WO2021180242A1 - Method and apparatus for detecting anomaly in diagnostic data, and computer device and storage medium - Google Patents

Abstract

A method and apparatus for detecting an anomaly in diagnostic data, and a computer device and a storage medium, which belong to the field of intelligent medical treatment. The method comprises: respectively matching diagnostic data of a target patient with two rules, i.e. a preset medical treatment rule and a medical treatment mining rule, so as to obtain two types of candidate information, and by means of fusing the two types of candidate information, obtaining multi-dimensional third candidate information in which the medical treatment rule and the medical treatment mining rule are combined; using a disease recognition model to recognize the diagnostic data, so as to acquire fourth candidate information, such that the flexibility of recognition of the diagnostic data is improved and the recognition speed is fast; and determining suspected disease information of the target patient by means of combining the fourth candidate information and the third candidate information, so as to determine, according to the suspected disease information, whether the diagnostic data is anomalous, and thereby achieving the aim of quickly and effectively confirming a misdiagnosis.

Description

Diagnostic data abnormal detection method, device, computer equipment and storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 27, 2020, the application number is 202011161090.4, and the invention title is "diagnostic data anomaly detection method, device, computer equipment and storage medium". The entire content of the application is approved The reference is incorporated in the application.

Technical field

This application relates to the field of digital medicine, and in particular to a method, device, computer equipment, and storage medium for detecting abnormal diagnostic data.

Background technique

Misdiagnosis means that the doctor gives the wrong diagnosis to the patient due to various reasons. Misdiagnosis is very common. According to survey data, the misdiagnosis rate of diseases is usually around 30%. Misdiagnosis can lead to serious consequences, such as wrong treatment plans and delays in patient treatment. Therefore, timely detection of misdiagnosis is very necessary.

The inventor realizes that the existing misdiagnosis detection method is based on the doctors compiling medical rules for each disease according to medical knowledge to perform misdiagnosis detection on the patient’s diagnosis, by judging whether the diagnosed disease meets the medical rules corresponding to the disease, if not Satisfying the rules means that this diagnosis is misdiagnosed. However, the existing misdiagnosis detection methods have problems: medical rules are compiled by doctors based on medical knowledge, which consumes a lot of energy, high time cost, low flexibility, and low accuracy of misdiagnosis detection.

Summary of the invention

Aiming at the problems of poor flexibility and low detection accuracy of existing misdiagnosis detection methods, a diagnostic data abnormality detection method, device, computer equipment and storage medium aiming to improve the flexibility and detection accuracy of misdiagnosis detection are now provided.

To achieve the foregoing objective, the first aspect of the present application provides a method for detecting anomalies in diagnostic data, including:

Obtain the diagnostic data of the target patient;

Matching the diagnostic data with preset medical rules to obtain first candidate information;

Matching the diagnostic data with medical mining rules to obtain second candidate information;

Fusing the first candidate information and the second candidate information to generate third candidate information;

Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.

To achieve the foregoing objective, a second aspect of the present application provides a diagnostic data abnormality detection device, including:

The acquiring unit is used to acquire the diagnostic data of the target patient;

The first matching unit is configured to match the diagnostic data with preset medical rules to obtain first candidate information;

The second matching unit is used to match the diagnostic data with medical mining rules to obtain second candidate information;

A fusion unit for fusing the first candidate information and the second candidate information to generate third candidate information;

An identification unit, configured to use a disease identification model to identify the diagnostic data to obtain fourth candidate information;

The processing unit is configured to fuse the third candidate information and the fourth candidate information to obtain suspected disease information, and determine whether the diagnosis data is abnormal according to the suspected disease information.

In order to achieve the above objective, the third aspect of the present application provides a diagnostic data abnormality detection device, including: a memory and at least one processor, the memory stores instructions; the at least one processor calls all the devices in the memory The instructions, so that the diagnostic data abnormality detection device executes the steps of the diagnostic data abnormality detection method as shown below:

Obtain the diagnostic data of the target patient;

Matching the diagnostic data with preset medical rules to obtain first candidate information;

Matching the diagnostic data with medical mining rules to obtain second candidate information;

Fusing the first candidate information and the second candidate information to generate third candidate information;

Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.

