WO2020119096A1 - Data analysis-based hospital evaluation method and related product - Google Patents

Abstract

A data analysis-based hospital evaluation method and a related product. The method is applied to an electronic device, and comprises: obtaining N preset risk indicators of a hospital to be evaluated according to a preset correspondence between hospitals and risk indicators, N being an integer greater than 1 (S101); combining the N preset risk indicators according to a preset rule to obtain a plurality of combination solutions (S102); sequentially inputting the plurality of combination solutions into a pre-trained combination solution recognition model to obtain a plurality of output results, and determining a target combination solution from the plurality of combination solutions according to the plurality of output results (S103); and evaluating said hospital according to the target combination solution (S104). The method facilitates improving the accuracy of evaluation of hospitals.

Description

Hospital evaluation method and related products based on data analysis

This application requires the priority of the Chinese patent application submitted to the China Patent Office on December 13, 2018 with the application number 2018115258014 and the application name "Database Analysis-based Hospital Evaluation Method and Related Products", the entire contents of which are incorporated by reference in In this application.

Technical field

This application relates to the field of big data technology, specifically to a hospital evaluation method and related products based on data analysis.

Background technique

With the continuous strengthening of the national basic medical security system and the continuous improvement of the economic level, the improvement of the life of the insured people, people are paying more and more attention to physical health, and gradually shifting their focus to seeing doctors and health care. , Leading to the continuous increase of medical expenses in hospitals. There are subjective and objective reasons for its growth. Among them, objective reasons include: improvement of living standards, enhancement of health awareness, and systematic health examinations attracting people’s attention; Heavier, leading to faster patient growth; the development and application of high-tech medical equipment, polymer medical materials, new drugs, special drugs, etc. increase medical costs. Subjective reasons: Hospitals and medical staff have some additional means such as indiscriminate use of drugs, indiscriminate charges, and indiscriminate inspections for their own benefit.

At present, there are many reasons for the increase in medical expenses, and some reasons are not the main reasons for the increase in medical expenses. Therefore, the rationality of the current evaluation of the increase in medical expenses of hospitals is not targeted, the accuracy is low, the form is single, and it is easy for the hospital Misjudged. There is an urgent need to provide a method for objectively evaluating the increase in hospital medical costs.

Summary of the invention

Embodiments of the present application provide a hospital evaluation method and related products based on data analysis, with a view to obtaining target risk indexes among preset risk indexes, and improving the pertinence and accuracy of hospital evaluation.

In a first aspect, an embodiment of the present application provides a hospital evaluation method based on data analysis. The method includes:

According to the correspondence between preset hospitals and risk indicators, obtain N preset risk indicators of the hospital to be evaluated, N is an integer greater than 1;

Combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

Inputting the plurality of combination schemes to a pre-trained combination scheme recognition model in sequence to obtain multiple output results, and determining a target combination scheme among the plurality of combination schemes according to the plurality of output results;

Evaluate the hospital according to the target combination plan.

In a second aspect, an embodiment of the present application provides a hospital evaluation electronic device based on data analysis, the electronic device includes:

An acquisition unit for acquiring N preset risk indicators of the hospital to be evaluated;

A combination unit, configured to combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

A determining unit, configured to sequentially input the multiple combination schemes into a pre-trained combination scheme recognition model to obtain multiple output results, and determine the target combination scheme among the multiple combination schemes according to the multiple output results;

The evaluation unit is used for evaluating the hospital according to the target combination plan.

In a third aspect, an embodiment of the present application provides an electronic device, including one or more processors, one or more memories, one or more transceivers, and one or more programs, where the one or more programs are Stored in the memory and configured to be executed by the one or more processors, the program includes instructions for performing the steps in the method as described in the first aspect.

According to a fourth aspect, an embodiment of the present application provides a computer-readable storage medium that stores a computer program for electronic data exchange, where the computer program causes a computer to execute the method according to the first aspect.

The implementation of the embodiments of the present application has the following beneficial effects:

It can be seen that in the embodiment of the present application, N preset risk indicators of the hospital are obtained, and the risk indicators are combined to obtain multiple combination schemes, and the multiple combination schemes are input into a pre-trained combination scheme recognition model to obtain The target combination plan evaluates the hospital based on the target combination plan, improves the accuracy of the hospital evaluation, provides data reference for the medical system reform, and improves the persuasion of the medical system reform.

BRIEF DESCRIPTION

1 is a schematic flowchart of a hospital evaluation method based on data analysis provided by an embodiment of the present application;

FIG. 1A is a schematic flowchart of a combination scheme based on a sliding window frame selection sub-matrix provided by an embodiment of the present application; FIG.

2 is a schematic flowchart of another hospital evaluation method based on data analysis provided by an embodiment of the present application;

3 is a schematic flowchart of another hospital evaluation method based on data analysis provided by an embodiment of the present application;

4 is a schematic structural diagram of a hospital evaluation electronic device based on data analysis provided by an embodiment of the present application;

5 is a block diagram of functional units of a hospital evaluation electronic device based on data analysis provided by an embodiment of the present application.

detailed description

The terms "first", "second", "third", and "fourth" in the description and claims of the present application and the accompanying drawings are used to distinguish different objects, not to describe a specific order . In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally also includes Other steps or units inherent to these processes, methods, products or equipment.

Referring to FIG. 1, FIG. 1 is a schematic flowchart of a hospital evaluation method based on data analysis provided by an embodiment of the present application. The method is applied to an electronic device, and the method includes steps S101 to S104:

Step S101: Acquire N preset risk indicators of the hospital to be evaluated according to the correspondence between preset hospitals and risk indicators, where N is an integer greater than 1.

Among them, the risk index is an index that affects the medical quality of the hospital, and specifically refers to an index that has an impact on the growth of the medical cost of the hospital.

Optionally, the corresponding relationship between the hospital and the risk index is set in advance. Specifically, the preset risk index is set according to the evaluation object and evaluation dimension, that is, different risks can be set for different evaluation objects and evaluation dimensions index.

