputation characteristic value to form a comprehensive consideration factor to guide patients to reasonably choose a hospital or physician. It aims to provide patients with a more accurate and comprehensive basis for medical service selection through scientific data analysis and comprehensive evaluation methods.

[0009] (II) Technical solution: To achieve the above technical goals, when a patient calls the medical reputation of a hospital / physician, the basic disease indicator dataset D of historical patients is collected from the hospital information system, including basic information (such as age, gender and BMI index), past medical history and physical sign measurement data, and then the following steps are performed.

[0010] 2.1 Step S1, data collection: The basic disease index dataset D is constructed into a data matrix d with rows representing different historical patients and columns representing different indicators. m .

[0011] 2.1.1 Step S100, data cleaning: Fill missing values, delete or correct outliers in the basic disease indicator dataset D. Then use the Z-score standardization method to standardize all indicators in the basic disease indicator dataset D to ensure that all variables have the same mean and variance.

[0012] (1) Missing value filling: ; where x missing is a missing value, X i are non-missing values, n is the number of non-missing values; (2) Outlier processing (Z-score method): ; Among them, x i is the i-th original value in the basic disease index dataset D, μ a is the mean of the current indicator on all non-outlier patients, σ a is the standard deviation of the current indicator on all non-outlier patients. Data points with |Z|>3 are considered outliers; (3) Z-score standardization: ; Where x i ′ is the standardized i-th element in the basic disease index dataset D, x i is its original value, μd is the mean of the current indicator on all patients with non-missing values; σ d is the standard deviation of the current indicator on all patients with non-missing values.

[0013] 2.1.2 Step S101, constructing data matrix d m : ; Where x ij ′ represents the (standardized value of) the jth indicator of the ith historical patient, n is the number of patients, and m is the number of indicators.

[0014] 2.2 Step S2, assessment of the patient's underlying disease: Based on the data matrix d m Principal component analysis (PCA) was performed to obtain the evaluation factor A.

[0015] 2.2.1 Step S200, calculate the covariance matrix: The data matrix d m After being converted into a covariance matrix C that describes the strength and direction of the linear relationship between the indicators, eigenvalue decomposition is performed to obtain the eigenvalue E (expressed in matrix form) and the corresponding eigenvector F (expressed in diagonal matrix form): (1) Convert to covariance matrix C: ; in, Denotes the data matrix d m The mean vector of Denotes the data matrix d m The i-th row in is the index data of the i-th patient (a 1*m vector, where m is the number of indexes j), n is the number of patients; T represents the transpose operation of the matrix.

[0016] (2) Eigenvalue decomposition: C = F * E * F T ; Where F is the eigenvector matrix, E is the diagonal matrix containing the eigenvalues, and T represents the transpose operation of the matrix.

[0017] 2.2.2 Step S201, select the main component: According to the size of the eigenvalue E, the principal component P whose cumulative contribution rate reaches the preset threshold is selected; then the original data is projected into the new feature space F composed of the selected principal component P. n , calculate the new coordinates P of each data point c (i.e., principal component score); for the new coordinate P c After normalization and weighted summation based on contribution rate, we get the evaluation factor A: (1) Suppose k principal components are selected so that the cumulative contribution rate R C Reaching a preset threshold; (2) Calculate the cumulative contribution rate R C : ; Among them, E jj represents the j-th diagonal element of the diagonal matrix E; m is the number of indices j; (3) Project the original data into the new feature space F consisting of the selected principal components n Middle: P c =d m *F k ; Among them, F k is the matrix composed of the eigenvectors corresponding to the selected k principal components; P c is the principal component score; (4) Principal component score P c Perform normalization: ; Among them, P cn Represents the normalized value of the jth principal component score; (5) Based on the weighted sum of contribution rates, we get the evaluation factor A: ; in, It represents the sum of the variance contributions of the first k principal components and is used to calculate the weight of each principal component in the weighted summation.

[0018] 2.3 Step S3, patient mortality assessment: Calculate the death rate curve D of historical patients vc , using the ordered clustering algorithm based on the historical medical record data D' and the death rate curve D vc The patients were clustered and the evaluation factor B was calculated.

[0019] 2.3.1 Step S300, collecting historical medical record data D': Collect the disease type De and death time d t Then the data is coded, the time format is unified, and the length of hospital stay is calculated to form the historical medical record data D'.

[0020] 2.3.2 Step S301, determine the time window and death time distribution: Set the time window T of the patient in the historical medical record data D' w (If within one year after admission), the month of death is M d ; For each historical patient, analyze different diseases De and their corresponding death time d tThe distribution of the death rate curve D was obtained by time series analysis and curve fitting using the exponential model. vc : (1) Exponential model: D vc (t) = a·exp(b·t)+c; Where a, b and c are the parameters to be fitted; t is the time variable (hours, days, weeks or months, etc.); (2) Use the least squares method to minimize the sum of the squared differences between the observed values ​​and the model predicted values: Right now ; Where N is the number of observations. The exponential model is fitted and parameter estimated by using relevant functions or libraries in statistical software or programming languages ​​(such as Python, R, etc.), solving the above least squares problem, and obtaining the values ​​of parameters a, b, and c.

[0021] 2.3.3 Step S302, ordered cluster analysis: Using the ordered clustering algorithm (Fisher's optimal partitioning method) based on the historical medical record data D' and the death rate curve D vc Cluster the patients to obtain different risk groups. Quantify the evaluation indicators and calculate the evaluation factor B; The goal of Fisher's optimal segmentation method is to divide n patients into K ordered groups (i.e. risk groups) so that the similarities within the ordered groups are maximized and the differences between the ordered groups are maximized: (1) Objective function: Let G1, G2, …, G K It is a segmentation that divides n patients into K groups, and defines the objective function L(n,K) as: ; Among them, n i is the i-th group G i The number of patients in It is group G i The feature mean vector of the patients in , is the feature mean vector of all patients.

[0022] (2) Solving the optimal segmentation: Use the dynamic programming algorithm to solve the optimal segmentation. Define L(j,i) as the objective function value of the optimal segmentation that divides the first j patients into i groups. The recursive relationship is: ; in, is the feature mean vector from patient s+1 to patient j; Represents the difference between the feature mean vector from patient s+1 to patient j and the feature mean vector of all patients. represents the transpose of the difference vector (in the case of real numbers, it is the vector itself; in the case of complex numbers or more general vector spaces, the conjugate operation should be performed). s represents a limit on the patient number, which is used to divide different patient groups. L(s,i−1) represents the objective function value of the optimal partitioning of the first s patients into i−1 groups.

[0023] (3) Initialization: Because there is only one group, there is no internal difference; so when i=1, L(j,1)=0; (4) Calculation: Starting from i=2, gradually increase the number of groups until the predetermined number of groups K is reached; (5) Obtaining the optimal segmentation: By backtracking the dynamic programming table, the optimal segmentation of n patients into K groups is obtained; (6) Quantification: ; in, It is group G i The proportion of patients in the total number of patients, R i It is group G i The (normalized) evaluation factor B reflects the overall risk level of the patient.

[0024] 2.4 Step S4, comprehensive reputation feature collection and quantification stage: Collect data on social reputation, treatment level, service attitude, and patient satisfaction to form a comprehensive reputation characteristic value S.

[0025] 2.4.1 Step S400, collect social reputation, treatment level, service attitude, and patient satisfaction data: The characteristic sequence R1 that quantifies reputation through industry ranking; The characteristic sequence R2 that quantifies the treatment level through the cure rate; The characteristic sequence R3 of service attitude is quantified through the complaint rate survey results; The characteristic sequence R4 of quantifying patient satisfaction directly through satisfaction survey results.

[0026] 2.4.2 Step S401, determine the weight and calculate the comprehensive reputation feature value: Determine the weights W1, W2, W3 and W4 corresponding to the feature sequences R1, R2, R3 and R4 using statistical methods. Then calculate the comprehensive reputation feature value S: S=W1⋅R1+W2⋅R2+W3⋅R3+W4⋅R4.

[0027] 2.5 Step S5, DS evidence theory algorithm combination stage: The DS evidence theory algorithm is used to combine the evaluation factor A and the evaluation factor B to form a treatment factor S with the patient's basic disease assessment information and the patient's mortality assessment information. aving .

[0028] 2.5.1 Step S500, define the recognition framework and assign basic probability BPA: Define the identification framework I={LR,MR,HR}, where LR represents low risk, MR represents medium risk, and HR represents high risk; then assign basic probabilities BPA to evaluation factors A and B, denoted by m and A and m B ;in: (1) m A (LR) represents the support degree of evaluation factor A for low-risk propositions, m A (MR) represents the support degree of evaluation factor A for medium-risk propositions, m A (HR) indicates the support degree of evaluation factor A for high-risk propositions; (2) m B (LR) represents the support degree of evaluation factor B for low-risk propositions, m B (MR) represents the support degree of evaluation factor B for medium-risk propositions, m B (HR) indicates the support degree of evaluation factor B for high-risk propositions; 2.5.2 Step S501, merging basic probability BPA: (1) Use Dempster's combination rule to combine the basic probabilities m A and the basic probability m B , calculate the normalization constant D: ; Among them, X and Y are subsets of the recognition framework I; the symbol represents the empty set; (2) Basic probability BPA value after merger m A,B Calculated as: ; Among them, Z is also a subset of the recognition framework I.

