A device and method for determining the severity of a rare disease

Abstract

The present invention relates to the field of Internet technology, and in particular to a device and method for determining the severity of a rare disease. The device includes: a processor, a storage medium and a bus. The processor executes machine-readable instructions to determine the severity of a target rare disease type for a target user based on the similarity between detection data corresponding to the target rare disease type of the target user and rare disease feature detection data corresponding to the target rare disease type at different severities, thereby providing an effective treatment direction for a doctor, thereby avoiding unnecessary examinations and unreasonable treatments for the target user and shortening the treatment time of the target user.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to a device and method for determining the severity of a rare disease. Background Art

[0002] For rare diseases, clinicians see fewer patients and have limited experience. Inexperienced clinicians cannot accurately determine the current severity of the patient's illness, which leads to failure to choose the best treatment plan. Unclear causes may lead to unnecessary examinations and unreasonable treatments, referrals to multiple departments, and increased treatment time. Summary of the invention

[0003] In view of this, the purpose of this application is to provide a device and method for determining the severity of rare diseases, which can help doctors determine the type and severity of at least one target rare disease that a target user may have, and provide doctors with effective treatment directions, thereby avoiding unnecessary examinations and unreasonable treatments for the target user and shortening the treatment time for the target user.

[0004] In a first aspect, an embodiment of the present application provides a device for determining the severity of a rare disease, the device comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, when the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the following steps:

[0005] Obtain rare disease feature detection data corresponding to each rare disease type at different severity levels, and first detection data of a target user determined according to a target detection item subset; the target detection item subset is a subset selected from the entire set of rare disease detection items that meets preset conditions and has the highest rare disease detection accuracy; the entire set of rare disease detection items includes detection item identifiers corresponding to various rare disease types; the preset conditions include quantity constraints and cost constraints for the target detection item subset; the rare disease feature detection data corresponding to each severity level is determined based on multiple rare disease detection data sets corresponding to the rare disease type; the rare disease detection data set is obtained by clustering all rare disease detection data corresponding to the rare disease type;

[0006] Inputting the first test data into a rare disease detection model to obtain at least one initial rare disease type; the rare disease detection model is obtained by training the test sample data and the corresponding rare disease type label;

[0007] Determining a target rare disease type from all initial rare disease types based on second test data corresponding to the target user obtained from the test item identifier corresponding to the initial rare disease type;

[0008] The similarity between the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type and the rare disease feature test data corresponding to the target rare disease type at different severity levels is calculated using the following similarity formula;

[0009] ;

[0010] in, is the similarity between the detection data corresponding to the target user and the detection data of rare disease features, The number of detection item identifiers corresponding to the detection data corresponding to the target user, The number of detection times corresponding to the detection data of the target user, = the detection data value of the nth detection item at the ath detection time in the detection data corresponding to the target user minus the detection data value of the nth detection item at the ath detection time The value obtained after the detection data value at the detection time, = The value of the test data at the ath test time of the nth test item in the rare disease feature test data minus the value of the The difference between the detection data values ​​at the detection time is obtained. is the absolute value of the difference between the detection data value of the nth detection item at the i-th moment in the detection data corresponding to the target user and the detection data value of the nth detection item at the i-th moment in the rare disease feature detection data, is the first preset weight, is the second preset weight;

[0011] The severity with the highest similarity is determined as the severity of the target user for the target rare disease type.

[0012] In a possible implementation, the processor selects a target detection item subset from the full set of rare disease detection items through the following steps:

[0013] Obtaining at least one initial test item subset that meets preset conditions and is selected from the entire set of rare disease test items, and rare disease test item sample data corresponding to each initial test item subset; the rare disease test item sample data carries a rare disease type label;

[0014] Inputting the rare disease test item sample data corresponding to each initial test item subset into the rare disease detection model to obtain the rare disease type detection results of the rare disease test item sample data;

[0015] Determine the rare disease detection accuracy corresponding to each initial detection item subset based on the rare disease type detection results and rare disease type labels of the rare disease detection item sample data corresponding to each initial detection item subset;

[0016] The initial test item subset with the highest rare disease detection accuracy is determined as the target test item subset.

[0017] In a possible implementation, the processor obtains rare disease feature detection data corresponding to each rare disease type at different severity levels through the following steps:

[0018] Based on the detection item identifier corresponding to the rare disease type, obtain the corresponding rare disease detection data at different severity levels;

[0019] The rare disease detection data are clustered by a target clustering algorithm to obtain multiple rare disease detection data sets; the target clustering algorithm is a clustering algorithm with the most accurate clustering result selected from multiple initial clustering algorithms;

[0020] Count the number of rare disease detection data corresponding to each severity level in each rare disease detection data set;

[0021] The severity level with the largest number of rare disease detection data is determined as the target severity level corresponding to each rare disease detection data set;

[0022] According to all target rare disease detection data with severity corresponding to the target severity in each rare disease detection data set, the rare disease feature detection data corresponding to each target severity is determined.

[0023] In a possible implementation, the processor selects a target clustering algorithm from a plurality of initial clustering algorithms through the following steps:

[0024] Obtain test data for multiple rare disease samples;

[0025] Clustering all rare disease sample detection data using each initial clustering algorithm to obtain multiple rare disease sample detection data sets corresponding to each initial clustering algorithm;

[0026] For each rare disease sample detection data set corresponding to each initial clustering algorithm, determine a first data similarity of the rare disease sample detection data set corresponding to the initial clustering algorithm according to the data similarity between every two rare disease sample detection data in the rare disease sample detection data set;

[0027] Determine a second data similarity corresponding to the initial clustering algorithm according to the first data similarity of all rare disease sample detection data sets corresponding to the initial clustering algorithm;

[0028] The initial clustering algorithm with the highest second data similarity is determined as the target clustering algorithm.