For example, when the evaluation object is a hospital, and the evaluation dimension is the hospital’s medical expenses, the risk index can be set as the growth rate of the patient, the reimbursement ratio of medical insurance, the aging trend of the population, the scope of the medical insurance catalog, the number of beds in the hospital, etc. As another example, when the evaluation object is the cause, and the evaluation dimension is the increase in the medical cost of the cause, the risk index can be set to the medical research and development cost for the cause, the surgical cost when treating the cause, and the patient’s Care costs, etc. In addition, in the same evaluation dimension, different risk indicators of different hospitals can be set according to the information of the hospital, and the same risk index can be set for hospitals at the same latitude. For example, the risk index can be set according to the hospital level. When the medical expenses increase, the risk index of the first-class hospitals is set as the growth rate of patients, economic growth rate, medical insurance reimbursement ratio, and so on. This application does not limit the method of setting risk indicators.

This application takes the evaluation of the medical expenses of the hospital as an example for specific explanation, but it is not limited.

Step S102: Combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes.

Optionally, combining the N preset risk indicators according to a preset rule specifically includes: importing the N preset risk indicators as original input data into a database, using the crawler algorithm Python to call the database, and of course it can also be loaded by MATLAB In this database, it should be understood that Python is only an exemplary description and is not limited; the number of input rows m and columns n is obtained. Of course, the number of rows m and columns can be automatically set according to the number of N preset risk indicators For the size of n, set the size of m and n automatically to ensure that m*n is greater than or equal to N. It can also accept the size of the number of rows m and columns n entered by the user. Set the number of rows and columns of the matrix according to user needs Number, call the matrix generation function of Python, and use the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows as m and the number of columns as n, and then obtain the scale h of the input sliding window, Obtain the sliding window h*h, and obtain the input sliding step for the sliding window h*h; use the sliding window h*h to sequentially select multiple sub-matrices in the risk index matrix, so the sub-matrix The scale is also h*h, and all the elements in each sub-matrix of the multiple sub-matrices are used as a combined scheme to obtain the multiple combined schemes, where h≤m and h≤n.

Among them, the scale h can be set by the user, or can be automatically generated according to the values of m and n. In the automatic setting, h≤m and h≤n need to be satisfied, and this application does not make a unique limitation.

It can be understood that, when the sliding window h*h slides to the boundary of the risk indicator matrix m*n, that is, when the sliding window frame selects the last column or last row of the risk indicator matrix m*n, such as sliding The sliding window h*h will exceed the risk indicator matrix m*n, so you can set the sliding window to stop the sliding operation when sliding to the last column or last row of the risk indicator matrix m*n. All sub-matrices serve as the multiple combination schemes.

Further, in order to make full use of the risk indicators in the risk indicator matrix m*n, to obtain multiple combinations, when the sliding window frame selects the last column or last row of the risk indicator matrix m*n, continue to slide the sliding window h*h, perform padding operation at the same time, that is, execute the zero-add strategy for the part of the sliding window h*h that exceeds the risk indicator matrix m*n, until the sliding window h*h completely slides out of the risk indicator matrix m*n , Stop the sliding operation, and use all the sub-matrices selected by the frame as the multiple combination schemes.

Optionally, the total number of sub-matrices is obtained based on the above sliding operation, that is, the total number of combined schemes:

Among them, Q is the number of combined schemes, m is the number of rows of the risk indicator matrix, n is the number of columns of the risk indicator matrix, padding is the number of filled circles, h is the scale of the sliding window, and S is the sliding step.

For example, for example, FIG. 1A shows an example of a process of obtaining a combined solution by frame selection sub-matrix. As shown in FIG. 1A, assuming N=100, that is, there are 100 preset risk indicators in the hospital, which are respectively For K1, K2, K3, ... K100, enter m=10, n=10, then the scale of the risk index matrix is 10*10, set the sliding step size S=1, h=3, that is, the size of the sliding window is 3. *3, assuming no padding operation, ie padding = 0, so the number of sub-matrices available at this time is [(10+2*0-3+1)/1] ² = 64, that is, the 100 risks The index of 64 combination schemes.

Further, when using the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows m and the number of columns being n, first verify whether the input number of rows m and the number of columns n meet the requirements , That is, to determine whether the N preset risk indicators can be fully accommodated in the matrix m*n, the following operations need to be performed for the number m of rows and columns n that do not meet the requirements, which specifically include: as described above N=m*n, Use the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows as m and the number of columns as n; if the N<m*n, execute the zero-added strategy, for the N (M*nN) zeros are added to each risk indicator to ensure that the risk indicator after zero addition can fill the matrix m*n, and the risk indicator after zero addition is used as a matrix element to generate the number of rows as m and the number of columns as The risk index matrix m*n of n; if N>m*n, prompt to re-enter the number of rows m and the number of columns n, until the N≤m*n, the N risk indicators As a matrix element, a risk index matrix m*n in which the number of rows is m and the number of columns is n is generated.

Step S103: The multiple combination schemes are sequentially input into a pre-trained combination scheme recognition model to obtain multiple output results, and a target combination scheme among the multiple combination schemes is determined according to the multiple output results.