[0029] 2.5.3 Step S502, evaluation and decision making: Calculate the trust function Bel and the likelihood function Pl to form the rescue factor S aving Here is how: (1) Calculate the trust function Bel and the likelihood function Pl: ; ; (2) Based on the values ​​of the trust function Bel and the likelihood function Pl, the rescue factor S is formed aving The rule is that if the confidence of the low-risk LR proposition is greater than the likelihood of the high-risk HR proposition, then it can be considered that the treatment effect is good, so the treatment factor S aving The value of will also be larger: S aving = Bel(LR) - Pl(HR); Among them, Bel(LR) represents the confidence of low-risk propositions, and Pl(HR) represents the likelihood of high-risk propositions. aving The larger the value, the better the treatment effect, and the lower the overall risk after combining the patient's underlying disease assessment information and mortality assessment information.

[0030] 2.6 Step S6, hospital / doctor selection stage: Rescue factor S aving Combined with the comprehensive reputation characteristic value S, it becomes a comprehensive consideration factor C that can help patients choose hospitals / doctors CF .

[0031] 2.6.1 Step S600, determine weight: Using entropy weight method to determine the rescue factor S aving The weight W5 of and the weight W6 of the comprehensive reputation feature value S: (1) Constructing the evaluation matrix EM: There are n hospitals / doctors to be evaluated. Each hospital / doctor has two evaluation indicators: treatment factor S aving And the comprehensive reputation eigenvalue S. Construct an n*2 evaluation matrix EM, where any element x of the evaluation matrix EM ij Represents the value of the i-th hospital / physician on the j-th evaluation index.

[0032] (2) Data standardization: Due to the treatment factor S aving Both the comprehensive reputation characteristic value S are positive indicators (the larger the value, the better), and the same standardization operation is used: ; Among them, x ij ′ is the standardized value of the i-th hospital / physician on the j-th evaluation index.

[0033] (3) For each evaluation index j, calculate the entropy value e j : ; Among them, p ij represents the proportion of the standardized value of the i-th hospital / physician on the j-th evaluation index ( ); k is a constant equal to , where n is the number of hospitals / physicians to be evaluated.

[0034] (4) For each evaluation index j, calculate its difference coefficient g j :g j =1−e j ; (5) For each evaluation index j, calculate the corresponding weight w j : ; 2.6.2 Step S601, calculate comprehensive consideration factors: Combined with rescue factor S aving and the comprehensive reputation characteristic value S, and the comprehensive consideration factor C is calculated by weighted summation CF :CCF=W5*Saving+W6*S; Patients can choose hospitals / physicians based on the comprehensive consideration factor values. The higher the value, the better the hospital / physician performs in terms of medical technology, service quality, and reputation.

[0035] 3. Mechanisms for solving technical problems: 3.1 How to reflect seasonal changes in physiological functions or disease spectrum and quantify data of each characteristic: The present invention collects detailed medical records of patients and sets a time window to analyze the distribution of the patients' death months and observe whether there are concentrated death months or specific death rate curves. That is, the time series analysis and curve fitting technology are used to describe the trend of death rate changes over time, thereby reflecting the seasonal changes in physiological functions or disease spectrum. At the same time, for the data quantification of each feature, the present invention uses a variety of specific indicators to quantify features such as reputation to ensure the objectivity and comparability of the data.

[0036] 3.2 How to avoid the subjective problem of relying on the subjective judgment of experts or managers to set weights: The present invention adopts an objective data-driven method (entropy weight method) when setting weights, rather than relying on the subjective judgment of experts or managers. The entropy weight method determines the weight by analyzing the degree of discreteness of each feature data itself. The greater the degree of discreteness, the greater the amount of information provided by the feature, and thus the greater the weight in the comprehensive evaluation. It effectively avoids the deviation caused by subjective weight setting, making the weight distribution more scientific and reasonable.

[0037] 3.3 How to comprehensively combine different quantitative data based on a data-driven approach to better adapt to changes in the hospital environment and quantify the data of each feature to obtain a more comprehensive reputation evaluation result: The present invention adopts a variety of data-driven methods, among which principal component analysis is used to extract important comprehensive indicators of patients' underlying diseases, ordered cluster analysis is used to assess patients' mortality risk, and DS evidence theory is used to combine treatment factors and comprehensive reputation characteristic values. It can not only effectively quantify the data of each feature, but also organically combine quantitative data from different sources and different properties to form a comprehensive consideration factor. By continuously collecting and analyzing the latest medical data, the present invention can dynamically reflect changes in hospital environment and patient needs, thereby obtaining more comprehensive and accurate reputation evaluation results, and providing patients with more accurate and scientific medical service selection basis.

[0038] Second, the patient condition assessment system based on multi-source information fusion mechanism: The system is used to implement the above-mentioned patient condition assessment method based on multi-source information fusion mechanism, including: (1) The data collection module is responsible for automatically collecting the patient's basic disease indicator data set from the hospital information system: it has the functions of interface development, data capture, data cleaning and preprocessing.

[0039] (2) A data analysis module that uses principal component analysis to extract comprehensive indicators of patients' underlying diseases and uses ordered cluster analysis to assess patients' mortality risk: It has the functions of PCA algorithm implementation, ordered clustering algorithm implementation (Fisher optimal segmentation method), time series analysis and curve fitting.

[0040] (3) Feature quantification and weight allocation module responsible for executing the entropy weight method: It has the functions of formulating quantification standards, calculating weights using the entropy weight method, and weight allocation strategy.

[0041] (4) Fusion module responsible for implementing the DS evidence theory algorithm: using the Dempster combination rule to integrate the treatment factors (evaluation factors A and B) based on the patient's underlying disease and mortality risk with the hospital's comprehensive reputation characteristic value to form a comprehensive consideration factor. This includes the implementation of the DS evidence theory algorithm, the calculation of the basic probability assignment (BPA), and the calculation of the trust function and likelihood function.

[0042] (5) Result explanation module for evaluating the reputation of hospitals or physicians: interprets the integrated consideration factors and provides patients with recommendations on medical treatment options.

[0043] Compared with the prior art, the present invention has the following beneficial effects: 1. Improve the comprehensiveness and objectivity of the evaluation: By comprehensively considering the severity of the patient's underlying disease, mortality risk, hospital reputation, treatment level and other factors, the present invention can provide a more comprehensive and objective hospital reputation evaluation result. It helps patients to more accurately understand the overall strength and service quality of the hospital, and provide strong support for medical selection.

[0044] Second, reflecting seasonal changes in physiological functions and disease spectrum: By analyzing the distribution of the months of death of patients and fitting the death rate curve, the present invention can reveal the seasonal changes in physiological functions or disease spectrum. This is of great significance for disease prevention, allocation of medical resources and improvement of medical services, and helps hospitals better cope with seasonal health challenges.

[0045] 3. Reduce subjective influence: In terms of weight setting, the present invention adopts objective data-driven methods such as entropy weight method to avoid the bias caused by relying on the subjective judgment of experts or managers. This objective and scientific weight allocation method helps to improve the fairness and credibility of the evaluation results.

[0046] 4. Enhanced data-driven decision support: This invention makes full use of data analysis techniques, including principal component analysis, ordered cluster analysis and DS evidence theory, to organically combine quantitative data from different sources and properties to form a comprehensive consideration factor. This data-driven approach not only improves the accuracy and efficiency of the evaluation, but also provides a powerful decision support tool for hospital managers.

[0047] 5. Adapt to changes in hospital environment: Since the present invention is based on continuous data collection and analysis, it can dynamically reflect changes in hospital environment and patient needs. This makes the evaluation results closer to the actual situation, helping hospitals to adjust service strategies and optimize resource allocation in a timely manner to better meet patients' needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 It is a schematic diagram of the method flow of the present invention; Figure 2 Schematic diagram of the execution method of steps S2 to S5 of the present invention; Figure 3 It is a schematic diagram of the system composition of the present invention; Figure 4 This is a schematic diagram for comparing the principal component analysis results of step S2 in the first embodiment of the present invention; Figure 5 This is a schematic diagram for visualizing the clustering effect of step S3 in the first embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below; It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0051] Embodiment 1: Figure 1 As shown, this embodiment discloses a patient condition assessment method based on a multi-source information fusion mechanism. When a patient calls the medical reputation of a hospital / doctor, the hospital information system first calls out the basic disease index data set D of historical patients, including basic information (age and BMI index), past medical history, physical sign measurement data, laboratory test data (such as blood routine or liver and kidney function) and health records (eating habits, exercise habits and sleep status).

[0052] Specifically, the basic disease index dataset D is a matrix with n rows and m columns, where n is the number of patients and m is the number of corresponding basic disease indicators. ij Represents the jth underlying disease index value of the i-th patient: ; For each basic disease indicator D in the basic disease indicator dataset D ij , normalization is performed as follows: (1)

[0053] Among them, D ij ′ is the standardized value, min(D j ) and max(D j ) are the minimum and maximum values ​​of the jth underlying disease indicator in all patients.

[0054] (2) For some special underlying disease indicators, such as BMI or blood pressure, they all have standard normal ranges. In this case, choose to standardize these indicators to their normal ranges instead of using the minimum and maximum values ​​of the entire data set. For example, the normal range of BMI is 18.5 to 24.9, then: ; Or if you need to take into account the indicator values ​​that are beyond the normal range, use piecewise functions for processing; (3) For laboratory test data with different units and magnitudes, the influence of the units and magnitudes is eliminated based on the Z-score normalization method in step S100.

[0055] Then, the following steps S1 to S6 of the method are executed.

[0056] In this embodiment, regarding step S1, data collection: construct the basic disease index data set D into a data matrix d with rows representing different historical patients and columns representing different indicators. m The specific scheme includes the following steps S100~S101.