In order to achieve the above objectives, the fourth aspect of the present application provides a computer-readable storage medium, which stores instructions in the computer-readable storage medium, which when run on a computer, causes the computer to execute the following diagnosis data abnormality Steps of the detection method:

Obtain the diagnostic data of the target patient;

Matching the diagnostic data with preset medical rules to obtain first candidate information;

Matching the diagnostic data with medical mining rules to obtain second candidate information;

Fusing the first candidate information and the second candidate information to generate third candidate information;

Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.

The diagnostic data abnormality detection method, device, computer equipment and storage medium provided in this application can match the target patient’s diagnostic data with preset medical rules and medical mining rules to obtain two kinds of candidate information. The two kinds of candidate information are fused to obtain the multi-dimensional third candidate information that combines medical rules and medical mining rules; the disease recognition model is used to identify the diagnostic data to obtain the fourth candidate information, which improves the flexibility of identifying diagnostic data, and The recognition speed is fast; the suspected disease information of the target patient is determined by combining the fourth candidate information and the third candidate information, so as to determine whether the diagnosis data is abnormal according to the suspected disease information, so as to achieve the purpose of quickly and effectively confirming the misdiagnosis.

Description of the drawings

FIG. 1 is a flowchart of an embodiment of the method for detecting abnormality of diagnostic data according to this application;

FIG. 2 is a flowchart of an embodiment of obtaining diagnostic data of a target patient in this application;

FIG. 3 is a flowchart of an embodiment of obtaining first candidate information in this application;

FIG. 4 is a flowchart of an embodiment of generating the medical mining rule according to historical sample data in this application;

FIG. 5 is a flowchart of an embodiment of obtaining second candidate information in this application;

FIG. 6 is a flowchart of an embodiment of fusing the third candidate information and the fourth candidate information to obtain suspected disease information;

FIG. 7 is a block diagram of an embodiment of the diagnostic data abnormality detection device described in this application;

FIG. 8 is a hardware architecture diagram of an embodiment of the computer device of this application.

Detailed ways

In order to make the purpose, technical solutions, and advantages of this application clearer and clearer, the following further describes the application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the application, and are not used to limit the application. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

It should be noted that the embodiments in the application and the features in the embodiments can be combined with each other if there is no conflict.

The diagnostic data abnormality detection method, device, computer equipment and storage medium provided in this application are suitable for the field of intelligent medical business. In this application, the diagnostic data of the target patient can be matched with preset medical rules and medical mining rules to obtain two kinds of candidate information. By fusing the two kinds of candidate information to obtain a combination of medical rules and medical mining rules The multi-dimensional third candidate information; the disease recognition model is used to identify the diagnostic data to obtain the fourth candidate information, which improves the flexibility of identifying the diagnostic data, and the recognition speed is fast; by combining the fourth candidate information and the third candidate information Determine the suspected disease information of the target patient in order to determine whether the suspected disease information contains the disease type in the diagnostic data, that is, whether the diagnostic data is abnormal, if it is, it means that the diagnostic data is normal; if not, it means that the diagnostic data is abnormal and misdiagnosed. So as to achieve the purpose of quickly and effectively confirming the misdiagnosis.

Example one

Referring to FIG. 1, a method for detecting anomalies in diagnostic data in this embodiment includes the following steps:

S1. Obtain the diagnostic data of the target patient.

Further, referring to step S1 shown in FIG. 2 may include the following steps:

S11. Receive medical data of the target patient sent by the user terminal, where the medical data includes: basic information of the target patient, the target disease type, and multiple medical entities.

Among them, the basic information of the target patient may include: identification number of the target patient (such as ID card, medical insurance card number, etc.), age, gender, chief complaint, current medical history, family history and other information; the target disease type can be a disease classification number ; The medical entity can be the number of the test item, for example: blood test items (such as blood pressure, hemoglobin, platelets, etc.), urine test items (such as protein, ketone body, glucose), etc.