Optionally, sequentially inputting the plurality of combination schemes to the pre-trained combination scheme recognition model specifically includes: sequentially inputting medical data corresponding to the plurality of combination schemes as input data to the pre-trained combination scheme recognition model , Which specifically includes: acquiring a plurality of preset risk indicators in each of the plurality of combination plans, and determining an input data set and a verification set of each combination plan based on the medical database of the hospital, That is, the medical data at different times of the medical database is obtained, and the medical data corresponding to the multiple preset risk indicators in each combination plan are screened in the medical data, and the medical data corresponding to the multiple preset risk indicators are used as At this moment, the input data of the combination plan is obtained based on the medical database. The input data set of the combination plan is obtained; based on the medical database, the actual impact result set of the N preset risk indicators on the hospital (that is, the input data set is obtained) Corresponding to the changes in the hospital's medical expenses in the same period), the actual impact result set is used as the verification set; wherein, determining the target combination plan among the multiple combination plans according to the multiple output results specifically includes: The input data set of any one of the plurality of combination schemes is input to the pre-trained combination scheme recognition model, so as to obtain the prediction result of the combination scheme on the medical cost growth of the hospital, and the prediction result is verified with the Set fitting to get the first degree of fit, input the input data sets of multiple combined schemes to the pre-trained combined scheme recognition model in turn to obtain a prediction result set, and then, the prediction result set and the verification set Corresponding to the fitting, the fitting degree of the multiple combination schemes is obtained, and the combination scheme corresponding to the maximum fitting degree is taken as the target combination scheme. Or, obtain multiple fitting degrees whose fitting degree is greater than the fitting degree threshold, determine multiple combining schemes corresponding to the multiple fitting degrees, and use each of the multiple combining schemes as a set of things to obtain Multiple thing sets, use the multiple thing sets as a thing database, set the minimum support degree, determine the frequent item set in the thing database based on the FP-Growth algorithm and the minimum support degree, and use the frequent item set as the target combination scheme .

The fitting threshold may be 0.6, 0.7, 0.75, 0.8, or other values.

Optionally, the above fitting the prediction result to the verification result to obtain a degree of fit specifically includes: vectorizing the prediction result and the verification result to obtain a first feature vector of the prediction result, and a second feature corresponding to the verification result Vector, calculate the Euclidean distance between the first eigenvector and the second eigenvector, and use the Euclidean distance as the degree of fit between the prediction result and the verification result. Of course, calculating the Euclidean distance is just an example and can be determined by other calculation methods. The degree of fit between the prediction result and the verification result, for example, Euclidean distance, Chebyshev distance, etc.

Step S104: Evaluate the hospital according to the target combination plan.

Optionally, evaluating the hospital according to the target combination plan specifically includes: determining a plurality of preset evaluation dimensions of the hospital, acquiring multiple preset risk indicators in the target combination plan; determining the multiple Suppose that the risk index is scored on each of the plurality of preset evaluation dimensions to obtain multiple score results; weight the multiple score results according to the weight values of the multiple preset evaluation dimensions To get the score of the hospital. For example, Table 1 shows an example of a mapping relationship based on the target combination scheme to obtain the scoring results. Table 1 shows that when the preset indexes in the target combination scheme are K1, K2, and K3, respectively, the score result for the hospital is α*a+β*b+γ*c, α+β+γ=1.

Table 1

Wherein, the value range of the scoring result is 0-100, and the higher the score, the lower the risk index of the hospital in the increase of medical expenses.

It can be seen that in the embodiment of the present application, the N preset risk indicators of the hospital to be evaluated are obtained, the risk indicators are imported into the database, and the N risk indicators are combined using Python, which improves the diversity of the combination scheme, and, slides The window selection of the combination scheme has high randomness, avoiding the subjectivity of the artificial combination of preset risk indicators, making the target combination scheme in this application more convincing; in addition, based on medical data, multiple combination schemes are entered into the pre Recognize the model of the trained combination plan, determine the target combination plan according to the prediction results, and determine the target combination plan according to the model to obtain the high accuracy of the target combination plan; Finally, evaluate the hospital according to the target combination plan to improve the accuracy of the hospital evaluation. The medical system reform provides data reference and improves the persuasion of the medical system reform.

In a possible example, the method further includes:

Determine whether the scoring result is less than a threshold, and if so, send the target combination plan to the terminal device of the hospital to display the target combination plan on the information display interface of the terminal device and the prompt information, the prompt information It is used to prompt the hospital to adjust the medical system of the hospital based on the risk index in the target combination plan.

Wherein, the threshold may be 10, 20, 30, 50 or other values.

It can be seen that in this example, when the scoring result is less than the threshold, the scoring result is fed back to the terminal equipment of the hospital in time, prompting the hospital to make reforms to the medical system, improving the efficiency of the medical system reform, and the medical system reform is based on the combined plan The risk indicators in the list have improved the relevance of the medical system reform.

In a possible example, the method further includes:

Comparing the fitting degree of the multiple combination schemes with the fitting degree threshold, if the fitting degrees in the multiple combination schemes are all less than the fitting degree threshold, it is suggested that the multiple combination schemes are not satisfied Demand, re-enter the size of the sliding window or re-enter the sliding step size, so as to re-frame the sub-matrix, and re-obtain multiple combined solutions until the corresponding fitting degree of the new multiple combined solutions is greater than the fitting degree threshold, Otherwise, the operation of re-entering the size or sliding step of the sliding window is repeatedly performed.

It can be seen that in this example, by adjusting the size or sliding step size of the sliding window, a more accurate target combination plan is obtained, and the hospital is evaluated according to the target combination plan, which further improves the accuracy and pertinence of the hospital evaluation.

Referring to FIG. 2, FIG. 2 is a schematic flowchart of another hospital evaluation method based on data analysis provided by an embodiment of the present application. The method is applied to an electronic device. The method includes steps S201 to S207:

Step S201: Acquire N preset risk indicators of the hospital to be evaluated according to the correspondence between preset hospitals and risk indicators, where N is an integer greater than 1.

Step S202: Combine the N risk indicators according to a preset rule to obtain multiple combination schemes.

Step S203: Based on the medical database of the hospital, determine the training set and the verification set of each of the plurality of combination plans to obtain the training set and the verification set of the plurality of combination plans.

Wherein, the medical database is composed of medical data at multiple times, and the medical data includes medical data related to medical expenses and medical data related to the N preset risk indicators.