[0029] In a possible implementation, the processor calculates the data similarity between every two rare disease sample detection data using the following formula:

[0030] ;

[0031] in, Detect the data similarity between two rare disease samples, The number of identical test item identifiers corresponding to the two rare disease sample test data. is the total number of test item identifiers after deduplication corresponding to the detection data of two rare disease samples, is the absolute value of the difference between the test data value of the i-th test item in the first rare disease sample test data and the test data value of the i-th test item in the second rare disease sample test data, and The preset weights.

[0032] In a possible implementation, the processor determines the target rare disease type from all initial rare disease types based on the second test data corresponding to the target user acquired based on the test item identifier corresponding to the initial rare disease type through the following steps:

[0033] Determine the similarity between the second test data and the rare disease comprehensive feature test data corresponding to each initial rare disease type as the first test probability corresponding to each initial rare disease type; the rare disease comprehensive feature test data corresponding to each initial rare disease type is obtained by combining the test data corresponding to each initial rare disease type at all severity levels;

[0034] Inputting the second detection data into the rare disease detection model to obtain a second detection probability corresponding to each initial rare disease type;

[0035] Calculate the target detection probability corresponding to each initial rare disease type according to the first detection probability and the second detection probability corresponding to each initial rare disease type;

[0036] The initial rare disease type with the highest target detection probability is determined as the target rare disease type.

[0037] In a possible implementation manner, the processor is further configured to perform the following steps:

[0038] A knowledge graph of rare disease treatment information is obtained; the knowledge graph of rare disease treatment information includes multiple triples; the first entity in each triple is text information including the type of rare disease, the severity of the rare disease, and test data, and the second entity in each triple refers to treatment text information;

[0039] According to the target rare disease type, the severity of the target rare disease type for the target user, and the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type, the treatment text information of the target user for the target rare disease type is queried from the rare disease treatment information knowledge graph.

[0040] In a second aspect, the embodiments of the present application further provide a method for determining the severity of a rare disease, the method for determining the severity of a rare disease comprising:

[0041] Obtain rare disease feature detection data corresponding to each rare disease type at different severity levels, and first detection data of a target user determined according to a target detection item subset; the target detection item subset is a subset selected from the entire set of rare disease detection items that meets preset conditions and has the highest rare disease detection accuracy; the entire set of rare disease detection items includes detection item identifiers corresponding to various rare disease types; the preset conditions include quantity constraints and cost constraints for the target detection item subset; the rare disease feature detection data corresponding to each severity level is determined based on multiple rare disease detection data sets corresponding to the rare disease type; the rare disease detection data set is obtained by clustering all rare disease detection data corresponding to the rare disease type;

[0042] Inputting the first test data into a rare disease detection model to obtain at least one initial rare disease type; the rare disease detection model is obtained by training the test sample data and the corresponding rare disease type label;

[0043] Determining a target rare disease type from all initial rare disease types based on second test data corresponding to the target user obtained from the test item identifier corresponding to the initial rare disease type;

[0044] The similarity between the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type and the rare disease feature test data corresponding to the target rare disease type at different severity levels is calculated using the following similarity formula;

[0045] ;

[0046] in, is the similarity between the detection data corresponding to the target user and the detection data of rare disease features, The number of detection item identifiers corresponding to the detection data corresponding to the target user, The number of detection times corresponding to the detection data of the target user, = the detection data value of the nth detection item at the ath detection time in the detection data corresponding to the target user minus the detection data value of the nth detection item at the ath detection time The value obtained after the detection data value at the detection time, = The value of the test data at the ath test time of the nth test item in the rare disease feature test data minus the value of the The difference between the detection data values ​​at the detection time is obtained. is the absolute value of the difference between the detection data value of the nth detection item at the i-th moment in the detection data corresponding to the target user and the detection data value of the nth detection item at the i-th moment in the rare disease feature detection data, is the first preset weight, is the second preset weight;

[0047] The severity with the highest similarity is determined as the severity of the target user for the target rare disease type.

[0048] In a possible implementation, a target detection item subset is selected from the entire set of rare disease detection items through the following steps:

[0049] Obtaining at least one initial test item subset that meets preset conditions and is selected from the entire set of rare disease test items, and rare disease test item sample data corresponding to each initial test item subset; the rare disease test item sample data carries a rare disease type label;

[0050] Inputting the rare disease test item sample data corresponding to each initial test item subset into the rare disease detection model to obtain the rare disease type detection results of the rare disease test item sample data;

[0051] Determine the rare disease detection accuracy corresponding to each initial detection item subset based on the rare disease type detection results and rare disease type labels of the rare disease detection item sample data corresponding to each initial detection item subset;

[0052] The initial test item subset with the highest rare disease detection accuracy is determined as the target test item subset.

[0053] In a possible implementation, the following steps are performed to obtain the rare disease feature detection data corresponding to each rare disease type at different severity levels:

[0054] Based on the detection item identifier corresponding to the rare disease type, obtain the corresponding rare disease detection data at different severity levels;

[0055] The rare disease detection data are clustered by a target clustering algorithm to obtain multiple rare disease detection data sets; the target clustering algorithm is a clustering algorithm with the most accurate clustering result selected from multiple initial clustering algorithms;

[0056] Count the number of rare disease detection data corresponding to each severity level in each rare disease detection data set;

[0057] The severity level with the largest number of rare disease detection data is determined as the target severity level corresponding to each rare disease detection data set;

[0058] According to all target rare disease detection data with severity corresponding to the target severity in each rare disease detection data set, the rare disease feature detection data corresponding to each target severity is determined.

[0059] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for determining the severity of a rare disease as described in any one of the second aspects are executed.

