WO2021189983A1 - Medical case search method and apparatus based on interactive feedback, and readable storage medium - Google Patents

Abstract

A medical case search method and apparatus based on interactive feedback, and a computer-readable storage medium. The method comprises: according to a preset matching rule, searching an existing medical case database for first-level similar medical cases that are similar to an initial medical case (S1); sequencing the similar medical cases according to the degree of similarity between the similar medical cases and the initial medical case, and feeding a preset number of similar medical cases back to a user according to the sequence (S2); according to feedback information of the user on the similar medical cases, performing supervised learning on the similar medical cases by means of a learning model (S3); and according to a supervised learning result of the learning model, determining second-level similar medical cases, which have a higher degree of similarity to the initial medical case, and feeding same back to the user until a final similar medical case, the degree of similarity of which to the initial medical case meets a preset requirement, is determined (S4). By means of the method, a medical case that best matches a medical condition is quickly and effectively found from a massive amount of medical cases.

Description

Case search method, device and readable storage medium based on interactive feedback

This application requires the priority of the patent application whose application number is 202011118337.4, the filing date is October 19, 2020, and the invention-creation title is "Case search method, device and readable storage medium based on interactive feedback".

Technical field

This application relates to the field of artificial intelligence technology, and in particular to a case search method, device, electronic device, and computer-readable storage medium based on interactive feedback.

Background technique

At present, with the development of computer technology, retrieval has become a common method for obtaining information in daily life. In the medical field, similar case retrieval is of great significance in scientific research and clinical practice. For example, some diseases are special due to various reasons, or the symptoms are complex, or the clinical manifestations are similar to other diseases, or the accompanying multiple complications mainly cover up the root cause of the disease. The diagnosis and treatment of these diseases are often complicated and cumbersome, which results in patients not being diagnosed in time and receiving corresponding treatments, and eventually leading to miss the optimal treatment window. When a patient sees a doctor, if the doctor quickly finds a case similar to the patient, and can make an effective judgment through the diagnosis and treatment path and effect of the similar case in time.

At present, in the process of diagnosis and treatment of these diseases, it is an efficient way to use cured precedents as a reference basis. Doctors search for the most suitable cases for the patient to assist in diagnosis and treatment. However, the current search of similar cases still has some shortcomings:

1) For more complicated conditions: How to accurately search for such cases mainly depends on the experience of doctors, but experience is limited after all, and the search efficiency and results of different doctors will be different;

2) The number of cases is large, and the search speed is slow: the most matching case cannot be effectively searched out from a large number of cases;

3) Existing cases such as searches are all one-way responses, and there is no search method for users to feedback the search results.

The inventor is aware of the above-mentioned problems existing in the traditional case retrieval method, and the present application urgently needs to provide a case search method based on interactive feedback.

Summary of the invention

This application provides a case search method, device, electronic device, and computer-readable storage medium based on interactive feedback, the main purpose of which is to quickly and effectively search for a case that best matches the condition from a large number of cases.

In order to achieve the above objectives, the interactive feedback-based case search method provided by this application includes:

According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.

In order to solve the above problems, the present application also provides a case search device based on interactive feedback, the device includes:

The case search module is used to search for first-level similar cases similar to the initial case from the existing case database according to preset matching rules;

The case ranking module is used for ranking the similar cases according to their similarity to the initial case, and feeding back a preset number of similar cases to the user according to the ranking;

The case learning module is used to perform supervised learning on the similar cases through a learning model according to the user's feedback information on the similar cases;

The case feedback module is used to determine and feed back to the user a second-level similar case that is more similar to the initial case according to the supervised learning result of the learning model, until it is determined that the similarity to the initial case reaches the expected level. Set the required final similar cases.

In order to solve the above-mentioned problems, the present application also provides an electronic device, which includes:

At least one processor; and,

A memory communicatively connected with the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can perform the following steps:

According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.

In order to solve the above-mentioned problems, the present application also provides a computer-readable storage medium having at least one instruction stored in the computer-readable storage medium, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned Case search method based on interactive feedback.