Optionally, determining the training set and verification set of each combination plan specifically includes: based on the medical database of the hospital, obtaining medical data corresponding to multiple risk indicators in each combination plan, and medical treatment corresponding to the multiple risk indicators The data is used as the training data of the combined scheme, the training data of the combined scheme at multiple times can be obtained to obtain the training data set of the combined scheme, and based on the medical database, the training data set of the multiple combined schemes can be obtained; based on the Medical database to obtain the actual impact results of the N preset risk indicators on the hospital (that is, the changes in the hospital's medical expenses in the same period corresponding to the training data set), and take the actual impact results on the hospital as the The verification data set of the Togo combination scheme, thereby obtaining the training data set and the verification data set of the multiple combination schemes.

Step S204: The training sets of the plurality of combination schemes are sequentially input to the initial model for training, and a trained combination scheme recognition model is obtained.

Optionally, input the training data set of any one of the multiple combined schemes to the initial model to perform a forward operation to obtain a predicted influence result set of the combined scheme, and the predicted influence result set and the combined scheme Verify the fitting of the data set to obtain the fit degree set of the combined scheme, calculate the average value of the fit degree set, and use the average value as the fit degree of the combined scheme. If the fit degree is greater than the first threshold, complete Train the initial model to obtain the pre-trained model, otherwise, perform reverse training on the initial model based on the loss function in the initial model, update the weight gradient in the initial model until the input of the combined scheme When the degree of fit between the prediction impact result set obtained from the training data set and the verification set is greater than the first threshold or the number of times of performing reverse training is greater than the second threshold, the training of the initial model is completed, and the Recognized model of the trained combined scheme.

Wherein, the first threshold may be 0.6, 0.7, 0.75, 0.8 or other values.

The second threshold may be 500, 1000, 3000, 5000, 10000, or other values.

Step S205: Based on the medical database, determine input data sets of the plurality of combination schemes, input the input data sets into the trained combination scheme recognition model, and obtain an output result.

Optionally, determining the input data set of the plurality of combination plans specifically includes: acquiring a plurality of preset risk indicators in each of the plurality of combination plans, and obtaining the medical database based on the medical database The most recently entered medical data in the medical data is obtained from the medical data corresponding to the multiple preset risk indicators in each combination plan, and the medical data corresponding to the multiple preset risk indicators is used as the current The input data of the combination plan to obtain the input data set of the multiple combination plans; based on the medical data input most recently, obtain the actual impact results of the N preset risk indicators on the hospital (that is, the most recent medical data input Changes in the hospital’s medical expenses), using the actual impact results as a verification set, inputting the input data of any one of the multiple combination schemes to the pre-trained combination scheme recognition model to obtain the output of the combination scheme As a result (that is, the prediction result of the medical cost increase of the hospital), the input data sets of the plurality of combination schemes are sequentially input to the pre-trained combination scheme recognition model to obtain respective output results.

Step S206: Determine a target combination plan among the plurality of combination plans according to the output result.

Optionally, the prediction results of the multiple combination schemes are respectively fitted to the verification set to obtain respective fitting degrees, and the combination scheme corresponding to the maximum fitting degree is used as the target combination scheme. Or, obtain a plurality of fitting degrees whose fitting degree is greater than a fitting degree threshold in the plurality of combining schemes, determine a plurality of combining schemes corresponding to the plurality of fitting degrees, and combine each of the plurality of combining schemes As a thing set, the solution obtains multiple thing sets, uses the multiple thing sets as a thing database, sets the minimum support degree, determines the frequent item set in the thing database based on the FP-Growth algorithm and the minimum support degree, and treats the frequent The item set serves as the target combination plan.

Optionally, fitting the prediction result to the verification result to obtain respective fitting degrees, specifically including: vectorizing the prediction result and the verification result to obtain a first feature vector of the prediction result, and a second corresponding to the verification result Feature vector, calculate the Euclidean distance between the first feature vector and the second feature vector, and use the Euclidean distance as the fit between the prediction result and the verification result. Of course, calculating the Euclidean distance is just an example, and other calculation methods can also be used. Determine the fit of the prediction results to the verification results, for example, Euclidean distance, Chebyshev distance, etc.

Wherein, the first threshold may be 0.6, 0.7, 0.75, 0.8 or other values.

Step S207: Evaluate the hospital according to the target combination plan.

It can be seen that in the embodiment of the present application, the N preset risk indicators of the hospital to be evaluated are obtained, the risk indicators are imported into the database, and the N risk indicators are combined using Python, which improves the diversity of the combination scheme, and, slides The window selection of the combination scheme has high randomness, avoiding the subjectivity of the artificial combination of preset risk indicators, making the target combination scheme in this application more convincing; in addition, based on medical data, multiple combination schemes are entered into the pre Recognize the model of the trained combination plan, determine the target combination plan according to the prediction results, and determine the target combination plan according to the model to obtain the high accuracy of the target combination plan; Finally, evaluate the hospital according to the target combination plan to improve the accuracy of the hospital evaluation. The medical system reform provides data reference and improves the persuasion of the medical system reform.

Referring to FIG. 3, FIG. 3 is a schematic flowchart of another hospital evaluation method based on data analysis provided by an embodiment of the present application. The method is applied to an electronic device, and the method includes steps S301 to S305:

Step S301: Acquire N preset risk indicators of the hospital to be evaluated according to the correspondence between preset hospitals and risk indicators, where N is an integer greater than 1.

Step S302: Combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes.

Step S303: The multiple combination schemes are sequentially input to the pre-trained combination scheme recognition model to obtain multiple output results, and the target combination scheme among the multiple combination schemes is determined according to the multiple output results.

Step S304: Evaluate the hospital according to the target combination plan, and obtain an evaluation result of the hospital.

Step S305: Send the evaluation result of the hospital to the network side device to display the score result of the hospital.

Optionally, establish an evaluation system for each hospital, embed the combined scheme identification model in the evaluation system, obtain the evaluation results of the hospital at different periods based on the medical data of the hospital at different periods, and then upload the evaluation results To the network-side device, so that the network-side device displays the evaluation results on the hospital’s information sharing platform (for example, the main page of the hospital’s official website), so that patients can query the hospital’s evaluation results of the increase in medical expenses on the information sharing platform, It provides data reference for patients to seek medical treatment. Among them, the evaluation results can be used to determine whether the increase in medical expenses is reasonable or whether the hospital's ability to regulate the increase in medical expenses or whether the increase in medical expenses matches the current economy, etc., and is not limited.