[0060] The embodiment of the present application provides a device and method for determining the severity of a rare disease, the device comprising: a processor, a storage medium and a bus, the processor executing machine-readable instructions to perform the following steps: inputting the first detection data of the target user determined according to the target detection item subset into the rare disease detection model to obtain the initial rare disease type; determining the target rare disease type from all initial rare disease types based on the second detection data corresponding to the target user obtained based on the detection item identifier corresponding to the initial rare disease type; calculating the similarity between the detection data corresponding to the target user obtained based on the detection item identifier corresponding to the target rare disease type and the rare disease feature detection data corresponding to the target rare disease type at different severity levels; determining the severity with the highest similarity as the severity of the target user for the target rare disease type. The present application can determine the severity of the target user for the target rare disease type based on the similarity between the detection data of the target user obtained based on the detection item identifier corresponding to the target rare disease type of the target user and the rare disease feature detection data corresponding to the target rare disease type at different severity levels, providing doctors with effective treatment directions, thereby avoiding unnecessary examinations and unreasonable treatments for the target user, and shortening the treatment time of the target user. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 A schematic diagram of the structure of a device for determining the severity of a rare disease provided in an embodiment of the present application is shown;

[0063] Figure 2 A flow chart showing a method for determining the severity of a rare disease provided in an embodiment of the present application;

[0064] Figure 3 A flow chart showing another method for determining the severity of a rare disease provided in an embodiment of the present application;

[0065] Figure 4 A flow chart showing another method for determining the severity of a rare disease provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0067] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0068] In order to enable those skilled in the art to use the content of this application, the following implementation is provided in conjunction with a specific application scenario, the "Internet technology field". For those skilled in the art, the general principles defined herein can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application. Although this application is mainly described around the "Internet technology field", it should be understood that this is only an exemplary embodiment.

[0069] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0070] The following is a detailed description of a device for determining the severity of a rare disease provided in an embodiment of the present application.

[0071] Reference Figure 1As shown, it is a structural schematic diagram of a device for determining the severity of a rare disease provided by an embodiment of the present application. The device for determining the severity of a rare disease includes: a processor 101, a storage medium 102 and a bus 103. The storage medium 102 (which can be a general-purpose memory, not limited thereto) stores machine-readable instructions executable by the processor 101. When the electronic device is running, the processor 101 communicates with the storage medium 102 through the bus. When the processor 101 executes the machine-readable instructions stored in the storage medium 102, the following steps are performed:

[0072] Step 1: Obtain rare disease feature detection data corresponding to each rare disease type at different severity levels, and first detection data of the target user determined based on the target detection item subset.

[0073] In an embodiment of the present application, a target detection item subset is a subset selected from a full set of rare disease detection items that meets preset conditions and has the highest rare disease detection accuracy; the full set of rare disease detection items includes detection item identifiers corresponding to various rare disease types; the preset conditions include quantity constraints and cost constraints (cost constraints include economic cost constraints and / or time consumption constraints) for the target detection item subset; the rare disease feature detection data corresponding to each severity level is determined based on multiple rare disease detection data sets corresponding to the rare disease type; the rare disease detection data set is obtained by clustering all rare disease detection data corresponding to the rare disease type.

[0074] Among them, rare disease types involve rare diseases in many aspects such as the blood system (such as hemophilia, etc.), the nervous system (such as myasthenia gravis, neuromyelitis optica, etc.), and the cardiovascular system (such as idiopathic pulmonary hypertension, Fabry disease, etc.). Each rare disease corresponds to multiple severity levels (such as early, middle, late, etc.). Each rare disease type corresponds to a rare disease characteristic data at each severity level. The rare disease characteristic data includes a test data value corresponding to at least one test item identifier (which can be the name or number of the test item, etc.). The rare disease characteristic data can be expressed as {test data value corresponding to test item 1, test data value corresponding to test item 2, ..., test data value corresponding to test item n}. The target user refers to a user who needs to determine the severity of the rare disease. The first test data includes a test data value corresponding to at least one test item identifier, and the type and number of the test item identifier corresponding to the first test data are the same as the type and number of the test item identifier corresponding to the target test item subset. The test item identifier corresponding to the rare disease type refers to the identifier of all test items (such as blood pressure test, blood sugar test, etc.) required to determine whether the user suffers from the rare disease type.

[0075] For example, if the rare disease types include rare disease 1 and rare disease 2, and the severity levels include early, middle and late stages, then the rare disease feature detection data corresponding to each rare disease type at different severity levels include: rare disease feature detection data a corresponding to rare disease 1 in the early stage, rare disease feature detection data b corresponding to rare disease 1 in the middle stage, rare disease feature detection data c corresponding to rare disease 1 in the late stage, rare disease feature detection data d corresponding to rare disease 2 in the early stage, rare disease feature detection data e corresponding to rare disease 2 in the middle stage, and rare disease feature detection data f corresponding to rare disease 2 in the late stage.

[0076] For example, the target detection item subset includes test item identification 1 and test item identification 2, then the first detection data of the target user determined according to the target detection item subset includes: the detection data value corresponding to the test item identification 1 and the detection data value corresponding to the test item identification 2.

[0077] Here, the target user only needs to test the test items included in the target test item subset, without having to test all the test items in the rare disease test item set. In addition, the target test item subset needs to meet quantity constraints and cost constraints, which can effectively reduce the number of tests and the test cost of the target user.