[0057] Specifically, step S100, data cleaning: The purpose of data cleaning is to ensure the integrity and accuracy of the data for subsequent analysis and modeling. This step includes three steps: missing value filling, outlier processing, and Z-score standardization. The Z-score standardization method standardizes all indicators to ensure that all variables have the same mean and variance: (1) Missing value filling: ; where x missing is a missing value, X i is the non-missing value, n is the number of non-missing values; by calculating the mean of the non-missing values ​​to fill the missing values, the overall distribution characteristics of the data can be maintained.

[0058] (2) Outlier processing (Z-score method): ; Among them, x i is the i-th original value in the basic disease index dataset D, μ a is the mean of the current indicator on all non-outlier patients, σ a is the standard deviation of the current indicator on all non-outlier patients. Data points with |Z|>3 are considered outliers; for outliers, you can choose to delete, modify, or use other methods based on the preset dictionary to ensure the accuracy and reliability of the data.

[0059] (3) Z-score standardization: First, calculate the mean and standard deviation of each indicator. Then, for each data point, we subtract the mean from its original value and divide it by the standard deviation to get its Z-score value. This Z-score value reflects the degree of deviation of the data point from the mean, in units of standard deviation: ; Where x i ′ is the standardized i-th element in the basic disease index dataset D, x i is its original value, μd is the mean of the current indicator on all patients with non-missing values; σ d is the standard deviation of the current indicator on all patients with non-missing values. Through standardization, we can eliminate the dimensional differences between different variables and make them comparable.

[0060] It is understandable that since Z-score standardization is based on the statistical characteristics of data, it is applicable to data of various types and distributions. For laboratory test data, although different test items may have different units and magnitudes, through Z-score standardization, we can convert them into comparable standard scale data, thus facilitating subsequent analysis and modeling.

[0061] Specifically, step S101, constructing a data matrix d m : ; Where x ij ′ represents the (standardized value of) the jth indicator of the ith historical patient, n is the number of patients, and m is the number of indicators. Data matrix d m It will serve as the basis for subsequent multi-source information fusion and hospital reputation evaluation.

[0062] In this embodiment, regarding step S2, the patient's underlying disease assessment: based on the data matrix d obtained in step S1 m Perform principal component analysis (PCA) to obtain the evaluation factor A. The specific scheme includes the following steps S200~S201. Figure 2 This step is an important part of helping patients obtain a more objective medical reputation of hospitals / doctors, because it quantifies the patient's underlying disease characteristics and provides key input for subsequent multi-source information fusion.

[0063] Specifically, step S200, calculate the covariance matrix: transform the data matrix d m After being converted into a covariance matrix C that describes the strength and direction of the linear relationship between the indicators, eigenvalue decomposition is performed to obtain the eigenvalue E and the corresponding eigenvector F: (1) Convert to covariance matrix C: ; in, Denotes the data matrix d m The mean vector of Denotes the data matrix d m The i-th row in is the index data of the i-th patient (a 1*m vector, where m is the number of indexes j); T represents the transpose operation of the matrix. The covariance matrix C describes the data matrix d mThe strength and direction of the linear relationship between the indicators in the covariance matrix. By calculating the deviation of each patient's indicator data from the mean vector and averaging their outer products, the covariance matrix can be obtained. This step is to capture the correlation between indicators and lay the foundation for subsequent eigenvalue decomposition and principal component selection.

[0064] (2) Eigenvalue decomposition: C = F * E * F T ; Among them, eigenvalue decomposition decomposes the covariance matrix C into the product of the eigenvector matrix F, the diagonal matrix E containing the eigenvalues, and the transpose of F. The eigenvalues ​​and eigenvectors reflect the degree and direction of the data variation in the corresponding directions, respectively. Through eigenvalue decomposition, the main variation direction in the data, namely the principal component, can be identified.

[0065] Specifically, step S201, select the principal component: select the principal component P whose cumulative contribution rate reaches the preset threshold according to the size of the eigenvalue E; then project the original data into the new feature space F composed of the selected principal component P n , calculate the new coordinates P of each data point c (i.e., principal component score); for the new coordinate P c After normalization and weighted summation based on contribution rate, we get the evaluation factor A: (1) Suppose k principal components are selected so that the cumulative contribution rate R C Reach the preset threshold; the cumulative contribution rate reflects the proportion of the selected principal component in explaining the total variation of the data, which can determine how many principal components need to be selected; (2) By calculating the ratio of the sum of the first k eigenvalues ​​to the sum of all eigenvalues, we can get the cumulative contribution rate R C : ; Among them, E jj Represents the jth diagonal element of the diagonal matrix E; this step is to check whether the selected principal component reaches the preset cumulative contribution rate threshold.

[0066] (3) The original data matrix d m Projected to a new feature space F consisting of the selected k principal components n In the new space, we can get the coordinates of each data point, that is, the principal component score P c :P c =d m *F k ; Among them, F k is the matrix composed of the eigenvectors corresponding to the selected k principal components; P c is the principal component score; this step is to convert the original data into a representation in a new coordinate system based on the principal component, so as to facilitate subsequent analysis and processing.

[0067] (4) Principal component score P c Normalization is performed to eliminate the dimensional differences between different principal component scores and make them comparable: ; Among them, P cn Represents the normalized value of the jth principal component score; this step is to ensure that the contribution of each principal component score is fair in the subsequent weighted summation.

[0068] (5) By taking the weighted sum of the variance contribution rate of each principal component, we can obtain the evaluation factor A for comprehensively evaluating the patient's underlying disease status: ; in, It represents the sum of the variance contribution rates of the first k principal components, and is used to calculate the weight of each principal component in the weighted summation. This step is to combine the information of each principal component to form an evaluation index that can comprehensively reflect the patient's underlying disease status.

[0069] It can be understood that through PCA processing, the original high-dimensional data is reduced to several main principal components, thereby simplifying the complexity of data analysis while retaining the main information of the data. Because PCA is an objective data analysis method, it is based on the statistical characteristics of the data, avoiding subjective interference from human factors and improving the accuracy and objectivity of the evaluation. Evaluation factor A, as a comprehensive evaluation indicator of the patient's underlying disease status, can provide key input for subsequent multi-source information fusion and hospital reputation evaluation, thereby improving the accuracy and reliability of the entire evaluation system.

[0070] Furthermore, the python execution program of the above step S2 is as follows: import numpy as np from sklearn.decomposition import PCA from sklearn.preprocessing import StandardScaler # The data matrix dm has been obtained in step S1 np.random.seed(0) dm = np.random.randn(10, 5) # 10 patients, 5 indicators # Step S200: Calculate the covariance matrix mean_vector = np.mean(dm, axis=0) cov_matrix = np.cov(dm, rowvar=False) # rowvar=False means that each column in the data matrix represents a variable # Step S201: Select the principal component and obtain the evaluation factor A # Perform PCA using scikit-learn's PCA class pca = PCA() # Standardize data scaler = StandardScaler() dm_standardized = scaler.fit_transform(dm) # Fit the PCA model and select principal components pca.fit(dm_standardized) # Set the cumulative contribution rate threshold (95%) cumulative_variance_ratio_threshold = 0.95 num_components = np.argmax(pca.explained_variance_ratio_.cumsum() >=cumulative_variance_ratio_threshold) + 1 # Extract the eigenvectors and principal component scores corresponding to the selected principal components principal_components = pca.components_[:num_components] principal_scores = pca.transform(dm_standardized)[:, :num_components] # Normalize the principal component scores principal_scores_normalized = (principal_scores - np.min(principal_scores, axis=0)) / (np.max(principal_scores, axis=0) - np.min(principal_scores, axis=0)) #According to the weighted sum of contribution rate, we get the evaluation factor A explained_variance_ratios = pca.explained_variance_ratio_[:num_components] evaluation_factor_A = np.sum(explained_variance_ratios * principal_scores_normalized, axis=1) # Output evaluation factor A print(evaluation_factor_A) In the above program, the numpy.cov function is used to calculate the data matrix d m covariance matrix of the data matrix. Note that the rowvar=False parameter means that each column in the data matrix represents a variable (i.e., indicator) and each row represents an observation (i.e., patient). Use scikit-learn's PCA class to perform PCA. First, standardize the data. Then fit the PCA model and obtain the variance contribution ratio of each principal component through the explained_variance_ratio_ attribute. Extract the eigenvectors and principal component scores corresponding to the selected principal components. Normalize the principal component scores to ensure that their contributions are fair when weighted summed. Weighted summation based on the contribution ratio is used to obtain the evaluation factor A for each patient.

[0071] For example, let a data matrix d m The data points are distributed as Figure 4 As shown in Part A of ; there are multiple dimensions in the original data, and the distribution of data points is difficult to observe directly in two-dimensional or three-dimensional space. PCA projects the data onto new coordinate axes through linear transformation, and these coordinate axes (i.e., principal components) are sorted according to the size of the data variance. In the scatter plot after dimensionality reduction (such as Part B of 4), the data points are redistributed along the direction of the principal components. And if there are obvious clustering or linear relationships in the original data, these relationships will be more obvious in the scatter plot after dimensionality reduction.

[0072] In this embodiment, regarding step S3, patient mortality assessment: In order to more accurately assess the medical quality of the hospital, especially for the key indicator of patient mortality, this method introduces the concept of mortality rate curve and combines it with the ordered clustering algorithm for analysis. This step aims to understand the patient mortality pattern in a more refined way through in-depth analysis of historical medical records, thereby providing a more objective hospital / physician reputation assessment. This step is mainly responsible for calculating the historical patient mortality rate curve D vc , using the ordered clustering algorithm based on the historical medical record data D' and the death rate curve D vcThe patients are clustered and the evaluation factor B is calculated. The specific scheme includes the following steps S300~S302. Figure 2 .