It can be seen that in the embodiment of the present application, the N preset risk indicators of the hospital to be evaluated are obtained, the risk indicators are imported into the database, and the N risk indicators are combined using Python, which improves the diversity of the combination scheme, and, slides The window frame selection combination scheme has high randomness, avoiding the subjectivity brought by the artificial combination of preset risk indicators, and making the target combination scheme more convincing; in addition, inputting multiple combination schemes into the pre-trained combination scheme recognition model To get the target combination plan, determine the target combination plan according to the model, and improve the accuracy; evaluate the hospital according to the target combination plan, improve the accuracy of the hospital evaluation, provide data reference for the medical system reform, and improve the persuasion of the medical system reform. Upload the evaluation results to the network-side device, display the evaluation results, and provide data references for patients to seek medical treatment.

Consistent with the above embodiments shown in FIGS. 1, 2, and 3, please refer to FIG. 4, which is a schematic structural diagram of a hospital evaluation electronic device 400 based on data analysis provided by an embodiment of the present application, as shown in FIG. 4 As shown, the electronic device 400 includes a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are different from the one or more application programs, and the one or more programs are stored in In the above memory, and configured to be executed by the above processor, the above program includes instructions for performing the following steps;

According to the correspondence between preset hospitals and risk indicators, obtain N preset risk indicators of the hospital to be evaluated, N is an integer greater than 1;

Combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

Inputting the plurality of combination schemes to a pre-trained combination scheme recognition model in sequence to obtain multiple output results, and determining a target combination scheme among the plurality of combination schemes according to the plurality of output results;

Evaluate the hospital according to the target combination plan.

In a possible example, in terms of combining the N preset risk indicators according to a preset rule to obtain multiple combination schemes, the instructions in the above program are specifically used to perform the following operations:

Import the N preset risk indicators into the database; use the crawler algorithm Python to call the database to obtain the number of input rows m and columns n; call the matrix generation function of Python to generate the N risk indicators as matrix elements The risk index matrix m*n where the number of rows is the m and the number of columns is the n; obtain the scale h of the input sliding window to obtain the sliding window h*h; obtain the input sliding against the sliding window h*h Step; slide the sliding window h*h in the risk index matrix m*n in sequence according to the sliding step, select multiple sub-matrices in the frame, and select all elements in each of the multiple sub-matrices As a combination scheme, multiple combination schemes are obtained, where h≤m and h≤n.

In a possible example, in terms of generating the risk index matrix m*n with the number of rows as the m and the number of columns as the matrix elements using the N risk indicators as matrix elements, the instructions in the above program are specifically used to execute The following operations:

As the N=m*n, the N risk indicators are used as matrix elements to generate a risk indicator matrix m*n with the number of rows as the m and the number of columns as the n; if the N<m*n, Implement a zero addition strategy, add (m*nN) zeros to the N risk indicators, and use the added risk indicators as matrix elements to generate a risk indicator matrix with the number of rows and the number of columns as n m*n; if N>m*n, prompt to re-enter the number of rows m and the number of columns n, until the N≤m*n, use the N risk indicators as matrix elements to generate the number of rows as The m and the number of columns are the risk index matrix m*n of the n.

In a possible example, in evaluating the hospital according to the target combination scheme, the instructions in the above procedure are specifically used to perform the following operations:

Determining multiple preset evaluation dimensions corresponding to the hospital; acquiring multiple preset risk indicators in the target combination scheme; determining each of the multiple preset risk indicators in the multiple preset evaluation dimensions A plurality of scoring results are obtained by scoring results in a preset evaluation dimension; the plurality of scoring results are weighted according to weight values of the plurality of preset evaluation dimensions to obtain a scoring result for the hospital.

In a possible example, the instructions in the above program are also used to perform the following operations:

Determine whether the scoring result is less than a threshold, and if so, send the target combination plan to the terminal device of the hospital to display the target combination plan on the information display interface of the terminal device and the prompt information, the prompt information It is used to prompt the hospital to adjust the medical system of the hospital based on the risk index in the target combination plan.

In a possible example, the instructions in the above program are also used to perform the following operations:

Based on the medical database of the hospital, obtain medical data corresponding to multiple risk indicators in any one of the multiple combined solutions, and determine that the medical data corresponding to the multiple risk indicators is the training of the combined solution A data set to obtain the training data set of the multiple combined solutions; based on the medical database, obtain the actual impact results of the N preset risk indicators on the hospital, and determine that the actual impact results are the multiple The verification set of the combined scheme; input the training data set of any one of the multiple combined schemes into the initial model to perform a forward operation to obtain the predicted influence result set of the combined scheme, and then the predicted influence result set Fitting with the verification set of the combination scheme to obtain a fit degree corresponding to the combination scheme, performing reverse training according to the fit degree, and obtaining the pre-trained combination scheme recognition model.

In a possible example, in fitting the prediction impact result set to the verification set of the combined scheme to obtain a fitting degree corresponding to the combined scheme, the instructions in the above program are specifically used to perform the following operations: Vectorizing the prediction result and the verification result to obtain a first feature vector corresponding to the prediction result and a second feature vector corresponding to the verification result; calculating the first feature vector and the second feature For the Euclidean distance of the vector, use the Euclidean distance as the degree of fit between the prediction result and the verification result.

In a possible example, in performing reverse training according to the fitting degree to obtain the pre-trained combined scheme recognition model, the instructions in the above program are specifically used to perform the following operations:

If the degree of fit is greater than the first threshold, complete the training of the initial model to obtain the pre-trained combination scheme recognition model, otherwise, perform an inversion on the initial model based on the loss function in the initial model To training, update the weight gradient in the initial model until the degree of fit between the predicted impact result set obtained by inputting the training data set and the verification set is greater than the first threshold or the number of times of performing reverse training is greater than At the second threshold, the training of the initial model is completed, and the pre-trained combination scheme recognition model is obtained.