[0078] Specifically, the processor 101 selects a target detection item subset from the full set of rare disease detection items through the following steps:

[0079] a. Obtain at least one initial test item subset that meets preset conditions and is selected from the entire set of rare disease test items, and rare disease test item sample data corresponding to each initial test item subset;

[0080] In an embodiment of the present application, the subsets of the complete set of rare disease detection items that meet the preset conditions are all initial detection item subsets. The initial detection item subset includes at least one detection item identifier in the complete set of rare disease detection items. The quantity constraint in the preset conditions is used to characterize the maximum number of detection item identifiers contained in the initial detection item subset. The cost constraint in the preset conditions is used to characterize the maximum cost required for the user to complete the detection of all detection items in the initial detection item subset. The types and quantities of detection item identifiers corresponding to the rare disease detection item sample data are the same as the types and quantities of detection item identifiers contained in the initial detection item subset. The rare disease detection item sample data carries a label of at least one rare disease type, i.e., a rare disease type label.

[0081] b. Input the rare disease detection item sample data corresponding to each initial detection item subset into the rare disease detection model to obtain the rare disease type detection results of the rare disease detection item sample data.

[0082] In the embodiment of the present application, the rare disease detection model is a pre-trained deep learning model, the input is the test data containing the test data value corresponding to at least one test item identifier, and the output is the rare disease type of the test data. The rare disease type detection result contains at least one rare disease type.

[0083] c. Determine the rare disease detection accuracy corresponding to each initial detection item subset based on the rare disease type detection results and rare disease type labels of the rare disease detection item sample data corresponding to each initial detection item subset.

[0084] In the embodiment of the present application, for each initial detection item subset, the target number of rare disease detection item sample data whose rare disease type labels are included in the rare disease type detection results is counted from all rare disease detection item sample data corresponding to the initial detection item subset. The ratio of the target number corresponding to the initial detection item subset to the total number of rare disease detection item samples corresponding to the initial detection item subset is determined as the rare disease detection accuracy corresponding to the initial detection item subset.

[0085] For example, if the rare disease type label includes rare disease type 1 and rare disease type 2, and the rare disease type test result includes rare disease type 1, rare disease type 2, and rare disease type 3, since the rare disease type test result includes rare disease type 1 and rare disease type 2 in the rare disease type label, the rare disease type label is included in the rare disease type test result. If the rare disease type label includes rare disease type 1 and rare disease type 3, and the rare disease type test result includes rare disease type 1 and rare disease type 2, since the rare disease type test result does not include rare disease type 3 in the rare disease type label, the rare disease type label is not included in the rare disease type test result.

[0086] Here, since the initial detection item subset is a subset that meets the preset conditions selected from the entire set of rare disease detection items, the initial detection item subset generally only contains some detection item identifiers of a certain rare disease type, but cannot contain all detection item identifiers corresponding to the rare disease type. The results detected by the initial detection item subset are difficult to be completely consistent with the label. Therefore, as long as the rare disease type label of the rare disease detection item sample data is included in the rare disease type detection result, the test result can be considered correct, and the rare disease detection item sample data will be included when counting the target number.

[0087] d. Determine the initial detection item subset with the highest rare disease detection accuracy as the target detection item subset.

[0088] Specifically, the following steps are used to obtain the rare disease feature detection data corresponding to each rare disease type at different severity levels:

[0089] (1) Based on the detection item identifier corresponding to the rare disease type, obtain the corresponding rare disease detection data at different severity levels.

[0090] In the implementation manner of the present application, each type of rare disease corresponds to a plurality of rare disease detection data to varying degrees, and the rare disease detection data can be obtained from the medical record database related to rare diseases in the hospital.

[0091] The rare disease detection data includes at least one detection data value corresponding to a detection item identifier. In addition, the detection item identifiers corresponding to different rare disease detection data may be different.

[0092] (2) Cluster the rare disease detection data through the target clustering algorithm to obtain multiple rare disease detection data sets.

[0093] In the implementation manner of the present application, the target clustering algorithm is a clustering algorithm selected from multiple initial clustering algorithms with the most accurate clustering result; the initial clustering algorithm can be any existing clustering algorithm. The rare disease detection data are clustered by the target clustering algorithm to combine the rare disease detection data belonging to the same cluster into a rare disease detection data set.

[0094] Furthermore, the processor selects a target clustering algorithm from a plurality of initial clustering algorithms through the following steps:

[0095] i. Obtain test data of multiple rare disease samples.

[0096] In the embodiment of the present application, the rare disease sample test data includes a test data value corresponding to at least one test item identifier. The corresponding test item identifiers in different rare disease sample test data may be different.

[0097] ii. Cluster all rare disease sample detection data using each initial clustering algorithm to obtain multiple rare disease sample detection data sets corresponding to each initial clustering algorithm.

[0098] iii. For each rare disease sample detection data set corresponding to each initial clustering algorithm, determine the first data similarity of the rare disease sample detection data set corresponding to the initial clustering algorithm according to the data similarity between every two rare disease sample detection data in the rare disease sample detection data set.

[0099] In the implementation manner of the present application, the higher the first data similarity of the rare disease sample detection data set, the closer the severity levels corresponding to different rare disease sample detection data in the rare disease sample detection data set are, that is, the better the clustering effect of the initial clustering algorithm.

[0100] Here, the data similarity between each two rare disease sample test data is calculated by the following formula:

[0101] ;

[0102] in, Detect the data similarity between two rare disease samples, The number of identical test item identifiers corresponding to the two rare disease sample test data. is the total number of test item identifiers after deduplication corresponding to the detection data of two rare disease samples, is the absolute value of the difference between the test data value of the i-th test item in the first rare disease sample test data and the test data value of the i-th test item in the second rare disease sample test data, and The preset weights.