[0073] Specifically, step S300, collect historical medical record data D': first, collect medical records of patients with specific diseases De (such as heart disease, cancer, etc.) from the hospital information system, including the patient's admission date, discharge date (if applicable), death date (if applicable), etc. Encode these data to ensure that the data format is unified, especially the time format needs to be standardized, so as to facilitate subsequent analysis.

[0074] Based on the admission and discharge (or death) dates, the length of hospital stay for each patient is calculated, which is the basic data for understanding the progression of the disease and the effectiveness of treatment. The above information is integrated to form a historical medical record dataset D' containing key information such as disease type, death time, and length of hospital stay.

[0075] In practice, the factors of patient death are relatively complex. Simply taking mortality into consideration can easily lead to inaccurate data. Specifically, step S301 is performed to determine the time window and death time distribution: the time window T of the patient in the historical medical record data D' is set. w (For example, within one year after admission, to exclude the interference of death caused by long-term chronic diseases on the evaluation of short-term treatment effects), the month of death M d ; At the same time, for each historical patient, analyze different diseases De and their corresponding death time d t To achieve this goal, this step uses the exponential model to analyze the death time d under different diseases De. t Perform curve fitting to obtain the death rate curve D corresponding to the disease De vc : (1) Exponential model: D vc (t) = a·exp(b·t)+c; Among them, a, b and c are the parameters to be fitted; t is the time variable (days, months, etc.); the exponential model is chosen based on its ability to better capture the trend of growth or decay over time, and is suitable for describing the dynamic changes of disease progression and mortality risk.

[0076] (2) Use the least squares method to minimize the sum of the squared differences between the observed values ​​and the model predicted values: Right now ; Among them, Dvc(t i ) is the predicted value of the exponential model. N is the number of observations, i.e., at a specific time point t iWhether the patient has died (the number of deaths or the probability of death at each time point in the fitting process of the death rate curve can be regarded as an observed value); i It is the actual death data, including the time of death, cause of death (if related to the disease), etc. The actual death data is used to calculate the death rate curve, that is, by analyzing the deaths at different time points, to reveal the changing trend of the death rate over time.

[0077] It should be pointed out that the actual death data is the key input required to fit the death rate curve. It is essentially the same as the observed value, but is described from a different perspective. The observed value focuses more on the numerical performance at a specific time point, while the actual death data reflects the patient's death more comprehensively.

[0078] It is understandable that the exponential model is fitted and parameter estimated through relevant functions or libraries in statistical software or programming languages ​​(such as Python, R, etc.), and the above least squares problem is solved to obtain the values ​​of parameters a, b and c. And by introducing the death rate curve and exponential model, this method can more finely characterize the relationship between patient death and time. Compared with a single mortality rate indicator, it can better reflect the effectiveness and timeliness of medical intervention, thereby improving the accuracy of hospital / physician reputation evaluation. Time series analysis helps to discover abnormal fluctuations in mortality rates within a specific time period, which may indicate problems in medical processes, treatment plans or resource allocation, and provide improvement directions for hospital management.

[0079] Furthermore, the Python execution program of the above steps S300 to S301 is as follows: import pandas as pd import numpy as np from scipy.optimize import curve_fit import matplotlib.pyplot as plt # CSV file contains historical medical record data # Data columns include: patient_id, disease_type, admission_date, discharge_date (or death_date) data_file = 'historical_medical_records.csv' df = pd.read_csv(data_file) # Step S300: Collect historical medical record data D' df_disease = df[df['disease_type'] == 'heart_disease'] # Convert date format, calculate hospitalization time and death time df_disease['admission_date'] = pd.to_datetime(df_disease['admission_date']) df_disease['death_date'] = pd.to_datetime(df_disease['death_date'].fillna(df_disease['discharge_date'])) # If there is no death date, use the discharge date df_disease['hospital_stay'] = (df_disease['death_date'] - df_disease['admission_date']).dt.days # Only keep data of patients who died within one year after admission (time window Tw) df_disease_within_tw = df_disease[df_disease['hospital_stay'] <= 365] # Statistics of death month Md df_disease_within_tw['death_month'] = df_disease_within_tw['death_date'].dt.month # Step S301: Determine the time window and death time distribution, and fit the death speed curve Dvc # Aggregate data and count the number of deaths by month monthly_deaths = df_disease_within_tw.groupby('death_month').size() # Time variable t (month is used as the time unit here) t = monthly_deaths.index.values # Actual death data y (number of deaths per month) y = monthly_deaths.values # Define exponential model function def exp_model(t, a, b, c): return a * np.exp(b * t) + c # Use the curve_fit function to fit the curve and get the parameters a, b, c params, covariance = curve_fit(exp_model, t, y) a, b, c = params # Print fitting parameters print(f"Fitted parameters: a={a}, b={b}, c={c}") # Use the fitted model to predict the number of deaths y_pred = exp_model(t, a, b, c) # Calculate the sum of squared differences (i.e. the residual sum of squares) rss = np.sum((y - y_pred) ** 2) print(f"Residual Sum of Squares (RSS): {rss}") # Visualize the fitting results plt.scatter(t, y, label='Observed Deaths') plt.plot(t, y_pred, label='Fitted Death Speed ​​Curve', color='red') plt.xlabel('Month') plt.ylabel('Number of Deaths') plt.legend() plt.title('Death Speed ​​Curve Fitting') plt.show() In the above program, the pandas library is used to read historical medical record data in CSV format. Patient data for specific diseases (such as heart disease) are filtered. The date format is converted and the length of hospital stay and death time are calculated. If the date of death is missing, the discharge date is used instead (in actual applications, more complex logic may be required to handle this situation). Only patient data that died within one year of admission are retained to meet the setting of the time window Tw. An exponential model function exp_model is defined to describe the change in the death rate over time. The scipy.optimize.curve_fit function is used to fit the actual death data to obtain the optimal values ​​of the model parameters a, b, and c.

[0080] Then, step S302 is performed, ordered cluster analysis: an ordered clustering algorithm (Fisher optimal segmentation method) is used to analyze the historical medical record data D' and the death rate curve D vc Cluster the patients to obtain different risk groups. Quantify the evaluation indicators and calculate the evaluation factor B; The goal of Fisher's optimal segmentation method is to divide the n patients under each disease type De into K ordered groups (i.e., risk groups) so that the similarities within the ordered groups are maximized, while the differences between the ordered groups are maximized; through this process, we can quantitatively assess the risk level of patients and calculate the evaluation factor B, providing an objective basis for the subsequent hospital reputation evaluation: (1) Objective function: Let G1, G2, …, G K It is a segmentation that divides n patients into K groups, and the objective function is defined as: ; Among them, n i is the i-th group G i The number of patients in It is group G i The feature mean vector of the patients in , is the feature mean vector of all patients. This formula calculates the difference between each group and the overall mean and sums them weighted to reflect the overall differences between groups.

[0081] (2) Solving the optimal segmentation: Use the dynamic programming algorithm to solve the optimal segmentation. Define L(j,i) as the objective function value of the optimal segmentation that divides the first j patients into i groups. Use the dynamic programming algorithm to solve the optimal segmentation, and gradually calculate the objective function value of the optimal segmentation through the recursive relationship. The recursive relationship is: ; in, is the feature mean vector from patient s+1 to patient j; Represents the difference between the feature mean vector from patient s+1 to patient j and the feature mean vector of all patients. represents the transpose of the difference vector (in the case of real numbers, it is the vector itself; in the case of complex numbers or more general vector spaces, the conjugate operation is performed). s represents a limit on the patient number used to divide different patient groups. L(s,i−1) represents the objective function value of the optimal partitioning of the first s patients into i−1 groups.

[0082] (3) Initialization: Because there is only one group, there is no internal difference; so when i=1, L(j,1)=0; (4) Calculation: Starting from i=2, gradually increase the number of groups until the predetermined number of groups K is reached; (5) Obtaining the optimal segmentation: By backtracking the dynamic programming table, the optimal segmentation of n patients into K groups is obtained; (6) Quantification: ; in, It is group G i The proportion of patients in the total number of patients, R i It is group G i The (normalized) evaluation factor B reflects the overall risk level of the patient.

[0083] For example, the death rate curve D of six diseases De is set vc like Figure 5 As shown in part (A) of the diagram, the visualization effect of the groups segmented by the above clustering method is as follows Figure 5 As shown in area (B).

[0084] It is understandable that through ordered cluster analysis, patients can be divided into risk groups based on objective data such as their historical medical records and mortality rate curves, thus avoiding the bias of subjective assessment. By calculating the evaluation factor B, the overall patient risk level can be accurately quantified, providing reliable data support for hospital reputation assessment. Due to the use of dynamic programming algorithms, the number of groups K can be flexibly adjusted to meet the assessment needs in different scenarios, thereby improving the flexibility and applicability of the assessment. Through cluster analysis and quantitative assessment, hospitals and physicians can be provided with objective evidence of patient risk levels, assisting them in making more reasonable medical decisions.

[0085] Furthermore, the Python execution program of the above step S302 is as follows: import numpy as np def fisher_optimal_partitioning(data, K): """ Fisher's optimal partitioning method was used to cluster the data.