In a possible example, in order to sequentially input the multiple combination schemes into a pre-trained combination scheme recognition model and obtain multiple output results, the instructions in the above program are specifically used to perform the following operations:

The medical data corresponding to the multiple combination schemes are sequentially input as input data to the pre-trained combination scheme recognition model to obtain a prediction result corresponding to each combination scheme, and the prediction result is medical treatment for the hospital Forecast results of cost increase.

Referring to FIG. 5, FIG. 5 shows a block diagram of a possible functional unit composition of the electronic device 500 of the hospital evaluation method based on data analysis involved in the above embodiment. The electronic device 500 includes an acquisition unit 510, a combination unit 520, and a determination Unit 530, evaluation unit 540, among them;

The obtaining unit 510 obtains N preset risk indicators of the hospital to be evaluated according to the correspondence between preset hospitals and risk indicators, where N is an integer greater than 1;

A combination unit 520, configured to combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

The determining unit 530 is configured to sequentially input the multiple combination schemes into a pre-trained combination scheme recognition model to obtain multiple output results, and determine the target combination scheme among the multiple combination schemes according to the multiple output results ;

The evaluation unit 540 is used to evaluate the hospital according to the target combination plan.

In a possible example, in terms of combining the N preset risk indicators according to a preset rule to obtain multiple combination solutions, the combining unit 520 is specifically configured to: import the N preset risk indicators into a database; and It is used to call the database using the crawler algorithm Python to obtain the input row number m and column number n; and to call Python's matrix generation function, using the N risk indicators as matrix elements to generate the row number for the m and The number of columns is the risk index matrix m*n of the n; and the size h of the sliding window used to obtain the input to obtain the sliding window h*h; and the sliding step for the input to the sliding window h*h Long; and used to sequentially slide the sliding window h*h in the risk index matrix m*n according to the sliding step, select a plurality of sub-matrixes, and select the All elements are used as one combination plan to obtain multiple combination plans, where h≤m and h≤n.

In a possible example, in terms of generating the risk index matrix m*n with the number of rows and the number of columns using the N risk indicators as matrix elements, the combination unit 520 is specifically used for: The N=m*n, using the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows as the m and the number of columns as the n; and used as N<m* n, execute a zero-added strategy, add (m*nN) zeros to the N risk indicators, and use the added risk indicators as matrix elements to generate a risk with the number of rows and the number of columns as n Index matrix m*n; and for N>m*n as mentioned above, prompting to re-enter the number of rows m and columns n, until the N≤m*n, the N risk indicators as a matrix The risk index matrix m*n where the number of element generation rows is m and the number of columns is n.

In a possible example, in evaluating the hospital according to the target combination scheme, the evaluation unit 540 is specifically configured to: determine a plurality of preset evaluation dimensions corresponding to the hospital; and obtain the target combination scheme A plurality of preset risk indicators in; and a score result for determining the plurality of preset risk indicators in each of the plurality of preset evaluation dimensions to obtain multiple score results; and It is used to weight the plurality of scoring results according to the weight values of the plurality of preset evaluation dimensions to obtain a score result for the hospital.

In a possible example, the electronic device 500 further includes a judgment unit 550 for judging whether the scoring result is less than a threshold, and if so, sending the target combination plan to the terminal device of the hospital to The information display interface of the terminal device displays the target combination plan and prompt information, and the prompt information is used to prompt the hospital to adjust the medical system of the hospital based on the risk index in the target combination plan.

In a possible example, the electronic device 500 further includes a training unit 560, which is used to obtain correspondences between multiple risk indicators in any one of the multiple combination schemes based on the medical database of the hospital Medical data corresponding to the multiple risk indicators, determine the medical data corresponding to the multiple risk indicators as the training data set of the combined scheme, and obtain the training data set of the multiple combined schemes; and for obtaining the N based on the medical database An actual impact result of a preset risk index on the hospital, determining that the actual impact result is a verification set of the multiple combined schemes; and training data for combining any one of the multiple combined schemes The set is input to the initial model to perform a forward operation to obtain a predicted impact result set of the combined scheme, and the predicted impact result set is fitted to the combined set verification set to obtain a degree of fit corresponding to the combined scheme, Perform reverse training according to the fitting degree to obtain the pre-trained combination scheme recognition model.

In a possible example, in fitting the prediction impact result set to the verification set of the combined scheme to obtain a fitting degree corresponding to the combined scheme, the training unit 560 is specifically configured to: The result is vectorized with the verification result to obtain a first feature vector corresponding to the prediction result and a second feature vector corresponding to the verification result; calculating the Euclidean distance between the first feature vector and the second feature vector , Using the Euclidean distance as the degree of fit between the prediction result and the verification result.

In a possible example, in performing reverse training according to the degree of fit to obtain the pre-trained combination scheme recognition model, the training unit 560 is specifically configured to: if the degree of fit is greater than the first threshold , Complete the training of the initial model, and obtain the pre-trained combination scheme recognition model, otherwise, perform reverse training on the initial model based on the loss function in the initial model, update the initial model Weight gradient until the fit between the predicted impact result set obtained by inputting the training data set and the verification set is greater than the first threshold or the number of times of performing reverse training is greater than the second threshold, the Model training to obtain the pre-trained combination scheme recognition model.

In a possible example, in order to input the multiple combination schemes to the pre-trained combination scheme recognition model in turn and obtain multiple output results, the determining unit 530 is specifically configured to: correspond to the multiple combination schemes The medical data of is sequentially input as input data to the pre-trained combination scheme recognition model to obtain a prediction result corresponding to each combination scheme, and the prediction result is a prediction result of the medical cost increase of the hospital.

An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, and the computer program enables the computer to execute any of the methods described in the above method embodiments based on Some or all steps of the hospital evaluation method for data analysis.