S12. Extract the medical entity in the medical data, and generate the diagnostic data.

In this embodiment, the diagnostic data is composed of the ID of the medical entity, such as: [medical entity X1, medical entity X2, medical entity X3, ...], X in the medical entity represents the ID of the medical entity.

S2. Match the diagnostic data with preset medical rules to obtain first candidate information.

Wherein, the preset medical rules are rules set in advance according to medical knowledge (doctors sorted out from medical knowledge), including multiple medical rules, each medical rule includes at least one medical entity, and each medical rule includes at least one medical entity. The medical rule corresponds to a type of disease.

As an example and not a limitation, the presentation of medical rules is generally, for example, disease 1 → (medical entity 1, medical entity 3): it means that both medical entity 1 and medical entity 3 appear in the diagnostic data, and disease 1 can be added to the target. In the patient’s first candidate information; disease 2 → (medical entity 1, medical entity 5, medical entity 10): indicates that medical entity 1, medical entity 5, and medical entity 10 appear in the diagnostic data at the same time to add disease 2 to the target The patient’s first candidate information.

Further, referring to step S2 shown in FIG. 3, it may include the following steps:

S21. The multiple medical entities in the diagnostic data are matched with each of the medical rules in the preset medical rules to obtain the matching degree of the disease type matched with the diagnostic data.

In this embodiment, when the medical rules are matched, all medical entities in the diagnosis data are matched with all the entities in each medical rule. If all the entities in the medical rules match all or part of the entities in the diagnosis data , It can be confirmed that the diagnostic data matches the disease type corresponding to the medical rule. If some entities in the medical rule match all or part of the entities in the diagnostic data, it means that the diagnostic data does not match the disease type corresponding to the medical rule .

For example: matching the diagnostic data (including multiple medical entities) of the target patient to the preset medical rules, and the following results are obtained. Suppose there are 5 rules in total, corresponding to 3 types of diseases, see Table 1:

Table 1

Medical rules

Type of disease

Rule matching result

suitability

Rule 1	Disease 1	match	1
Rule 2	Disease 1	Mismatch	0
Rule 3	Disease 2	Mismatch	0
Rule 4	Disease 2	Mismatch	0
Rule 5	Disease 3	match	1

In the preset medical rule, the matching degree of the disease type corresponding to the matched medical rule is set to 1. If there are multiple medical rules for a disease type, as long as the diagnosis data matches any one, the matching degree of the diagnosis data for the disease is set to 1. For example, both rule 1 and rule 2 in the above table correspond to disease 1, although the diagnosis data only Rule 1 is matched, but the matching degree of the diagnosis data to disease 1 is also set to 1.

S22. Extract the matching degrees of all disease types that match the diagnosis data, and generate the first candidate information.

Combining with Table 1, extracting the disease type with a matching degree of 1, the first candidate information is: {disease 1:1, disease 3:1}.

In an embodiment, before step S3 is executed, the method may further include: generating the medical mining rule according to historical sample data.

It should be noted that the historical sample data includes multiple pieces of historical medical data, and each piece of historical medical data includes the disease type of the historical patient and multiple medical entities. The medical mining rule includes a plurality of mining rules, each of the mining rules includes at least one of the medical entities, and each of the mining rules corresponds to a type of disease.

Further, referring to FIG. 4, generating the medical mining rule based on historical sample data includes the following steps:

A1. Classify the historical medical data in the historical sample data according to the disease type to generate a disease type set.

Specifically, the weight value of each medical entity and the corresponding disease type is calculated; medical entities whose weight value is greater than or equal to the weight threshold in the historical sample data are extracted one by one.

The historical medical data is classified according to the disease type, and the weight value is calculated for each medical entity in the historical medical data corresponding to each disease type. Filter out the medical entities whose weight value is less than the weight threshold (the weight threshold is between 0 and 1, [0,1]). The purpose of filtering medical entities of each disease type is to reduce interference, remove the interference of frequently occurring medical entities with low discrimination and low importance, and improve the quality of data.