[0103] Among them, in the first rare disease sample data, the test data value corresponding to the test item identifier included in the second rare disease sample test data but not included in the first rare disease sample test data is set to 0 In the second rare disease sample data, the test data value corresponding to the test item identifier included in the first rare disease sample test data but not included in the second rare disease sample test data is set to 0

[0104] For example, in the two rare disease sample test data, the test data value corresponding to the test item identifier 1 in the first rare disease sample test data is 97, the test data value corresponding to the test item identifier 2 is 3, and the test data value corresponding to the test item identifier 5 is 40. In the two rare disease sample test data, the test data value corresponding to the test item identifier 1 in the second rare disease sample test data is 93, the test data value corresponding to the test item identifier 2 is 5, and the test data value corresponding to the test item identifier 6 is 42. Therefore, the number of identical test item identifiers (test item identifier 1 and test item identifier 2) corresponding to the two rare disease sample test data Equal to 2, the total number of test item identifiers (test item identifier 1, test item identifier 2, test item identifier 5, and test item identifier 6) after deduplication corresponding to the test data of two rare disease samples is equal to 4, = the absolute value of the difference in the test data values ​​corresponding to the detection item identification 1 + the absolute value of the difference in the test data values ​​corresponding to the detection item identification 2 + the absolute value of the difference in the test data values ​​corresponding to the detection item identification 5 + the absolute value of the difference in the test data values ​​corresponding to the detection item identification 6 = |97-93|+|3-5|+|40-0|+|0-42|=88, and then .

[0105] Here, the calculation formula of the data similarity takes into account the similarity of the number of detection item identifiers and the similarity of the detection data values ​​between the detection data of two rare disease samples.

[0106] iv. Determine the second data similarity corresponding to the initial clustering algorithm based on the first data similarity of all rare disease sample detection data sets corresponding to the initial clustering algorithm.

[0107] In the embodiment of the present application, the average or weighted average of the first data similarities of all rare disease sample detection data sets can be determined as the second data similarity corresponding to the initial clustering algorithm. The greater the second data similarity, the more accurate the clustering result of the initial clustering algorithm.

[0108] v. Determine the initial clustering algorithm with the highest second data similarity as the target clustering algorithm.

[0109] (3) Count the number of rare disease detection data corresponding to each severity level in each rare disease detection data set.

[0110] In an embodiment of the present application, each rare disease detection data set may contain rare disease detection data corresponding to multiple severity levels. If the number of rare disease detection data corresponding to a certain severity level is larger, it means that the possibility that the target severity level corresponding to the rare disease detection data set is that severity level is greater.

[0111] (4) The severity level with the largest number of rare disease detection data is determined as the target severity level corresponding to each rare disease detection data set.

[0112] (5) Determine the rare disease feature detection data corresponding to each target severity level based on all target rare disease detection data with severity levels corresponding to the target severity levels in each rare disease detection data set.

[0113] In an implementation manner of the present application, for each rare disease detection data set, the detection data values ​​at multiple moments corresponding to each detection item identifier in the target rare disease detection data under the target severity are weighted averaged to obtain the detection data values ​​at multiple moments corresponding to each detection item identifier in the rare disease feature detection data corresponding to each target severity.

[0114] Among them, the weight value when taking the weighted average of the test data values ​​corresponding to each test item identifier in the target rare disease test data under the target severity is determined according to the diagnosis time and the level of the diagnosis doctor corresponding to the target rare disease test data. The earlier the diagnosis time and the higher the level of the diagnosis doctor, the greater the weight value.

[0115] Step 2: Input the first detection data into a rare disease detection model to obtain at least one initial rare disease type.

[0116] In an embodiment of the present application, the rare disease detection model is obtained by training with detection sample data and corresponding rare disease type labels.

[0117] Step three: based on the second test data corresponding to the target user obtained by the test item identifier corresponding to the initial rare disease type, determine the target rare disease type from all the initial rare disease types.

[0118] In the embodiment of the present application, the detection item identifier corresponding to the second detection data is the same as the detection item identifier corresponding to the initial rare disease type. If the detection item identifier corresponding to the initial rare disease type is included in the target detection item subset, the detection data value corresponding to the detection item identifier in the first detection data is determined as the detection data value corresponding to the detection item identifier in the second detection data, without the target user rechecking the detection item corresponding to the detection item identifier.

[0119] Here, the target rare disease type determined in the embodiment of the present application can be one or more rare disease types. Through the method of the embodiment of the present application, the doctor can be provided with at least one rare disease type that the target user may have, and the doctor can be provided with effective treatment direction, thereby avoiding unnecessary examinations and unreasonable treatments for the target user, and shortening the treatment time and cycle of the target user.

[0120] Specifically, the processor 101 determines the target rare disease type from all the initial rare disease types by obtaining the second test data corresponding to the target user based on the test item identifier corresponding to the initial rare disease type through the following steps:

[0121] a. Determine the similarity between the second detection data and the rare disease comprehensive feature detection data corresponding to each initial rare disease type as the first detection probability corresponding to each initial rare disease type.

[0122] In the implementation mode of the present application, the rare disease comprehensive characteristic detection data corresponding to each initial rare disease type is obtained by combining the detection data corresponding to each initial rare disease type at all severity levels, and is used to characterize the characteristics of the detection data values ​​corresponding to various detection item identifiers of each initial rare disease type. Each initial rare disease type corresponds to a unique rare disease comprehensive characteristic detection data, and the rare disease comprehensive characteristic detection data contains a detection data value corresponding to at least one detection item identifier.

[0123] b. Input the second detection data into the rare disease detection model to obtain the second detection probability corresponding to each initial rare disease type.

[0124] In an implementation manner of the present application, the second detection data is input into a rare disease detection model, the output of the fully connected layer of the rare disease detection model is extracted, and the second detection probability corresponding to each initial rare disease type is obtained.

[0125] c. Calculate the target detection probability corresponding to each initial rare disease type based on the first detection probability and the second detection probability corresponding to each initial rare disease type.

[0126] In the implementation of the present application, the average or weighted average of the first detection probability and the second detection probability corresponding to each initial rare disease type is determined as the target detection probability corresponding to each initial rare disease type. The higher the target detection probability, the greater the probability that the target user has a rare disease of the initial rare disease type.