According to the preset matching rules, the embodiment of this application searches the existing case database for first-level similar cases that are similar to the initial case; sorts the similar cases according to their similarity to the initial case, and sorts them according to the degree of similarity to the initial case. Sorting feeds back a preset number of similar cases to the user; according to the user’s feedback information on the similar cases, supervised learning of the similar cases through the learning model; determining and feeding back to the user according to the supervised learning results of the learning model A second-level similar case with a higher degree of similarity to the initial case is determined until a final similar case whose degree of similarity to the initial case reaches the preset requirement is determined. In the embodiments of the present application, the search results are continuously updated through feedback on the search results. Among them, the classification of cases is divided by clustering, and the search results are supervised learning through the learning model, so as to quickly and effectively learn from a large number of cases. The case that best matches the condition of the disease is searched; the search method of this application makes different responses through direct or indirect feedback from the user, continuously updates the search content, realizes the interaction between the engine and the user, and improves the search efficiency of the user.

Description of the drawings

FIG. 1 is a schematic flowchart of a case search method based on interactive feedback provided by an embodiment of this application;

2 is a schematic diagram of modules of a case search device based on interactive feedback provided by an embodiment of this application;

3 is a schematic diagram of the internal structure of an electronic device for implementing a case search method based on interactive feedback provided by an embodiment of the application;

The realization, functional characteristics, and advantages of the purpose of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings.

Detailed ways

It should be understood that the specific embodiments described here are only used to explain the present application, and are not used to limit the present application.

This application provides a case search method based on interactive feedback. Referring to FIG. 1, it is a schematic flowchart of a case search and identification method based on interactive feedback provided by an embodiment of this application. The method can be executed by a device, and the device can be implemented by software and/or hardware.

In this embodiment, the case search method based on interactive feedback includes:

S1: According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

S2: Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

S3: Perform supervised learning on the similar case through a learning model according to the user's feedback information on the similar case;

S4: According to the supervised learning result of the learning model, determine and feed back to the user secondary similar cases that are more similar to the initial case, until it is determined that the similarity to the initial case reaches the preset required final similarity Case.

The above-mentioned case search method based on interactive feedback based on the artificial intelligence of this application can facilitate doctors or clinical researchers (hereinafter referred to as users) to obtain cases that need to be known and referred to. The search method continuously updates the search results from the user's feedback on the search results, and searches for a case that best matches the condition from a large number of cases.

In the case database, the cases are stored in text form. The cases include but are not limited to pathological data such as patient information, clinical manifestations, and final diagnosis results.

In step S1, searching for a first-level similar case similar to the initial case from the existing case database according to the preset matching rules, including the following steps:

Setting the preset matching rules according to the clinical manifestations and diagnosis results of the patients included in the initial case;

According to the preset matching rule, the cases in the case database are matched with the initial case one by one, so as to determine a first-level similar case that matches the initial case.

When the user initiates a search, that is, after entering the initial case, the system parses the initial case to generate a search command, and searches among the massive cases in the case database.

In step S2, the similar cases are sorted according to their similarity to the initial case, and a preset number of similar cases are fed back to the user according to the sort. The searched similar cases are fed back to the user according to the degree of similarity to the initial case. The search engine executes the search command and returns a batch of case texts in descending order of similarity.

When calculating the degree of similarity, it can be defined by calculating the L ₂ norm with other case texts after encoding.

In order to improve the search efficiency, before the user performs the search, the system uses the k-means algorithm to cluster the case files in advance on the case text in the case database. The operation is as follows:

S21: Code the case text in the database after serialization to obtain {x _i };

S22: Initialize the category center vector {μ ₁ ,μ ₂ ,…,μ _k }, k is selected according to the number of case texts when conditions allow (such as system memory, the maximum number of iterations), such as the database contains 1 million cases Text, then k is 1,000;

S23: Divide each text code according to the central vector calculation L ₂ norm, that is, d _ij =‖x _j -μ _i ‖ ₂ ;

S24: Calculate the new category center vector {μ′ ₁ ,μ′ ₂ ,…,μ′ _k };

S25: Iterate steps S23 and S24 until the difference between the corresponding center values of the two center vectors is less than the threshold or exceeds the maximum number of iterations;

S26: Calculate the center vector of the initial case and each category of cases to calculate the L ₂ norm distance, and feed back all the cases in the category with the smallest distance to the user according to the degree of similarity.