Among them, the calculation formula of the weight value: weight[medical entity i, disease j]=(the number of times the medical entity i appears in the disease j)/(how many diseases the medical entity i appears in). i represents the ID of the medical entity; j represents the ID of the medical entity. If weight[medical entity i, disease j]<threshold, remove the medical entity.

A2. Use a frequent set mining algorithm to filter the historical medical data in the disease type set, and generate mining rules corresponding to the disease type.

The frequent set mining algorithm is used to filter multiple historical sample data corresponding to each disease type after filtering based on the preset support (min_support) and confidence (min_confidence) thresholds to obtain the mining corresponding to each disease type Rules, such as: for disease 1, three data-based rules are obtained after meeting two thresholds, namely: {[medical entity 1, medical entity 3, medical entity 9], [medical entity 1, medical entity 3, medical entity 5. Medical entity 7], [medical entity 1, medical entity 5, medical entity 7, medical entity 9, medical entity 10]}. The mining rules corresponding to each disease type are arranged in descending order of support.

Frequent-Pattern Growth (FP-growth) is an iterative method called layer-by-layer search. For example, k-itemsets are used to explore (k+1)-itemsets. First, find the set of frequent 1-itemsets, denote the set as L ₁ , use set L ₁ to find the set L _{2 of} frequent 2-items sets, and then use the set L ₂ to find L ₃ , so By analogy, until no frequent k-itemsets can be found, a database scan is required to _{find each L k.} S3. Match the diagnostic data with medical mining rules to obtain second candidate information.

Further, referring to step S3 shown in FIG. 5, it may include the following steps:

S31. Match a plurality of the medical entities in the diagnostic data with each of the mining rules in the medical mining rules to obtain the matching degree of the disease type matching the diagnostic data.

The diagnostic data is matched with all mining rules. For a disease type, if a certain rule of the disease type is matched, the support degree corresponding to the rule is used as the matching degree of the data to the disease; if there is no match For any rule of this disease type, the matching degree of historical medical data to this disease type is set to 0. By analogy, the possibility list of all disease types of the diagnosis data is obtained, and the disease types in the list are sorted according to the degree of matching from large to small. Note that if a piece of data matches multiple rules for a disease, the rule with the largest matching degree is selected.

For example: for disease 1, the support threshold is set to 0.7, corresponding to the 3 mining rules in Table 2, and the support corresponding to each mining rule is as follows:

Table 2

serial number	Mining rules		Support
1	Medical entity 1, medical entity 3, medical entity 9	0.80
2	Medical entity 1, medical entity 3, medical entity 5, medical entity 7	0.75
3	Medical entity 1, medical entity 5, medical entity 7, medical entity 9, medical entity 10	0.70

When the diagnostic data contains 4 medical entities: [medical entity 1, medical entity 2, medical entity 3, medical entity 9], for disease 1, the diagnostic data corresponds to rule 1 in Table 2 (this piece of data contains rule 1 All three medical entities in ), that is, the probability that the diagnostic data corresponds to disease 1 is 0.80.

S32. Extract the matching degrees of all disease types that match the diagnosis data, and generate the second candidate information.

Take 10 disease types as an example: match the diagnosis data with the mining rules corresponding to all disease types, as shown in Table 3:

table 3

serial number	Rule matching result	Support
Disease 1	Match a rule	0.80
Disease 2	No rules were matched	0
Disease 3	No rules were matched	0
Disease 4	Match a rule	0.85
Disease 5	No rules were matched	0
Disease 6	No rules were matched	0
Disease 7	Match a rule	0.70
Disease 8	No rules were matched	0
Disease 9	No rules were matched	0
Disease 10	No rules were matched	0

Three disease types are obtained, and the support (matching degree) of each disease type that matches is extracted. The second candidate information is: {disease 1:0.82, disease 4:0.85, disease 7:0.70}.

It should be emphasized that, in order to further ensure the privacy and security of the aforementioned preset medical rules and medical mining rules, the aforementioned preset medical rules and medical mining rules can also be stored in a node of a blockchain. The blockchain referred to in this application is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain, essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of the information (anti-counterfeiting) and the generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

S4. Fusion of the first candidate information and the second candidate information to generate third candidate information.