[0086] parameter: data (numpy.ndarray): patient feature data, shape is (n_samples, n_features) K (int): number of groups to be divided return: partitions (list of lists): list of patient indices for each group B (float): evaluation factor """ n = data.shape[0] #Number of patients # Initialize the dynamic planning table L = np.zeros((n + 1, K + 1)) # Initialize the split point record table for backtracking split_points = np.zeros((n + 1, K + 1), dtype=int) # Calculate the overall mean overall_mean = np.mean(data, axis=0) # Dynamic programming to solve the optimal segmentation for i in range(2, K + 1): # Start with 2 groups and gradually increase to K groups for j in range(i, n + 1): # Traverse the first j patients min_L = float('inf') best_s = 0 for s in range(i - 1, j): # Find the optimal split point s # Calculate the mean from s+1 to j mean_s_j = np.mean(data[s:j], axis=0) # Calculate the objective function value L_value = L[s, i - 1] + ((j - s) / n) * np.sum((mean_s_j - overall_mean) ** 2) # Update the minimum objective function value and the best split point if L_value < min_L: min_L = L_value best_s = s L[j, i] = min_L split_points[j, i] = best_s # Backtrack the dynamic planning table to get the optimal segmentation partitions = [] j = n for i in range(K, 0, -1): s = split_points[j, i] partitions.append(list(range(s + 1, j + 1))) j = s partitions.reverse() # Because it is backtracking from K to 1, the list needs to be reversed # You can subtract 1 from the index list for each group to match Python's 0-based indexing partitions = [[idx - 1 for idx in group] for group in partitions] # Calculate the evaluation factor B B = 0 for group in partitions: group_data = data[group] group_mean = np.mean(group_data, axis=0) group_risk_score = np.sum((group_mean - overall_mean) ** 2) #risk score B += (len(group) / n) * group_risk_score return partitions, B # Perform cluster analysis K = 3 # Divide into 3 groups partitions, B = fisher_optimal_partitioning(data, K) # Output results print("Partitions:", partitions) print("Evaluation Factor B:", B) In the above program, the dynamic programming table L and the split point record table split_points are initialized. The overall mean overall_mean is calculated. A two-layer loop is used to traverse the first j patients and the number of groups i. In the inner loop, the optimal split point s is found, the mean and objective function value from s+1 to j are calculated, and the minimum objective function value and the best split point are updated. The dynamic programming table is backtracked from K to 1 to obtain the optimal split group index list. Traverse each group and calculate the group mean and risk score (the example here simply uses the sum of the squares of the difference between the mean and the overall mean as the risk score). The evaluation factor B is calculated based on the group size and risk score.

[0087] In this embodiment, regarding step S4, the comprehensive reputation feature collection and quantification stage: the core goal of this stage is to collect and quantify the reputation features of hospitals or physicians in multiple dimensions, and then form a comprehensive reputation feature value S to fully reflect the medical reputation of hospitals or physicians. This process is subdivided into two sub-steps: S400 and S401.

[0088] Specifically, step S400 collects data on social reputation, treatment level, service attitude, and patient satisfaction: (1) Collect social reputation data (R1): Use authoritative ranking organizations or publicly available industry reports to collect social reputation information of hospitals or physicians. Convert the ranking into a numerical feature sequence R1. The higher the ranking, the higher the R1 value, indicating a better social reputation.

[0089] (2) Collect treatment level data (R2): Count the cure rate of hospitals or physicians, that is, the proportion of successfully cured cases to total cases. Convert the cure rate into the characteristic sequence R2. The higher the cure rate, the larger the R2 value, indicating a higher treatment level.

[0090] (3) Collect service attitude data (R3): Measure service attitude through the survey results of patient complaint rate. Convert the complaint rate into the characteristic sequence R3. The lower the complaint rate, the smaller the R3 value (or use reverse scoring so that low complaint rate corresponds to high R3 value), indicating better service attitude.

[0091] (4) Collect patient satisfaction data (R4): Directly use the patient satisfaction survey results. Convert the satisfaction score into the characteristic sequence R4. The higher the satisfaction, the larger the R4 value.

[0092] Specifically, step S401, determine the weights and calculate the comprehensive reputation feature value: determine the weights W1, W2, W3 and W4 corresponding to the feature sequences R1, R2, R3 and R4 according to the statistical method. Then calculate the comprehensive reputation feature value S: S=W1⋅R1+W2⋅R2+W3⋅R3+W4⋅R4.

[0093] The higher the S value, the better the overall reputation of the hospital or physician. By collecting and quantifying reputation characteristics in multiple dimensions, the medical reputation of a hospital or physician can be more comprehensively reflected, avoiding the one-sidedness of single indicator evaluation. Reputation information from different sources and of different natures is converted into a numerical feature sequence, and through weight distribution and comprehensive calculation, the objective quantification of reputation is achieved, which is convenient for comparison and ranking. The calculation process of the comprehensive reputation characteristic value S is transparent and traceable, and is based on actual data and statistical methods, which improves the credibility and persuasiveness of the evaluation results. The comprehensive reputation characteristic value S can serve as an important reference for patients to choose hospitals or physicians, helping patients make more informed medical decisions.

[0094] In this embodiment, regarding step S5, the DS evidence theory algorithm combination stage: the evaluation factor A and the evaluation factor B are combined using the DS evidence theory algorithm to form a treatment factor S with the patient's underlying disease evaluation information and the patient's mortality evaluation information. aving . See Figure 2 , the process includes defining the recognition framework and assigning the basic probability BPA (step S500), merging the basic probability BPA (step S501), and evaluating and making decisions (step S502).

[0095] Specifically, step S500, defining an identification framework and assigning a basic probability BPA: defining an identification framework I={LR, MR, HR}, where LR represents low risk, corresponding to a lower numerical range or probability interval, such as [0, 0.3]; MR represents medium risk, corresponding to an intermediate numerical range or probability interval, such as [0.3, 0.7]; HR represents high risk, corresponding to a higher numerical range or probability interval, such as [0.7, 1]; Set the scoring criteria for the underlying disease status, for example: mild (0-3 points), moderate (4-7 points), severe (8-10 points). Then assign the basic probability BPA to the evaluation factor A and the evaluation factor B, denoted by m respectively. A and m B ;in: (1) m A (LR) represents the support degree of evaluation factor A for low-risk propositions, m A (MR) represents the support degree of evaluation factor A for medium-risk propositions, m A(HR) indicates the support degree of evaluation factor A for high-risk propositions, that is, according to the patient's underlying disease status, a corresponding score S is given. A ; In order to calculate the support degree of each risk proposition, this embodiment selects to use a fuzzy membership function in the form of a bell-shaped function to describe the membership degree of elements in a fuzzy set, and its shape is similar to a normal distribution curve, and then calculates the support degree of the evaluation factor A: Define three bell-shaped functions to represent the membership of low-risk, medium-risk, and high-risk propositions respectively. Suppose the following parameters are used to define these functions: 1) The central point of the low-risk proposition (LR): c A,LR =2, width: σ A,LR =1.5; 2) The central point of the medium risk proposition (MR): c A,MR =5, width: σ A,MR =1.5; 3) The central point of high risk proposition (HR): c A,HR =8, width: σ A,HR =1.5; ; Among them, μ A,i (S A ) represents the membership of the evaluation factor A (i.e., the patient's underlying disease status) to the risk proposition i (i is low risk LR, medium risk MR, or high risk HR). Its output value is between 0 and 1, indicating that S A The degree of support for proposition X. C A,i Represents the center point of the bell-shaped function corresponding to risk proposition i. For different risk propositions (low risk, medium risk, high risk), the value of the center point will be different. It represents the most typical value of the evaluation factor A under the risk proposition. σ A,i Represents the width of the bell-shaped function corresponding to risk proposition i. The width determines the shape of the bell-shaped function, that is, the rate at which the membership changes with the value of the evaluation factor A. The larger the width, the flatter the function shape and the slower the membership changes; the smaller the width, the steeper the function shape and the faster the membership changes. The denominator is always positive by the operation of square sum plus 1, and when S A Close to c A,i When S A Stay away from c A,i When , the denominator value is larger, so the degree of membership is smaller.

[0096] Therefore, the support degree of evaluation factor A for low-risk, medium-risk and high-risk propositions are: m A (LR) = μ A,LR (S A ); m A (MR) = μ A,MR (S A ); m A (HR) = μ A,HR (S A ); (2) m B (LR) represents the support degree of evaluation factor B for low-risk propositions, m B (MR) represents the support degree of evaluation factor B for medium-risk propositions, m B (HR) represents the support degree of evaluation factor B for high-risk propositions; since its value range has been normalized to [0, 1], a similar bell-shaped function can be directly used to define the membership of low-risk, medium-risk and high-risk propositions, and give the corresponding score S B : 1) The central point of the low-risk proposition (LR): c B,LR =0.2, width: σ B,LR =0.15; 2) The central point of the medium risk proposition (MR): c B,MR =0.5, width: σ B,MR =0.15; 3) The central point of high risk proposition (HR): c B,HR =0.8, width: σ B,HR =0.15; ; The parameter signs are consistent with the membership function of evaluation factor A, and will not be repeated here. Therefore, the support degree of evaluation factor B for low-risk, medium-risk and high-risk propositions is: m B (LR) = μ B,LR (S B ); m B (MR) = μ B,MR (S B ); m B (HR) = μ B,HR (S B ); It should be noted that the support calculation should ensure that for any given score, the sum of the support for low-risk, medium-risk, and high-risk propositions is not 1, because the fuzzy membership function allows overlapping memberships. If the normalization requirement (i.e., the sum of the support is 1) needs to be met, other types of membership functions or post-processing can be considered. However, this step is essentially a fuzzy logic, and it is usually not mandatory that the sum of the memberships is 1.