After the user enters the search text, he only needs to calculate the _{distance to the center L 2} norm of each category text respectively. For example, the distance between the search text x and the i-th category center μ _i is:

d _i =‖x-μ _i ‖ ₂

Among them, d _{ij represents the L 2} norm distance between the searched case and the center of each category of cases;

x _j represents the search case, μ _i represents the i-th classification center vector;

Find the category with the smallest distance, and then return all cases in this category to the case text according to the degree of similarity.

In step S3, according to the user's feedback information on the similar cases, the supervised learning of the similar cases through the learning model includes the following steps:

The user’s feedback information on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity, among which,

The learning model is used to supervise and learn similar cases with a high degree of similarity to the initial case, where the logistic regression formula is used to calculate the probability:

Among them, x is the feature of the participant's data, and y is the label of the data.

According to the search results, the loss value of the function can be calculated, and the convergence of the model is achieved on the basis of the lowest loss value area. That is, according to the supervised learning result of the learning model, the case that is more similar to the initial case is fed back to the user, until the case that most matches the initial case is fed back to the user.

In step S4, the user's feedback information on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity,

Respectively extract the word frequency of similar cases with a high degree of similarity to the initial case and the word frequencies of similar cases with a low degree of similarity;

Modify the preset matching rule according to the extracted word frequency;

According to the modified preset matching rules and the supervised learning results of the learning model, further search the case database for secondary cases similar to the initial case, and give feedback to the user

In the embodiment of the present application, similar cases with a high degree of similarity and similar cases with a low degree of similarity are respectively strengthened and weakened. The user consults the similar cases fed back by the system, and returns the search after consulting the text of a certain case. engine. The search engine gets feedback, and if the user thinks that the case text description they have just looked up is close, then strengthen the operation. On the contrary, if the user thinks that the case text just checked is not useful, then the weakening operation is performed.

Strengthen and weaken operations, that is, the system records the case texts approved and disapproved by the user each time, and updates the two sets with labels 1 and 0 respectively.

Among them, strengthening and weakening operations also include:

1. Extraction of valid words in text:

The intensive operation extracts all words in the document in the tag 1 set and calculates the number of occurrences, and divides the number of occurrences by the number of occurrences of the word in all case texts in the database to obtain the word frequency. Add the first three words in word frequency to the "text contains" item in the search command.

In the same way, the weakening operation is to extract the word frequency of the documents in the tag 0 set, and add the first three words of the word frequency to the "text does not contain" item in the background search command.

2. Correct the search:

After the search command is updated, the distance to the center L_2 norm of the text of each category will be recalculated. If the distance is found to be less than the category of the current search, it will switch to the new category or merge the category to search. This correction is avoided to a certain extent The search results converge to categories that are not required by the user.

For the above-mentioned user feedback, search engines can also obtain the following ways:

① Directly ask whether pathology is needed;

②According to several inquiries, learn the user's habits to determine whether the pathology is needed.

According to the preset matching rules, the embodiment of this application searches the existing case database for first-level similar cases that are similar to the initial case; sorts the similar cases according to their similarity to the initial case, and sorts them according to the degree of similarity to the initial case. Sorting feeds back a preset number of similar cases to the user; according to the user’s feedback information on the similar cases, supervised learning of the similar cases through the learning model; determining and feeding back to the user according to the supervised learning results of the learning model A second-level similar case with a higher degree of similarity to the initial case is determined until a final similar case whose degree of similarity to the initial case reaches the preset requirement is determined. In the embodiments of the present application, the search results are continuously updated through feedback on the search results. Among them, the classification of cases is divided by clustering, and the search results are supervised learning through the learning model, so as to quickly and effectively learn from a large number of cases. The case that best matches the condition of the disease is searched; the search method of this application makes different responses through direct or indirect feedback from the user, continuously updates the search content, realizes the interaction between the engine and the user, and improves the search efficiency of the user.