[0127] d. Determine the initial rare disease type with the highest target detection probability as the target rare disease type.

[0128] Step 4: Calculate the similarity between the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type and the rare disease feature test data corresponding to the target rare disease type at different severity levels using the following similarity formula.

[0129] ;

[0130] in, is the similarity between the detection data corresponding to the target user and the detection data of rare disease features, The number of detection item identifiers corresponding to the detection data corresponding to the target user, The number of detection times corresponding to the detection data of the target user, = the detection data value of the nth detection item at the ath detection time in the detection data corresponding to the target user minus the detection data value of the nth detection item at the ath detection time The value obtained after the detection data value at the detection time, = The value of the test data at the ath test time of the nth test item in the rare disease feature test data minus the value of the The difference between the detection data values ​​at the detection time is obtained. is the absolute value of the difference between the detection data value of the nth detection item at the i-th moment in the detection data corresponding to the target user and the detection data value of the nth detection item at the i-th moment in the rare disease feature detection data, is the first preset weight, is the second preset weight.

[0131] For example, assuming that the detection item identifiers corresponding to the detection data corresponding to the target user include {detection item identifier 1, detection item identifier 2}, the number of detection moments corresponding to the detection data corresponding to the target user is 3, and the detection data corresponding to the target user at the first detection moment is {90, 5}, the detection data corresponding to the target user at the second detection moment is {85, 3}, and the detection data corresponding to the target user at the third detection moment is {80, 6}, the rare disease feature detection data at the first detection moment is {96, 4}, the rare disease feature detection data at the second detection moment is {92, 3}, and the rare disease feature detection data at the third detection moment is {93, 6}.

[0132] From the above example, we can know that: the number of detection item identifiers (detection item identifier 1, detection item identifier 2) corresponding to the detection data of the target user ; The number of target users corresponding to the detection data and corresponding detection time .

[0133]

[0134] Step 5: Determine the severity with the highest similarity as the severity of the target user for the target rare disease type.

[0135] Furthermore, the processor 101 is also used to perform the following steps: obtaining a knowledge graph of rare disease treatment information; the rare disease treatment information knowledge graph contains multiple triples; the first entity in each triple is a text information containing the type of rare disease, the severity of the rare disease, and the detection data, and the second entity in each triple refers to the treatment text information; according to the target rare disease type, the severity of the target rare disease type for the target user, and the detection data corresponding to the target user obtained based on the detection item identifier corresponding to the target rare disease type, query the target user's treatment text information for the target rare disease type from the rare disease treatment information knowledge graph.

[0136] Through the device provided in the embodiment of the present application, the target rare disease type of the target user can be determined. Since the target rare disease type in the embodiment of the present application may include multiple rare disease types, the severity levels corresponding to multiple rare disease types may also be obtained. The results of the embodiment of the present application only serve to assist doctors. It is necessary to inform the target user's doctor of the target user's rare disease type, severity level and specific treatment information to provide direction for the doctor, thereby reducing the treatment time.

[0137] Based on the same inventive concept, the embodiments of the present application also provide a method for determining the severity of a rare disease corresponding to the device for determining the severity of a rare disease. Since the principle of solving the problem by the method in the embodiments of the present application is similar to the device for determining the severity of a rare disease in the embodiments of the present application, the implementation of the method can refer to the implementation of the device, and the repeated parts will not be repeated.

[0138] Reference Figure 2 As shown, it is a flow chart of a method for determining the severity of a rare disease provided in an embodiment of the present application. The exemplary steps of the embodiment of the present application are described below:

[0139] S201. Obtain rare disease feature detection data corresponding to each rare disease type at different severity levels, and first detection data of a target user determined according to a target detection item subset.

[0140] Among them, the target detection item subset is a subset selected from the full set of rare disease detection items that meets the preset conditions and has the highest rare disease detection accuracy; the full set of rare disease detection items includes the detection item identifiers corresponding to various rare disease types; the preset conditions include the quantity constraints and cost constraints of the target detection item subset; the rare disease feature detection data corresponding to each severity level is determined based on multiple rare disease detection data sets corresponding to the rare disease type; the rare disease detection data set is obtained by clustering all rare disease detection data corresponding to the rare disease type.

[0141] Specifically, rare disease feature detection data corresponding to each rare disease type at different severity levels are obtained, including: obtaining rare disease detection data corresponding to different severity levels based on the detection item identifier corresponding to the rare disease type; clustering the rare disease detection data through a target clustering algorithm to obtain multiple rare disease detection data sets; the target clustering algorithm is a clustering algorithm with the most accurate clustering results selected from multiple initial clustering algorithms; counting the number of rare disease detection data corresponding to each severity level in each rare disease detection data set; determining the severity level with the largest number of rare disease detection data as the target severity level corresponding to each rare disease detection data set; and determining the rare disease feature detection data corresponding to each target severity level based on all target rare disease detection data with a severity level corresponding to the target severity level in each rare disease detection data set.

[0142] Furthermore, a target clustering algorithm is selected from multiple initial clustering algorithms, including: obtaining multiple rare disease sample detection data; clustering all rare disease sample detection data through each initial clustering algorithm to obtain multiple rare disease sample detection data sets corresponding to each initial clustering algorithm; for each rare disease sample detection data set corresponding to each initial clustering algorithm, according to the data similarity between every two rare disease sample detection data in the rare disease sample detection data set, determining the first data similarity of the rare disease sample detection data set corresponding to the initial clustering algorithm; according to the first data similarity of all rare disease sample detection data sets corresponding to the initial clustering algorithm, determining the second data similarity corresponding to the initial clustering algorithm; and determining the initial clustering algorithm with the highest second data similarity as the target clustering algorithm.