[0097] Specifically, step S501, merging the basic probability BPA, the purpose of this step is to merge the BPA values ​​from different evaluation factors to form a comprehensive BPA value. The Dempster combination rule is a method for merging trust functions, which takes into account the overlap (i.e., intersection) between different trust functions; including: (1) Use Dempster's combination rule to merge m A and m B , calculate the normalization constant D: ; Among them, the symbol represents the empty set; X can be any of {LR}, {MR}, {HR}, {LR, MR}, {LR, HR}, {MR, HR}, or {LR, MR, HR}, and similarly, Y can also be one of these subsets. In Dempster's combination rule, X and Y represent subsets that are assigned non-zero probabilities by different belief functions (or BPAs).

[0098] (2) The combined basic probability BPA value mA,B is calculated as: ; Among them, Z is also a subset of the recognition framework I, which represents the merged BPA (i.e., m A,B ) are assigned a subset of non-zero probabilities. In the Dempster combination rule, Z is a certain combination of the intersection of X and Y that produces a new BPA during the merge process. However, Z is not chosen arbitrarily, but is determined by the intersection of X and Y. Specifically, we consider the product of the BPA values ​​of X and Y during the merge process if and only if the intersection of X and Y is equal to Z. Therefore, Z is a specific result of the intersection of X and Y, which reflects the degree of support for different risk propositions by the merged trust function.

[0099] Combined BPA value m A,B(Z) reflects the comprehensive support of evaluation factors A and B for risk proposition Z. This value is obtained by considering all possible combinations of X and Y (their intersection is equal to Z) and calculating the sum of the products of their BPA values. Ultimately, these combined BPA values ​​will be used to calculate the trust function Bel and the likelihood function Pl, and then form the saving factor Saving. Therefore, the connection between X, Y and Z is not only that they are subsets of the identification framework I, but also that they are interconnected through the Dempster combination rule and jointly affect the final decision result.

[0100] Specifically, step S502, evaluation and decision: calculate the trust function Bel and the likelihood function Pl to form the treatment factor S aving Here is how: (1) Construct the trust function Bel and likelihood function Pl: ; ; Among them, the symbol represents the empty set; (2) Based on the values ​​of the trust function Bel and the likelihood function Pl, the rescue factor S is formed aving The rule is that if the trust of the low-risk LR proposition (in layman's terms, optimistic expectations) is greater than the likelihood of the high-risk HR proposition (in layman's terms, pessimistic expectations), then it can be considered that the treatment effect is good, so the treatment factor S aving The value of will also be larger; the method flow is as follows: 1) Calculate the trust function Bel(LR): It represents the trust in proposition Z, which is the sum of the BPA values ​​of all subsets containing Z. For a low-risk proposition LR, its trust Bel(LR) is calculated as: ; Among them, m a,b (Y) is the basic probability distribution of subset Y after the evaluation factors A and B are combined; the trust function Bel(Z) represents the trust in proposition Z (for the scenario of this step, proposition Z refers to the low-risk LR subset), which is the sum of the BPA values ​​of all subsets containing Z.

[0101] 2) Calculate the likelihood function PI(HR): represents the likelihood of proposition Z, which is the sum of the BPA values ​​of all subsets that intersect with Z. For a high-risk proposition HR, its likelihood Pl(HR) is calculated as: ; It is necessary to consider all subsets that intersect {HR} and add up their BPA values, including the BPA value directly assigned to {HR} and the BPA value assigned to a larger subset that contains {HR} (such as {MR,HR} or {LR,MR,HR}), if any.

[0102] 3) S aving = Bel(LR) - Pl(HR); Among them, the rescue factor S aving The larger the value is, the greater the trust of the low-risk LR proposition is than the likelihood of the high-risk HR proposition, that is, the higher the optimistic expectation of low risk is, the better the treatment effect is, and the lower the overall risk after combining the patient's underlying disease assessment information and mortality assessment information.

[0103] It is understandable that the DS evidence theory algorithm combines the evaluation factors A and B from different sources and properties to improve the accuracy and comprehensiveness of hospital reputation evaluation. aving The introduction of S provides an intuitive and comparable indicator for quantifying the treatment effect. aving The value of directly reflects the effect of treatment, providing strong decision support for hospital managers and patients. aving The application of DS evidence theory algorithm effectively solves the uncertainty and conflict problems in multi-source information fusion and improves the efficiency and accuracy of information fusion.

[0104] Furthermore, the Python execution program of the above step S5 is as follows: # Step S500: Define the recognition framework and assign basic probability BPA def define_framework_and_bpa(): # Define the recognition framework framework = {'LR': 'Low Risk', 'MR': 'Medium Risk', 'HR': 'HighRisk'} # Assign basic probability BPA to evaluation factors A and B mA = {'LR': 0.6, 'MR': 0.2, 'HR': 0.2} # Evaluate the BPA of factor A mB = {'LR': 0.5, 'MR': 0.3, 'HR': 0.2} # Evaluate the BPA of factor B return framework, mA, mB # Step S501: Merge basic probability BPA def combine_bpa(mA, mB): # Initialize the merged BPA mA_B = {} # Calculate the normalization constant D D = sum(mA[x] * mB[y] for x in mA for y in mB if x == y) # Only consider the case where X=Y, that is, direct allocation # The combined BPA can be obtained directly by multiplying the corresponding items of mA and mB and normalizing them for x in mA: mA_B[x] = mA[x] * mB[x] / D # Consider the case of direct allocation return mA_B # Step S502: Evaluation and decision making def calculate_bel_and_pl(mA_B): # Calculate the trust function Bel and likelihood function Pl bel = {'LR': 0, 'MR': 0, 'HR': 0} pl = {'LR': 0, 'MR': 0, 'HR': 0} # Calculate Bel(LR) bel['LR'] = mA_B.get('LR', 0) # If 'LR' is not in mA_B, it returns 0 # For the likelihood function, we need to consider all subsets that intersect with 'HR' pl['HR'] = mA_B.get('HR', 0) # If 'HR' is not in mA_B, it returns 0 # Calculate the saving factor saving = bel['LR'] - pl['HR'] return bel, pl, saving # Main program def main(): # Execute step S500 framework, mA, mB = define_framework_and_bpa() # Execute step S501 mA_B = combine_bpa(mA, mB) # Execute step S502 bel, pl, saving = calculate_bel_and_pl(mA_B) # Output results print("Merged BPA:", mA_B) print("Trust function Bel:", bel) print("Likelihood function Pl:", pl) print("Saving factor:", saving) # Run the main program if __name__ == "__main__": main() In the above program, define_framework_and_bpa function defines the recognition framework I (including 'LR', 'MR', 'HR') and the basic probability BPA of evaluation factors A and B. The combine_bpa function is responsible for combining the BPA values ​​of evaluation factors A and B. The combined BPA value is obtained by multiplying the corresponding terms and normalizing them. The calculate_bel_and_pl function calculates the values ​​of the trust function Bel and the likelihood function Pl, and calculates the saving factor Saving based on these values. The main function calls the above functions in sequence and outputs the results of the combined BPA, trust function Bel, likelihood function Pl and saving factor Saving.

[0105] In this embodiment, in order to help patients choose hospitals or doctors more objectively and comprehensively, we designed a comprehensive consideration factor CCF (Comprehensive Consideration Factor) as a decision-making basis. That is, step S6, hospital / doctor selection stage: the treatment factor S aving Combined with the comprehensive reputation characteristic value S, it becomes a comprehensive consideration factor C that can help patients choose hospitals / doctors CF The specific steps include the following steps S600~S601. The purpose is to quantify the comprehensive performance of hospitals or physicians in terms of medical technology, service quality, reputation and word of mouth.

[0106] Specifically, step S600, determine the weight: use the entropy weight method to determine the treatment factor S aving The weight W5 of and the weight W6 of the comprehensive reputation feature value S: (1) Constructing the evaluation matrix EM: There are n hospitals / doctors to be evaluated. Each hospital / doctor has two evaluation indicators: treatment factor S aving And the comprehensive reputation eigenvalue S. Construct an n*2 evaluation matrix EM, where any element x of the evaluation matrix EM ij Represents the value of the i-th hospital / physician on the j-th evaluation index.

[0107] (2) Data standardization: Due to the treatment factor S aving Both the comprehensive reputation characteristic value S are positive indicators (the larger the value, the better), and the same standardization operation is used: ; Among them, x ij ′ is the standardized value of the i-th hospital / physician on the j-th evaluation index.

[0108] (3) For each evaluation index j, calculate the entropy value e j : ; Among them, p ij represents the proportion of the standardized value of the i-th hospital / physician on the j-th evaluation index ( ); k is a constant equal to , where n is the number of hospitals / physicians to be evaluated.

[0109] (4) For each evaluation index j, calculate its difference coefficient g j :g j =1−e j ; (5) For each evaluation index j, calculate the corresponding weight w j , to quantify the contribution of this indicator in the comprehensive consideration factor CCF: ; Specifically, step S601, calculate the comprehensive consideration factor: combined with the treatment factor S aving and the comprehensive reputation characteristic value S, and the comprehensive consideration factor C is calculated by weighted summation CF :CCF=W5*Saving+W6*S; Patients can choose hospitals / physicians based on the comprehensive consideration factor values. The higher the value, the better the hospital / physician performs in terms of medical technology, service quality, and reputation.