As shown in Figure 2, it is a functional block diagram of the case search device based on interactive feedback in this application.

The apparatus 100 for case search based on interactive feedback described in this application can be installed in an electronic device. According to the realized functions, the device for searching a case based on interactive feedback may include: a case searching module 101, a case ranking module 102, a case learning module 103, and a case feedback module 104. The module described in this application can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

In this embodiment, the functions of each module/unit are as follows:

The case search module 101 is used to search for first-level similar cases similar to the initial case from the existing case database according to preset matching rules;

The case ranking module 102 is used for ranking the similar cases according to their similarity to the initial case, and feeding back a preset number of similar cases to the user according to the ranking;

The case learning module 103 is configured to perform supervised learning on the similar cases through a learning model according to user feedback information on the similar cases;

The case feedback module 104 is used to determine and feed back to the user a second-level similar case that is more similar to the initial case according to the supervised learning result of the learning model, until it is determined that the similarity to the initial case is reached Final similar cases with pre-determined requirements

Wherein, searching for a first-level similar case similar to the initial case from an existing case database according to a preset matching rule includes the following steps:

Setting the preset matching rules according to the clinical manifestations and diagnosis results of the patients included in the initial case;

According to the preset matching rule, the cases in the case database are matched with the initial case one by one, so as to determine a first-level similar case that matches the initial case.

Optionally, the similar cases are sorted according to their similarity to the initial case, and a preset number of similar cases are fed back to the user according to the sorting, and the k-means algorithm is used to compare the cases in the case database. The cluster division includes the following steps:

The first step: serialize the cases in the case database and encode them to obtain {x _i };

Step 2: Initialize the category center vector {μ ₁ ,μ ₂ ,...,μ _k }, where k is selected at the maximum number of iterations according to the number of cases;

The third step: the code of each case is divided into categories by _{calculating the L 2 norm according to the central vector;}

Step 4: According to the division result, calculate the category center vector {μ′ ₁ ,μ′ ₂ ,…,μ′ _k };

Step 5: Iterate the third and fourth steps until the difference between the corresponding center values of the center vectors of the two categories is less than the threshold or exceeds the maximum number of iterations;

Step 6: Calculate the center vector of the initial case and each category of cases to calculate the L ₂ norm distance, and feed back all the cases in the category with the smallest distance to the user according to the degree of similarity.

Wherein, the coding of each case is _{divided according to the central vector calculation L 2} norm, and the formula used is:

d _ij =‖x _j -μ _i ‖ ₂

Among them, d _{ij represents the L 2} norm distance between the searched case and the center of each category of cases;

x _j represents the search case, and μ _i represents the i-th classification center vector.

Wherein, according to the user's feedback information on the similar cases, the supervised learning of the similar cases through a learning model includes the following steps:

The user’s feedback information on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity, among which,

The learning model is used to supervise and learn similar cases with a high degree of similarity to the initial case, where the logistic regression formula is used to calculate the probability:

Among them, x is the feature of the participant's data, and y is the label of the data.

Wherein, the feedback information of the user on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity:

Respectively extract the word frequency of similar cases with a high degree of similarity to the initial case and the word frequencies of similar cases with a low degree of similarity;

Modify the preset matching rule according to the extracted word frequency;

According to the modified preset matching rules and the supervised learning results of the learning model, the case database is further searched for secondary cases similar to the initial case, and the user is fed back according to the preset Matching rules, searching for first-level similar cases similar to the initial case from the existing case database; sorting the similar cases according to their similarity to the initial case, and sorting the preset number of similar cases according to the sorting Feedback to the user; according to the user's feedback information on the similar case, supervise learning of the similar case through a learning model; determine and feed back to the user the degree of similarity to the initial case according to the supervised learning result of the learning model Higher second-level similar cases until the final similar cases whose similarity with the initial case meets the preset requirements are determined. In the embodiments of the present application, the search results are continuously updated through feedback on the search results. Among them, the classification of cases is divided by clustering, and the search results are supervised learning through the learning model, so as to quickly and effectively learn from a large number of cases. The case that best matches the condition of the disease is searched; the search method of this application makes different responses through direct or indirect feedback from the user, continuously updates the search content, realizes the interaction between the engine and the user, and improves the search efficiency of the user.