An embodiment of the present application further provides a computer program product, the computer program product includes a non-transitory computer-readable storage medium storing a computer program, the computer program is operable to cause the computer to execute as described in the above method embodiments Some or all steps of any hospital evaluation method based on data analysis.

It should be noted that, for the sake of simple description, the foregoing method embodiments are all expressed as a series of action combinations, but those skilled in the art should know that this application is not limited by the described action sequence, Because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the involved actions and modules are not necessarily required by this application.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or software program modules.

If the integrated unit is implemented in the form of a software program module and sold or used as an independent product, it may be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application essentially or part of the contribution to the existing technology or all or part of the technical solution can be embodied in the form of a software product, the computer software product is stored in a memory, Several instructions are included to enable a computer device (which may be a personal computer, server, network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. The aforementioned memory includes: U disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

The embodiments of the present application are described in detail above, and specific examples are used to explain the principles and implementation of the present application. The descriptions of the above embodiments are only used to help understand the method and core idea of the present application; at the same time, Those of ordinary skill in the art, based on the ideas of the present application, will have changes in specific implementations and application scopes. In summary, the content of this specification should not be construed as limiting the present application.

Claims (20)

1. A hospital evaluation method based on data analysis, characterized in that the method includes:

   According to the correspondence between preset hospitals and risk indicators, obtain N preset risk indicators of the hospital to be evaluated, N is an integer greater than 1;

   Combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

   Inputting the plurality of combination schemes to a pre-trained combination scheme recognition model in sequence to obtain multiple output results, and determining a target combination scheme among the plurality of combination schemes according to the plurality of output results;

   Evaluate the hospital according to the target combination plan.
2. The method according to claim 1, wherein the combining of the N preset risk indicators according to a preset rule to obtain multiple combination schemes specifically includes:

   Import the N preset risk indicators into the database;

   Use the crawler algorithm Python to call the database to obtain the number of input rows m and columns n;

   Invoking Python's matrix generation function, using the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows as the m and the number of columns as the n;

   Obtain the scale h of the input sliding window to obtain the sliding window h*h;

   Obtaining the input sliding step size for the sliding window h*h;

   Sliding the sliding window h*h sequentially in the risk index matrix m*n according to the sliding step, selecting a plurality of sub-matrices, and using all elements in each sub-matrix of the plurality of sub-matrices as a combination Scheme, a plurality of combined schemes are obtained, where h≤m and h≤n.
3. The method according to claim 2, wherein the generating of a risk index matrix m*n with the number of rows and the number of columns using the N risk indicators as matrix elements specifically includes:

   If the N=m*n, the N risk indicators are used as matrix elements to generate a risk indicator matrix m*n with the number of rows and the number of columns as n;

   If the N<m*n, execute a zero-add strategy, add (m*nN) zeros to the N risk indicators, and use the added risk indicators as matrix elements to generate rows for the m and columns The number is the risk index matrix m*n of n;

   If the N>m*n, prompt to re-enter the number of rows m and the number of columns n, until the N≤m*n, use the N risk indicators as matrix elements to generate the number of rows for the m and n The number of columns is the risk index matrix m*n of n.
4. The method according to claim 1, wherein the evaluation of the hospital according to the target combination scheme specifically includes:

   Determine multiple preset evaluation dimensions corresponding to the hospital;

   Acquiring multiple preset risk indicators in the target combination scheme;

   Determining the score results of the multiple preset risk indicators on each of the multiple preset evaluation dimensions, to obtain multiple score results;

   Weighting the plurality of scoring results according to the weight values of the plurality of preset evaluation dimensions to obtain a scoring result for the hospital.
5. The method according to claim 4, wherein the method further comprises:

   Judging whether the scoring result is less than the threshold, if so, sending the target combination plan to the terminal device of the hospital to display the target combination plan on the information display interface of the terminal device and the prompt information, the prompt information It is used to prompt the hospital to adjust the medical system of the hospital based on the risk index in the target combination plan.
6. The method according to claim 5, wherein the method further comprises:

   Based on the medical database of the hospital, obtain medical data corresponding to multiple risk indicators in any one of the multiple combined solutions, and determine that the medical data corresponding to the multiple risk indicators is the training of the combined solution A data set to obtain the training data set of the multiple combined schemes;

   Based on the medical database, acquiring the actual impact results of the N preset risk indicators on the hospital, and determining the actual impact results as the verification set of the multiple combined schemes;

   Input the training data set of any one of the plurality of combination schemes into the initial model to perform a forward operation to obtain a predicted impact result set of the combined scheme, and combine the predicted influence result set with the combined scheme Verify the fitting of the set to obtain the degree of fit corresponding to the combination scheme, and perform reverse training according to the degree of fit to obtain the pre-trained combination scheme recognition model.
7. The method according to claim 6, wherein the fitting of the predicted impact result set to the verification set of the combined scheme to obtain a fitting degree corresponding to the combined scheme specifically includes:

   Vectorizing the prediction result and the verification result to obtain a first feature vector corresponding to the prediction result and a second feature vector corresponding to the verification result;

   Calculate the Euclidean distance between the first feature vector and the second feature vector, and use the Euclidean distance as the degree of fit between the prediction result and the verification result.
8. The method according to claim 6, wherein the performing reverse training according to the fitting degree to obtain the pre-trained combination scheme recognition model specifically includes:

   If the degree of fit is greater than the first threshold, complete the training of the initial model to obtain the pre-trained combination scheme recognition model, otherwise, perform an inversion on the initial model based on the loss function in the initial model To training, update the weight gradient in the initial model until the degree of fit between the predicted impact result set obtained by inputting the training data set and the verification set is greater than the first threshold or the number of times of performing reverse training is greater than At the second threshold, the training of the initial model is completed, and the pre-trained combination scheme recognition model is obtained.
9. The method according to claim 8, wherein the sequentially inputting the plurality of combination schemes to a pre-trained combination scheme recognition model, and obtaining a plurality of output results specifically include:

   The medical data corresponding to the multiple combination schemes are sequentially input as input data to the pre-trained combination scheme recognition model to obtain a prediction result corresponding to each combination scheme, and the prediction result is medical treatment for the hospital Forecast results of cost increase.
10. A hospital evaluation electronic device based on data analysis, characterized in that the electronic device includes:

    An acquisition unit for acquiring N preset risk indicators of the hospital to be evaluated;

    A combination unit, configured to combine the N preset risk indicators according to a preset rule to obtain multiple combination schemes;

    A determining unit, configured to sequentially input the multiple combination schemes into a pre-trained combination scheme recognition model to obtain multiple output results, and determine the target combination scheme among the multiple combination schemes according to the multiple output results;

    The evaluation unit is used for evaluating the hospital according to the target combination plan.
11. The electronic device according to claim 10, characterized in that, in combining the N preset risk indicators according to a preset rule to obtain multiple combination schemes, the combination unit is specifically used to:

    Import the N preset risk indicators into the database;

    Use the crawler algorithm Python to call the database to obtain the number of input rows m and columns n;

    Invoking Python's matrix generation function, using the N risk indicators as matrix elements to generate a risk indicator matrix m*n with the number of rows as the m and the number of columns as the n;

    Obtain the scale h of the input sliding window to obtain the sliding window h*h;

    Obtaining the input sliding step size for the sliding window h*h;

    Sliding the sliding window h*h sequentially in the risk index matrix m*n according to the sliding step, selecting a plurality of sub-matrices, and using all elements in each sub-matrix of the plurality of sub-matrices as a combination Scheme, a plurality of combined schemes are obtained, where h≤m and h≤n.
12. The electronic device according to claim 11, characterized in that

    In terms of generating the risk index matrix m*n with the number of rows as the m and the number of columns as the matrix elements using the N risk indicators as matrix elements, the combination unit is specifically used for:

    If the N=m*n, the N risk indicators are used as matrix elements to generate a risk indicator matrix m*n with the number of rows and the number of columns as n;

    If the N<m*n, execute a zero-add strategy, add (m*nN) zeros to the N risk indicators, and use the added risk indicators as matrix elements to generate rows for the m and columns The number is the risk index matrix m*n of n;

    If the N>m*n, prompt to re-enter the number of rows m and the number of columns n, until the N≤m*n, use the N risk indicators as matrix elements to generate the number of rows for the m and n The number of columns is the risk index matrix m*n of n.
13. The electronic device according to claim 10, characterized in that

    In terms of evaluating the hospital according to the target combination plan, the evaluation unit is specifically used to:

    Determine multiple preset evaluation dimensions corresponding to the hospital;

    Acquiring multiple preset risk indicators in the target combination scheme;

    Determining the score results of the multiple preset risk indicators on each of the multiple preset evaluation dimensions, to obtain multiple score results;

    Weighting the plurality of scoring results according to the weight values of the plurality of preset evaluation dimensions to obtain a score result for the hospital.
14. The electronic device according to claim 13, wherein the electronic device further comprises a judgment unit;

    The judging unit is used to judge whether the scoring result is less than a threshold, if so, send the target combination plan to the terminal device of the hospital to display the target combination plan on the information display interface of the terminal device, and Prompt information, which is used to prompt the hospital to adjust the medical system of the hospital based on the risk index in the target combination plan.
15. The electronic device according to claim 14, wherein the electronic device further comprises a training unit;

    The training unit is configured to obtain medical data corresponding to multiple risk indicators in any one of the multiple combination plans based on the medical database of the hospital, and determine medical data corresponding to the multiple risk indicators For the training data set of the combined plan, obtain the training data set of the multiple combined plans; based on the medical database, obtain the actual impact results of the N preset risk indicators on the hospital, and determine the actual The impact result is the verification set of the multiple combination schemes; input the training data set of any one of the multiple combination schemes to the initial model to perform a forward operation to obtain the predicted influence result set of the combination scheme, Fitting the prediction impact result set to the verification set of the combined scheme to obtain a fit degree corresponding to the combined scheme, performing reverse training according to the fit degree, and obtaining the pre-trained combined scheme identification model.
16. The electronic device according to claim 15, characterized in that

    In terms of fitting the prediction impact result set to the verification set of the combined scheme to obtain a fitting degree corresponding to the combined scheme, the training unit is specifically configured to: match the predicted result with the verification result Vectorize to obtain the first feature vector corresponding to the prediction result and the second feature vector corresponding to the verification result; calculate the Euclidean distance between the first feature vector and the second feature vector, and convert the Euclidean distance As the degree of fit between the prediction result and the verification result.
17. The electronic device according to claim 15, characterized in that

    In performing reverse training according to the degree of fit to obtain the pre-trained combination scheme recognition model, the training unit is specifically configured to: if the degree of fit is greater than the first threshold, complete the initial Model training, the pre-trained combination scheme recognition model is obtained, otherwise, the reverse training is performed on the initial model based on the loss function in the initial model, and the weight gradient in the initial model is updated until the input When the degree of fit between the predicted impact result set obtained by the training data set and the verification set is greater than the first threshold or the number of times of performing reverse training is greater than the second threshold, the training of the initial model is completed, and the The pre-trained combination scheme recognition model is described.
18. The electronic device according to claim 17, characterized in that

    In order to sequentially input the plurality of combination schemes into a pre-trained combination scheme recognition model and obtain multiple output results, the determination unit is specifically configured to: use medical data corresponding to the plurality of combination schemes as input data It is sequentially input into the pre-trained combination scheme recognition model to obtain a prediction result corresponding to each combination scheme, and the prediction result is a prediction result of the increase in medical expenses of the hospital.
19. An electronic device characterized by comprising a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor , The program includes instructions for performing the steps in any of the methods of claims 1-9.
20. A computer-readable storage medium, characterized in that it is used to store a computer program, wherein the computer program causes a computer to execute the method according to any one of claims 1-9.

Images