[0110] It is understandable that the weights are determined by the entropy weight method, which avoids the subjectivity of artificial weighting and makes the comprehensive consideration factor CCF more objective. At the same time, the treatment factor and the comprehensive reputation characteristic value are integrated to comprehensively evaluate the performance of hospitals / physicians from multiple dimensions. Through standardized operations and weighted summation, quantitative data from different sources and different natures are organically combined to form a unified comprehensive consideration factor CCF, which is convenient for patients to compare different hospitals / physicians. Moreover, the comprehensive consideration factor CCF provides patients with an intuitive decision-making basis. The higher the value, the better the performance of the hospital / physician in terms of medical technology, service quality and reputation, which helps patients make more informed choices.

[0111] Furthermore, the Python execution program of step S6 is as follows: import numpy as np n = 5 # Number of hospitals / doctors Saving = np.array([85, 90, 78, 88, 92]) # Saving factor example data S = np.array([80, 85, 75, 82, 88]) # Comprehensive reputation feature value example data # Step S600: Determine weight def calculate_weights(em): # Data Standardization min_vals = np.min(em, axis=0) max_vals = np.max(em, axis=0) em_standardized = (em - min_vals) / (max_vals - min_vals) # Calculate entropy k = 1 / np.log(n) p = em_standardized / np.sum(em_standardized, axis=0) e = -k * np.sum(p * np.log(p + 1e-10), axis=0) # Add 1e-10 to prevent log(0) # Calculate the coefficient of variation and weight g = 1 - e w = g / np.sum(g) return w # Construct evaluation matrix EM EM = np.column_stack((Saving, S)) # Calculate weights weights = calculate_weights(EM) W5, W6 = weights[0], weights[1] # Step S601: Calculate the comprehensive consideration factor CCF CCF = W5 * Saving + W6 * S # Output results print("The weight of the saving factor W5:", W5) print("The weight of the comprehensive reputation feature value S W6:", W6) print("Comprehensive consideration factor CCF:", CCF) In the above program, the np.column_stack function is used to combine Saving and S into an evaluation matrix EM with n rows and 2 columns. A calculate_weights function is defined to accept the evaluation matrix em as input and return the weights of each evaluation indicator. Inside the function, the data is first standardized to eliminate the dimension effect. The np.min and np.max functions are used to calculate the minimum and maximum values ​​of each evaluation indicator, and then the standardization formula is used to convert the original data into standardized data.

[0112] Next, calculate the entropy value. Use the np.sum function to calculate the sum of the standardized data on each evaluation indicator, and then calculate the proportion p of the standardized value of each hospital / physician on each evaluation indicator. Finally, use the entropy formula to calculate the entropy value e of each evaluation indicator. Calculate the coefficient of difference g and weight w. The coefficient of difference g is 1 minus the entropy value e, and the weight w is the coefficient of difference g divided by the sum of the coefficients of difference.

[0113] The comprehensive consideration factor CCF is calculated by weighted summation using the calculated weights W5 and W6, as well as the original saving factor and the comprehensive reputation characteristic value S.

[0114] Embodiment 2: Based on step S302 of embodiment 1, this embodiment further provides a preferred technical solution.

[0115] Considering that in practice, different diseases De have obvious seasonal characteristics (for example, the incidence of cardiovascular disease in winter is significantly higher than in other seasons), and different diseases De have acute and chronic differences, the objective function of step S302 of the first embodiment can be further optimized.

[0116] (1) Introducing the time variable: Let t i is the onset or diagnosis time of patient i, and the death rate curve D is located by this time variable vc The death rate curve D vc Denote as a function of time D vc (t), then for patient i, the corresponding death rate curve starts at D vc (t i ).

[0117] (2) Consider seasonal factors: Let S(t) be a seasonal factor function (such as a weight function or penalty function) that reflects the seasonal changes in the morbidity or mortality of the disease De at time t. It can be constructed based on historical data or expert knowledge.

[0118] (3) Optimized objective function: Combining the above two factors, the original objective function is optimized as follows: ; Among them, d i,obs represents the actual death time (or survival time) of patient i. i,pred (t i ,S(t i )) represents the death time (or survival time) of patient i predicted by the model, which depends on the onset or diagnosis time t i and seasonal factors S(t i ). i is the standard deviation of the death time (or survival time) of patient i, which is used to measure the prediction error. λ is the weight coefficient of the penalty term. Penalty(G k ) is the value of group G k A penalty factor to prevent the group from being too large or too small.

[0119] It is understandable that the seasonal factor function S(t) reflects the seasonal changes in the incidence or mortality of the disease De at time t. When predicting the patient's death time (or survival time), the model not only considers the individual characteristics of the patient, but also the onset or diagnosis time t i The corresponding seasonal factor S(t i In this way, the model can more accurately reflect the impact of seasonal changes on the survival of patients with certain diseases.

[0120] Acute patients usually develop rapidly and have a short course of illness, while chronic patients have a longer course of illness. Patients with different onset or diagnosis times will have different survival and mortality rate curves. Therefore, by introducing the time variable t i and seasonal factors S(t i ), which can capture the differences between acute and chronic patients. In other words, the main purpose of the objective function is to minimize the sum of the prediction error and the penalty term. The differences between acute and chronic patients can be reflected to a certain extent by incorporating seasonal factors and considering the individual characteristics of patients (including the type of disease course). For example, if the model can more accurately predict the survival status of patients with different disease course types, the value of the objective function will be smaller.

[0121] Furthermore, let L(j,k) represent the minimum objective function value for dividing the first j patients into k groups. We can construct a (n+1)*(K+1) dynamic programming table to store these values. For each j=2,3,…,n and k=2,3,…,K, we have: ; Where Cost(i+1,j) represents the objective function value of dividing patients i+1 to j into a group.

[0122] When the dynamic programming table is full, you can backtrack from L(n,K) to find the optimal split point. The specific steps are as follows: P1. Initialize an empty list partitions to store the patient index of each group.

[0123] P2. Let j=n, k=K.

[0124] P3. When k>1, find the i that minimizes L(j,k) (i.e., the optimal split point).

[0125] P4. Add the indexes of patients i+1 to j to the empty list partitions.

[0126] P5. Update j=i, k=k−1, and repeat the above steps P3~P4.

[0127] P6. When k=1, add the remaining patients (i.e., the first j patients) to a new group and add their indexes to partitions.

[0128] The Python program to implement the above solution is as follows: def backtrack_dp_table(dp_table, split_points): partitions = [] j, k = len(dp_table) - 1, len(dp_table[0]) - 1 while k > 1: i = split_points[j, k] # Find the optimal split point partitions.append(list(range(i + 1, j + 1))) # Add patient index to the group j, k = i, k - 1 partitions.append(list(range(1, j + 1))) # Add the remaining patients to a new group partitions.reverse() # Because we are backtracking from the back to the front, we need to reverse the list # The index returned here is 1-based, and 0-based indexing requires subtracting 1 return [[idx - 1 for idx in group] for group in partitions] # dp_table and split_points are filled by dynamic programming algorithm (see Example 1 for details) dp_table = ... split_points = ... # Get the optimal segmentation by: optimal_partitions = backtrack_dp_table(dp_table, split_points) Embodiment 3: This embodiment further provides a standard deviation weight assignment method for weights W1, W2, W3, and W4 in step S401 of embodiment 1. This method assigns weights according to the degree of dispersion (i.e., standard deviation) of each feature sequence data. Features with greater dispersion have a more important position in the comprehensive evaluation: S4010, for the feature sequences R1, R2, R3 and R4, calculate their standard deviations σ1, σ2, σ3 and σ4 respectively.

[0129] S4011, normalized standard deviation: Since the standard deviations of different feature sequences may have different dimensions, normalization is required. Divide each standard deviation by the sum of all standard deviations to obtain the normalized standard deviation ratio: ;where i=1,2,3,4, σ i ' is the normalized σ i .

[0130] S4012, assignment: W1=σ1 / [σ1+σ2+σ3+σ4]; W2=σ2 / [σ1+σ2+σ3+σ4]; W3=σ3 / [σ1+σ2+σ3+σ4]; W4=σ4 / [σ1+σ2+σ3+σ4]; Among them, all the above expressions represent the characteristic sequence R i The standard deviation σ i The ratio of the feature sequence to the sum of the standard deviations of all feature sequences.

[0131] The scheme of this embodiment allocates weights based on the variability of the data, and considers that features with large variability are more important in the comprehensive evaluation. This method avoids the subjectivity of artificial weighting and ensures the objectivity and rationality of the weights. The weight allocation is completely based on the characteristics of the data and is not interfered by human factors, which improves the objectivity of the evaluation. It can also capture small changes in the data, which has a positive impact on the accuracy of the evaluation results.

[0132] Embodiment 4: Figure 3 As shown, this embodiment discloses a patient condition assessment system based on a multi-source information fusion mechanism: the system is used to implement the patient condition assessment method based on a multi-source information fusion mechanism described in all the above embodiments, and the system includes: (1) User side, including an interactive interface through which patients can access the reputation information of hospitals / doctors.

[0133] (2) Server side, including: (2.1) The data collection module is responsible for automatically collecting the patient's basic disease indicator data set from the hospital information system: it has the functions of interface development, data capture, data cleaning and preprocessing.

[0134] (2.2) A data analysis module that uses principal component analysis to extract comprehensive indicators of patients' underlying diseases and uses ordered cluster analysis to assess patients' mortality risk: It has the functions of PCA algorithm implementation, ordered clustering algorithm implementation (Fisher optimal segmentation method), time series analysis and curve fitting.

[0135] (2.3) Feature quantification and weight allocation module responsible for executing the entropy weight method: It has the functions of formulating quantification standards, calculating weights using the entropy weight method, and weight allocation strategy.