As shown in FIG. 3, it is a schematic diagram of the structure of an electronic device implementing a case search method based on interactive feedback in this application.

The electronic device 1 may include a processor 10, a memory 11, and a bus, and may also include a computer program stored in the memory 11 and running on the processor 10, such as a case search program 12 based on interactive feedback.

Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, magnetic disk, CD etc. The memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, for example, a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart media card (SMC), and a secure digital (Secure Digital) equipped on the electronic device 1. , SD) card, flash card (Flash Card), etc. Further, the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can be used not only to store application software and various data installed in the electronic device 1, such as the code of a data audit program, etc., but also to temporarily store data that has been output or will be output.

The processor 10 may be composed of integrated circuits in some embodiments, for example, may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits with the same function or different functions, including one or more Combinations of central processing unit (CPU), microprocessor, digital processing chip, graphics processor, and various control chips, etc. The processor 10 is the control unit of the electronic device, which uses various interfaces and lines to connect the various components of the entire electronic device, and runs or executes programs or modules (such as data) stored in the memory 11 Audit procedures, etc.), and call data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

The bus may be a peripheral component interconnect standard (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus and so on. The bus is configured to implement connection and communication between the memory 11 and at least one processor 10 and the like.

FIG. 3 only shows an electronic device with components. Those skilled in the art can understand that the structure shown in FIG. 2 does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure. Components, or a combination of certain components, or different component arrangements.

For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, thereby controlling power The device implements functions such as charge management, discharge management, and power consumption management. The power supply may also include any components such as one or more DC or AC power supplies, recharging devices, power failure detection circuits, power converters or inverters, and power status indicators. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

Further, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.

Optionally, the electronic device 1 may also include a user interface. The user interface may be a display (Display) and an input unit (such as a keyboard (Keyboard)). Optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the electronic device 1 and to display a visualized user interface.

It should be understood that the embodiments are only for illustrative purposes, and are not limited by this structure in the scope of the patent application.

The case search program 12 based on interactive feedback stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can realize:

According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.

Specifically, for the specific implementation method of the above-mentioned instructions by the processor 10, reference may be made to the description of the relevant steps in the embodiment corresponding to FIG. 1, which will not be repeated here.

Further, if the integrated module/unit of the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory) .

In the embodiment of the present application, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a case search method based on interactive feedback is implemented. The specific method is as follows:

According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.

The computer-readable storage medium may be non-volatile or volatile.

In the several embodiments provided in this application, it should be understood that the disclosed equipment, device, and method may be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional modules in the various embodiments 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-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional modules.

For those skilled in the art, it is obvious that the present application is not limited to the details of the foregoing exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or basic characteristics of the application.

Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of this application is defined by the appended claims rather than the above description, and therefore it is intended to fall into the claims. All changes in the meaning and scope of the equivalent elements of are included in this application. Any associated diagram marks in the claims should not be regarded as limiting the claims involved.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the application and not to limit them. Although the application has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the application can be Make modifications or equivalent replacements without departing from the spirit and scope of the technical solution of the present application.

Claims (20)

1. A case search method based on interactive feedback, wherein the method includes:

   According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

   Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

   According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

   According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.
2. The case search method based on interactive feedback according to claim 1, wherein the search for a first-level similar case similar to the initial case from an existing case database according to a preset matching rule comprises the following steps:

   Setting the preset matching rules according to the clinical manifestations and diagnosis results of the patients included in the initial case;

   According to the preset matching rule, the cases in the case database are matched with the initial case one by one, so as to determine a first-level similar case that matches the initial case.
3. The case search method based on interactive feedback according to claim 1, wherein the said similar cases are sorted according to their similarity to the initial case, and a preset number of similar cases are fed back to the user according to the sorting , Using the k-means algorithm to cluster and divide the cases in the case database, including the following steps:

   The first step: serialize the cases in the case database and encode them to obtain {x _i };

   Step 2: Initialize the category center vector {μ ₁ ,μ ₂ ,...,μ _k }, where k is selected at the maximum number of iterations according to the number of cases;

   The third step: the code of each case is divided into categories by _{calculating the L 2 norm according to the central vector;}

   Step 4: According to the division result, calculate the category center vector {μ′ ₁ ,μ′ ₂ ,…,μ′ _k };

   Step 5: Iterate the third and fourth steps until the difference between the corresponding center values of the center vectors of the two categories is less than the threshold or exceeds the maximum number of iterations;

   Step 6: Calculate the center vector of the initial case and each category of cases to calculate the L ₂ norm distance, and feed back all the cases in the category with the smallest distance to the user according to the degree of similarity.
4. The case search method based on interactive feedback according to claim 3, wherein the code of each case is _{divided into categories by calculating the L 2} norm according to the center vector, and the formula used is:

   d _ij =‖x _j -μ _i ‖ ₂

   Among them, d _{ij represents the L 2} norm distance between the searched case and the center of each category of cases;

   x _j represents the search case, and μ _i represents the i-th classification center vector.
5. The case search method based on interactive feedback according to claim 1, wherein the supervised learning of the similar cases through a learning model according to the user's feedback information on the similar cases comprises the following steps:

   The user’s feedback information on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity, among which,

   The learning model is used to supervise and learn similar cases with a high degree of similarity to the initial case, where the logistic regression formula is used to calculate the probability:

   

   Among them, x is the feature of the participant's data, and y is the label of the data.
6. The case search method based on interactive feedback according to claim 5, wherein the feedback information points of the user on the similar cases include: similar cases with a high degree of similarity and similar cases with a low degree of similarity:

   Respectively extract the word frequency of similar cases with a high degree of similarity to the initial case and the word frequencies of similar cases with a low degree of similarity;

   Modify the preset matching rule according to the extracted word frequency;

   According to the modified preset matching rules and the supervised learning result of the learning model, the second-level case similar to the initial case in the case database is further searched, and the user is fed back.
7. The case search method based on interactive feedback according to claim 1, wherein:

   The user performs an enhanced operation on the similar cases with a high degree of similarity, and the user performs a weakening operation on the similar cases with a low degree of similarity; wherein,

   The strengthening operation corresponds to the label 1 set, and the weakening operation corresponds to the label 0 set.
8. The case search method based on interactive feedback according to claim 7, wherein:

   The strengthening operation refers to: extracting all words in the document in the label 1 set, calculating the number of occurrences, and dividing the number of occurrences by the number of occurrences of the word in all case texts in the case database to obtain Word frequency, and add the top three words of the word frequency to the "text contains" item in the search command.
9. The case search method based on interactive feedback according to claim 7, wherein:

   The weakening operation refers to: extracting all words in the document in the tag 0 set, calculating the number of occurrences, dividing the number of occurrences by the number of occurrences of the words in all case texts in the case database to obtain the word frequency, and Add the top three words of the word frequency to the "text does not contain" item in the background search command.
10. A case search device based on interactive feedback, wherein the device includes:

    The case search module is used to search for first-level similar cases similar to the initial case from the existing case database according to preset matching rules;

    The case ranking module is used for ranking the similar cases according to their similarity to the initial case, and feeding back a preset number of similar cases to the user according to the ranking;

    The case learning module is used to perform supervised learning on the similar cases through a learning model according to the user's feedback information on the similar cases;

    The case feedback module is used to determine and feed back to the user a second-level similar case that is more similar to the initial case according to the supervised learning result of the learning model, until it is determined that the similarity to the initial case reaches the expected level. Set the required final similar cases.
11. An electronic device, wherein the electronic device includes:

    At least one processor; and,

    A memory communicatively connected with the at least one processor; wherein,

    The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can execute the following steps:

    According to preset matching rules, search for first-level similar cases similar to the initial case from the existing case database;