[0143] Here, the data similarity between each two rare disease sample test data is calculated by the following formula:

[0144] ;

[0145] in, Detect the data similarity between two rare disease samples, The number of identical test item identifiers corresponding to the two rare disease sample test data. is the total number of test item identifiers after deduplication corresponding to the detection data of two rare disease samples, is the absolute value of the difference between the test data value of the i-th test item in the first rare disease sample test data and the test data value of the i-th test item in the second rare disease sample test data, and The preset weights.

[0146] S202. Input the first detection data into a rare disease detection model to obtain at least one initial rare disease type.

[0147] Among them, the rare disease detection model is trained by testing sample data and corresponding rare disease type labels.

[0148] S203. Determine a target rare disease type from all initial rare disease types based on the second test data corresponding to the target user obtained based on the test item identifier corresponding to the initial rare disease type.

[0149] Specifically, the similarity between the second detection data and the rare disease comprehensive feature detection data corresponding to each initial rare disease type is determined as the first detection probability corresponding to each initial rare disease type; the rare disease comprehensive feature detection data corresponding to each initial rare disease type is obtained by combining the detection data corresponding to each initial rare disease type at all severity levels; the second detection data is input into the rare disease detection model to obtain the second detection probability corresponding to each initial rare disease type; based on the first detection probability and the second detection probability corresponding to each initial rare disease type, the target detection probability corresponding to each initial rare disease type is calculated; the initial rare disease type with the highest target detection probability is determined as the target rare disease type.

[0150] S204. Calculate the similarity between the detection data corresponding to the target user obtained based on the detection item identifier corresponding to the target rare disease type and the rare disease feature detection data corresponding to the target rare disease type at different severity levels through the following similarity formula.

[0151] ;

[0152] in, is the similarity between the detection data corresponding to the target user and the detection data of rare disease features, The number of detection item identifiers corresponding to the detection data corresponding to the target user, The number of detection times corresponding to the detection data of the target user, = the detection data value of the nth detection item at the ath detection time in the detection data corresponding to the target user minus the detection data value of the nth detection item at the ath detection time The value obtained after the detection data value at the detection time, = The value of the test data at the ath test time of the nth test item in the rare disease feature test data minus the value of the The difference between the detection data values ​​at the detection time is obtained. is the absolute value of the difference between the detection data value of the nth detection item at the i-th moment in the detection data corresponding to the target user and the detection data value of the nth detection item at the i-th moment in the rare disease feature detection data, is the first preset weight, is the second preset weight.

[0153] S205. Determine the severity with the highest similarity as the severity of the target user for the target rare disease type.

[0154] Here, the method provided in the embodiment of the present application can provide doctors with at least one target rare disease type and severity that the target user may have, and provide doctors with effective treatment directions, thereby avoiding unnecessary examinations and unreasonable treatments for the target user and shortening the treatment time and cycle for the target user.

[0155] Reference Figure 3 As shown, it is a flow chart of another method for determining the severity of a rare disease provided in an embodiment of the present application. The exemplary steps of the embodiment of the present application are described below:

[0156] S301. Obtain at least one initial detection item subset that meets preset conditions and is selected from the entire set of rare disease detection items, and rare disease detection item sample data corresponding to each initial detection item subset.

[0157] Among them, the sample data of rare disease detection items carry rare disease type labels.

[0158] S302. Inputting the rare disease detection item sample data corresponding to each initial detection item subset into a rare disease detection model to obtain a rare disease type detection result of the rare disease detection item sample data.

[0159] S303. Determine the rare disease detection accuracy corresponding to each initial detection item subset based on the rare disease type detection results and rare disease type labels of the rare disease detection item sample data corresponding to each initial detection item subset.

[0160] S304: Determine the initial detection item subset with the highest rare disease detection accuracy as the target detection item subset.

[0161] Here, the method provided by the embodiment of the present application can determine a subset of target detection items, avoiding the target user from detecting all detection items, thereby effectively reducing the cost for users to determine the type of rare diseases.

[0162] Reference Figure 4 As shown, it is a flow chart of another method for determining the severity of a rare disease provided in an embodiment of the present application. The exemplary steps of the embodiment of the present application are described below:

[0163] S401. Obtain a knowledge graph of rare disease treatment information.

[0164] The rare disease treatment information knowledge graph contains multiple triplets; the first entity in each triplet is text information including the type of rare disease, the severity of the rare disease, and the detection data, and the second entity in each triplet refers to the treatment text information.

[0165] S402. According to the target rare disease type, the severity of the target rare disease type for the target user, and the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type, query the treatment text information of the target user for the target rare disease type from the rare disease treatment information knowledge graph.

[0166] Here, through the method provided in the embodiment of the present application, the treatment text information of the target user for the target rare disease type can be determined, and effective treatment directions can be provided to doctors, thereby avoiding unnecessary examinations and unreasonable treatments for the target user and shortening the treatment time and cycle of the target user.

[0167] Corresponding to the above-mentioned method for determining the severity of a rare disease, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for determining the severity of a rare disease are executed.