[0136] (2.4) Fusion module responsible for implementing the DS evidence theory algorithm: using the Dempster combination rule to integrate the treatment factors (evaluation factors A and B) based on the patient's underlying disease and mortality risk with the hospital's comprehensive reputation characteristic value to form a comprehensive consideration factor. Including the implementation of the DS evidence theory algorithm, the calculation of the basic probability assignment (BPA), and the calculation of the trust function and likelihood function.

[0137] (2.5) Result interpretation module for evaluating hospital or physician reputation: interpreting the integrated consideration factors. The result interpretation module interprets the reputation information of the hospital / physician calculated in step S6 (i.e., the comprehensive consideration factor C CF ) is digitized and fed back to the user's interactive interface.

[0138] (3) Data end: including the existing hospital information system, the data collection module on the server side communicates with it and extracts relevant data information to the server side.

[0139] All the above embodiments only express the implementation methods of the relevant practical applications of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

[0140] For those skilled in the art, it can be further appreciated that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0141] At the same time, those skilled in the art can understand that all or part of the processes in all the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

Claims

1. A patient condition assessment method based on a multi-source information fusion mechanism, characterized in that: When a patient calls the medical reputation of a hospital / physician, after collecting the basic disease indicator dataset D from the hospital information system, the following steps are performed: S1, construct the basic disease indicator dataset D into a data matrix d with rows representing different historical patients and columns representing different indicators m ; S2, based on the data matrix d m Perform principal component analysis to obtain the evaluation factor A; S3, calculate the death rate curve D of historical patients vc , using the ordered clustering algorithm based on the historical medical record data D' and the death rate curve D vc Cluster the patients and calculate the evaluation factor B; S4, collects data on social reputation, treatment level, service attitude, and patient satisfaction to form a comprehensive reputation characteristic value S; S5, using the DS evidence theory algorithm to combine the evaluation factor A and the evaluation factor B to form a treatment factor S with the patient's basic disease assessment information and the patient's mortality assessment information aving ; S6, the rescue factor S aving Combined with the comprehensive reputation characteristic value S, it becomes a comprehensive consideration factor C that can help patients choose hospitals / doctors CF .

2. The method for assessing a patient's condition according to claim 1, characterized in that: In S2, the implementation method of the principal component analysis includes: S200, the data matrix d m After being converted into a covariance matrix C that describes the strength and direction of the linear relationship between the indicators, it is decomposed to obtain the eigenvalue E and the corresponding eigenvector F; S201, according to the size of the eigenvalue E, select the principal component P whose cumulative contribution rate reaches a preset threshold; then project the original data into a new feature space F formed by the selected principal component P n , calculate the new coordinates P of each data point c And perform normalization processing, and perform weighted summation according to the contribution rate to obtain the evaluation factor A.

3. The method for assessing a patient's condition according to claim 2, wherein: In S200, the covariance matrix C is: ； in, Denotes the data matrix d m The mean vector of Denotes the data matrix d m The i-th row in , n is the number of patients; T represents the transpose operation of the matrix; In the S200, the eigenvalue decomposition method is: C=F*E*F T ; Where T still represents the transpose operation of the matrix; In S201, the methods of the selection, the cumulative contribution rate, the projection, the normalization and the weighted summation are respectively as follows: I. The selection: Assume that k principal components are selected so that the cumulative contribution rate R C Reaching a preset threshold; II. The cumulative contribution rate R c : ; Among them, E jj represents the j-th diagonal element of the diagonal matrix E; m is the number of indices j; III. The projection: P c =d m *F k ; Among them, F k is the matrix composed of the eigenvectors corresponding to the selected k principal components; P c is the principal component score; F n is the new feature space; V. Normalization: ; Among them, P cn Represents the normalized value of the jth principal component score; VI. The weighted summation: ;in, Represents the sum of the variance contributions of the first k principal components.

4. The method for assessing a patient's condition according to claim 1, wherein: In S3, the time window T of the patient in the historical medical record data D' is set. w , and count the month of death M d ; For each historical patient, analyze different diseases De and their corresponding death time d t The distribution of the death rate curve D was obtained by time series analysis and curve fitting using the exponential model. vc ; The ordered clustering algorithm is the Fisher optimal segmentation method, which is based on the historical medical record data D' and the death rate curve D vc Clustering patients to obtain different risk groups; Quantify the evaluation index and calculate the evaluation factor B.

5. The method for assessing a patient's condition according to claim 4, characterized in that: In S3, the time series analysis and curve fitting methods are as follows: I. The index model: D vc (t)=a·exp(b·t)+c; where a, b and c are the parameters to be fitted; t is the time variable; II. Use the least squares method to minimize the sum of squared differences between the observed values ​​and the model's predicted values: ; where N is the number of observations; y i is the actual death data, Dvc(t i ) is the predicted value of the exponential model; solving the least squares problem to obtain the values ​​of parameters a, b and c.

6. The method for assessing a patient's condition according to claim 5, characterized in that: In S3, the method of the Fisher optimal segmentation method is to divide n patients into K ordered groups so that the similarity within the ordered groups is maximized and the difference between the ordered groups is maximized, and the method includes: I. Let G1, G2, …, G K It is a segmentation that divides n patients into K groups, and defines the objective function L(n,K) as: ； Among them, n i is the i-th group G i The number of patients in It is group G i The feature mean vector of the patients in , is the feature mean vector of all patients; II. Define L(j,i) as the objective function value of the optimal segmentation of the first j patients into i groups, and then obtain the following recursive relationship: ； in, is the feature mean vector from patient s+1 to patient j; represents the difference between the feature mean vector from patient s+1 to patient j and the feature mean vector of all patients; represents the transpose of the difference vector; s represents a limit of the patient number; L(s,i−1) represents the objective function value of the optimal segmentation of the first s patients into i−1 groups; III. Initialization: when i=1, L(j,1)=0; V. Starting from i=2, gradually increase the number of groups until the predetermined number of groups K is reached; VI. Obtain the optimal partitioning of n patients into K groups; VII. Quantification: ; in, It is group G i The proportion of patients in the total number of patients, R i is the group G i risk score.

7. The method for assessing a patient's condition according to claim 1, 2 or 4, characterized in that: In S5, the implementation method of the DS evidence theory algorithm includes: S500, define the identification framework I = {LR, MR, HR}, where LR represents low risk, MR represents medium risk, and HR represents high risk; then assign basic probabilities BPA to evaluation factors A and B, denoted as m and A and m B ; Among them, m A (LR) represents the support degree of evaluation factor A for low-risk propositions, m A (MR) represents the support degree of evaluation factor A for medium-risk propositions, m A (HR) represents the support degree of evaluation factor A for high-risk propositions; m B (LR) represents the support degree of evaluation factor B for low-risk propositions, m B (MR) represents the support degree of evaluation factor B for medium-risk propositions, m B (HR) indicates the support degree of evaluation factor B for high-risk propositions; S501, use Dempster's combination rule to combine the basic probability m A and the basic probability m B Combine and calculate the normalization constant D: ； Wherein, X and Y are subsets of the identification framework I; symbol represents the empty set; The combined basic probability BPA value m A,B Calculated as: ; Wherein, Z is also a subset of the identification framework I; S502, calculate the trust function Bel and the likelihood function Pl to form the rescue factor S aving : I. First calculate the trust function Bel and the likelihood function Pl: ； ； II. Then, based on the values ​​of the trust function Bel and the likelihood function Pl, the rescue factor S is formed. aving ; The rules are: S aving = Bel(LR)-Pl(HR); Among them, Bel(LR) represents the confidence of low-risk propositions, Pl(HR) represents the likelihood of high-risk propositions; the rescue factor S aving The larger the value, the better the treatment effect, and the lower the overall risk after combining the patient's underlying disease assessment information and mortality assessment information.

8. The method for assessing a patient's condition according to claim 1, 2 or 4, characterized in that: The implementation method of S6 includes: S600, using entropy weight method to determine the treatment factor S aving The weight W5 of and the weight W6 of the comprehensive reputation feature value S; S601, combined with rescue factor S aving and the comprehensive reputation characteristic value S, and the comprehensive consideration factor C is calculated by weighted summation CF :CCF=W5*Saving+W6*S.

9. The method for assessing a patient's condition according to claim 8, characterized in that: In S6, the implementation method of the entropy weight method includes: I. There are n hospitals / doctors to be evaluated. Each hospital / doctor has two evaluation indicators, namely, the treatment factor S aving and the comprehensive reputation characteristic value S; any element x of the evaluation matrix EM of n*2 ij represents the value of the i-th hospital / physician on the j-th evaluation index; II. Standardized operation: ; where x ij ′ is the standardized value of the i-th hospital / physician on the j-th evaluation index; III. For each evaluation index j, calculate its entropy value e j : ; Among them, p ij represents the proportion of the standardized value of the i-th hospital / physician on the j-th evaluation index; k is a constant; IV. For each evaluation index j, calculate its difference coefficient g j :g j =1−e j ; V. For each evaluation index j, calculate the corresponding weight w j : .

10. A system for implementing the patient condition assessment method according to any one of claims 1 to 9, characterized in that: include: The data collection module is responsible for automatically collecting the patient's basic disease indicator data set from the hospital information system; A data analysis module that uses principal component analysis to extract comprehensive indicators of patients' underlying diseases and an ordered cluster analysis to assess patients' mortality risk; Responsible for the feature quantization and weight allocation module that implements the entropy weight method; Responsible for implementing the fusion module of DS evidence theory algorithm; Results interpretation module to evaluate the reputation of a hospital or physician.

Images