    Sort the similar cases according to their similarity to the initial case, and feed back a preset number of similar cases to the user according to the sort;

    According to the user's feedback information on the similar cases, supervise learning of the similar cases through a learning model;

    According to the supervised learning result of the learning model, determine and feed back to the user a second-level similar case that is more similar to the initial case, until a final similar case that is similar to the initial case reaches a preset requirement is determined.
12. The electronic device according to claim 11, wherein:

    According to preset matching rules, searching for a first-level similar case similar to the initial case from an existing case database includes the following steps:

    Setting the preset matching rules according to the clinical manifestations and diagnosis results of the patients included in the initial case;

    According to the preset matching rule, the cases in the case database are matched with the initial case one by one, so as to determine a first-level similar case that matches the initial case.
13. The electronic device according to claim 11, wherein:

    The said similar cases are sorted according to their similarity to the initial case, and a preset number of similar cases are fed back to the user according to the sorting, and the cases in the case database are clustered and divided by the k-means algorithm , Including the following steps:

    The first step: serialize the cases in the case database and encode them to obtain {x _i };

    Step 2: Initialize the category center vector {μ ₁ ,μ ₂ ,...,μ _k }, where k is selected at the maximum number of iterations according to the number of cases;

    The third step: the code of each case is divided into categories by _{calculating the L 2 norm according to the central vector;}

    Step 4: According to the division result, calculate the category center vector {μ′ ₁ ,μ′ ₂ ,…,μ′ _k };

    Step 5: Iterate the third and fourth steps until the difference between the corresponding center values of the center vectors of the two categories is less than the threshold or exceeds the maximum number of iterations;

    Step 6: Calculate the center vector of the initial case and each category of cases to calculate the L ₂ norm distance, and feed back all the cases in the category with the smallest distance to the user according to the degree of similarity.
14. The electronic device according to claim 13, wherein:

    The coding of each case is _{divided into categories by calculating the L 2} norm according to the center vector, and the formula used is:

    d _ij =‖x _j -μ _i ‖ ₂

    Among them, d _{ij represents the L 2} norm distance between the searched case and the center of each category of cases;

    x _j represents the search case, and μ _i represents the i-th classification center vector.
15. The electronic device according to claim 11, wherein:

    The supervised learning of the similar cases through a learning model according to the user's feedback information on the similar cases includes the following steps:

    The user’s feedback information on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity, among which,

    The learning model is used to supervise and learn similar cases with a high degree of similarity to the initial case, where the logistic regression formula is used to calculate the probability:

    

    Among them, x is the feature of the participant's data, and y is the label of the data.
16. The electronic device according to claim 15, wherein the feedback information of the user on the similar cases includes: similar cases with a high degree of similarity and similar cases with a low degree of similarity:

    Respectively extract the word frequency of similar cases with a high degree of similarity to the initial case and the word frequencies of similar cases with a low degree of similarity;

    Modify the preset matching rule according to the extracted word frequency;

    According to the modified preset matching rules and the supervised learning result of the learning model, the second-level case similar to the initial case in the case database is further searched, and the user is fed back.
17. The electronic device according to claim 11, wherein:

    The user performs an enhanced operation on the similar cases with a high degree of similarity, and the user performs a weakening operation on the similar cases with a low degree of similarity; wherein,

    The strengthening operation corresponds to the label 1 set, and the weakening operation corresponds to the label 0 set.
18. The electronic device according to claim 17, wherein:

    The strengthening operation refers to: extracting all words and their appearance times in the documents in the label 1 set, dividing by the times of appearance of the words in all case texts in the case database, and obtaining the word frequency, and adding the word frequency before the word frequency. The three words are added to the "text contains" item in the search command.
19. The electronic device according to claim 17, wherein:

    The weakening operation is to extract all words and their appearance times in the document in the tag 0 set, divide by the times the words appear in all case texts in the case database to obtain the word frequency, and add the first three to the background search command The "text does not contain" item.
20. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for searching a case based on interactive feedback according to any one of claims 1 to 9 is realized.

Images