[0168] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0169] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0170] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0171] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the information processing method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0172] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A device for determining the severity of a rare disease, characterized in that: The device includes: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the following steps: Obtain rare disease feature detection data corresponding to each rare disease type at different severity levels, and first detection data of a target user determined according to a target detection item subset; the target detection item subset is a subset selected from the entire set of rare disease detection items that meets preset conditions and has the highest rare disease detection accuracy; the entire set of rare disease detection items includes detection item identifiers corresponding to various rare disease types; the preset conditions include quantity constraints and cost constraints for the target detection item subset; the rare disease feature detection data corresponding to each severity level is determined based on multiple rare disease detection data sets corresponding to the rare disease type; the rare disease detection data set is obtained by clustering all rare disease detection data corresponding to the rare disease type; Inputting the first detection data into a rare disease detection model to obtain at least one initial rare disease type; the rare disease detection model is obtained by training the detection sample data and the corresponding rare disease type label; Determining a target rare disease type from all initial rare disease types based on the second test data corresponding to the target user obtained from the test item identifier corresponding to the initial rare disease type; The similarity between the detection data corresponding to the target user obtained based on the detection item identifier corresponding to the target rare disease type and the rare disease feature detection data corresponding to the target rare disease type at different severity levels is calculated by the following similarity formula; ； in, is the similarity between the detection data corresponding to the target user and the detection data of rare disease features, The number of detection item identifiers corresponding to the detection data corresponding to the target user, The number of detection times corresponding to the detection data of the target user, = the detection data value of the nth detection item at the ath detection time in the detection data corresponding to the target user minus the detection data value of the nth detection item at the ath detection time The value obtained after the detection data value at the detection time is = The value of the test data at the ath test time of the nth test item in the rare disease feature test data minus the value of the The difference between the detection data values ​​at the detection time is obtained. is the absolute value of the difference between the detection data value of the nth detection item at the i-th moment in the detection data corresponding to the target user and the detection data value of the nth detection item at the i-th moment in the rare disease feature detection data, is the first preset weight, is the second preset weight; The severity with the highest similarity is determined as the severity of the target user for the target rare disease type.

2. The device for determining the severity of a rare disease according to claim 1, characterized in that: The processor selects the target detection item subset from the full set of rare disease detection items through the following steps: Acquire at least one initial detection item subset that meets preset conditions and is selected from the entire set of rare disease detection items, and rare disease detection item sample data corresponding to each initial detection item subset; the rare disease detection item sample data carries a rare disease type label; Inputting the rare disease detection item sample data corresponding to each initial detection item subset into the rare disease detection model to obtain the rare disease type detection result of the rare disease detection item sample data; Determine the rare disease detection accuracy rate corresponding to each initial detection item subset according to the rare disease type detection result of the rare disease detection item sample data corresponding to each initial detection item subset and the rare disease type label; The initial test item subset with the highest rare disease detection accuracy is determined as the target test item subset.

3. The device for determining the severity of a rare disease according to claim 1, characterized in that: The processor obtains rare disease feature detection data corresponding to each rare disease type at different severity levels through the following steps: Based on the detection item identifier corresponding to the rare disease type, obtain the corresponding rare disease detection data at different severity levels; Clustering the rare disease detection data by a target clustering algorithm to obtain multiple rare disease detection data sets; the target clustering algorithm is a clustering algorithm with the most accurate clustering result selected from multiple initial clustering algorithms; Count the number of rare disease detection data corresponding to each severity level in each rare disease detection data set; The severity level with the largest number of rare disease detection data is determined as the target severity level corresponding to each rare disease detection data set; According to all target rare disease detection data with severity corresponding to the target severity in each rare disease detection data set, the rare disease feature detection data corresponding to each target severity is determined.

4. The device for determining the severity of a rare disease according to claim 3, characterized in that: The processor selects a target clustering algorithm from a plurality of initial clustering algorithms by the following steps: Obtain test data for multiple rare disease samples; Clustering all rare disease sample detection data using each initial clustering algorithm to obtain multiple rare disease sample detection data sets corresponding to each initial clustering algorithm; For each rare disease sample detection data set corresponding to each initial clustering algorithm, determine a first data similarity of the rare disease sample detection data set corresponding to the initial clustering algorithm according to the data similarity between every two rare disease sample detection data in the rare disease sample detection data set; Determining a second data similarity corresponding to the initial clustering algorithm according to the first data similarity of all rare disease sample detection data sets corresponding to the initial clustering algorithm; The initial clustering algorithm with the highest second data similarity is determined as the target clustering algorithm.

5. The device for determining the severity of a rare disease according to claim 4, characterized in that: The processor calculates the data similarity between each two rare disease sample detection data using the following formula: ； in, Detect the data similarity between two rare disease samples, The number of identical test item identifiers corresponding to the two rare disease sample test data. is the total number of test item identifiers after deduplication corresponding to the detection data of two rare disease samples, is the absolute value of the difference between the test data value of the i-th test item in the first rare disease sample test data and the test data value of the i-th test item in the second rare disease sample test data, and The preset weights.

6. The device for determining the severity of a rare disease according to claim 1, characterized in that: The processor determines a target rare disease type from all initial rare disease types by obtaining the second test data corresponding to the target user based on the test item identifier corresponding to the initial rare disease type through the following steps: Determine the similarity between the second detection data and the rare disease comprehensive feature detection data corresponding to each initial rare disease type as the first detection probability corresponding to each initial rare disease type; the rare disease comprehensive feature detection data corresponding to each initial rare disease type is obtained by combining the detection data corresponding to each initial rare disease type at all severity levels; Inputting the second detection data into the rare disease detection model to obtain a second detection probability corresponding to each initial rare disease type; Calculate the target detection probability corresponding to each initial rare disease type according to the first detection probability and the second detection probability corresponding to each initial rare disease type; The initial rare disease type with the highest target detection probability is determined as the target rare disease type.

7. The device for determining the severity of a rare disease according to any one of claims 1 to 6, characterized in that: The processor is further configured to execute the following steps: Obtaining a rare disease treatment information knowledge graph; the rare disease treatment information knowledge graph includes a plurality of triples; the first entity in each triple is text information including the type of rare disease, the severity of the rare disease, and test data, and the second entity in each triple refers to treatment text information; According to the target rare disease type, the severity of the target rare disease type for the target user, and the test data corresponding to the target user obtained based on the test item identifier corresponding to the target rare disease type, the treatment text information of the target user for the target rare disease type is queried from the rare disease treatment information knowledge graph.

Images