Further, step S4 may include: according to the disease type in the first candidate information and the disease type in the second candidate information, the matching degree and the corresponding degree of matching of the same disease type in the first candidate information The corresponding matching degree in the second candidate information calculates a matching average value, and generates the third candidate information including the disease type matching average value.

In the embodiment, the matching degree in the first candidate information and the second candidate information is weighted and fused to obtain the fused third candidate information, for example, the third candidate information corresponding to the diagnostic data L _rule = {disease 4:0.95, Disease 2: 0.90, Disease 1: 0.5, ...}.

S5. Use a disease recognition model to recognize the diagnostic data to obtain fourth candidate information.

Wherein, the fourth candidate information includes the matching value of the disease type.

In this embodiment, the disease recognition model adopts the BERT (Bidirectional Encoder Representations from Transformers) model, and the input of the BERT is diagnostic data, and the output is the fourth candidate information.

The BERT model is a language model in the field of natural language processing, which can directly process natural language without conversion. However, the patient's consultation data not only contains unstructured data, that is, free text data, but also contains a large amount of structured data, and the BERT model can only process free text data. In order to enable the BERT model to process structured and unstructured data at the same time, this embodiment improves the BERT model. The unstructured data and structured data are spliced, and then input into the BERT model, where each word (word, w) of the unstructured data is input into a token, and each code of the structured data (code , C) Correspondingly input into a token. The BERT model is improved, the original segmentembedding layer in the BERT model is removed, and the original positionembedding layer is improved. Unstructured data, that is, text, is sequential, so the positionembedding token corresponding to unstructured data has positional embedded expression. However, structured data has no order, and all positionembedding tokens for all structured data are set to the same embedded expression.

Based on the Chinese pre-trained BERT model, structured data is added to the dictionary for expansion. On the basis of this Chinese pre-training model, pre-training is performed to update the embedding representation of structured data and the parameters of the model. On the basis of the pre-trained model, the downstream task of performing suspected disease judgment fine-tuning the model, where FC is a fully connected layer (fullyconneted layer), and the output is a list of suspected diseases, for example: fourth candidate information L _deep = {disease 2:0.98, disease 4:0.80, disease 1:0.2,...}.

S6. Fusion of the third candidate information and the fourth candidate information to obtain suspected disease information, and determine whether the diagnostic data is abnormal according to the suspected disease information.

Further, referring to step S6 shown in FIG. 6 may include the following steps:

S61. Extract the type of disease whose suspected value meets the preset condition, and generate the suspected disease information.

S62. Match the target disease type in the diagnostic data with the disease type in the suspected disease information. If it matches, it means that the diagnostic data is normal, and if it does not match, it means that the diagnostic data is abnormal. .

The third candidate information L _rule and the fourth candidate information L _deep are weighted and fused, and the formula is suspected disease information L = w _rule × L _rule + w _deep × L _deep , where w _rule and w _deep are preset coefficients, w _rule +w _deep =1. For example: _{the value of L rule} is 0.8 and the value of L _deep is 0.5, then the value of this disease in L is w _rule ×0.8+w _deep ×0.5. Take the first K (positive integer) disease types with the highest probability value to constitute the suspected disease information of the target patient. If the target disease type (the actual disease type) of the target patient is not in the suspected disease information, it means that the diagnostic data of the target patient is misdiagnosed.

In this embodiment, the diagnostic data abnormality detection method can match the target patient's diagnostic data with preset medical rules and medical mining rules to obtain two kinds of candidate information. The two kinds of candidate information are merged to Obtain multi-dimensional third candidate information that combines medical rules and medical mining rules; use disease recognition models to identify diagnostic data to obtain fourth candidate information, which improves the flexibility of identifying diagnostic data, and the recognition speed is fast; The fourth candidate information and the third candidate information determine the suspected disease information of the target patient, so as to determine whether the diagnosis data is abnormal according to the suspected disease information, so as to achieve the purpose of quickly and effectively confirming the misdiagnosis.

Example two

Referring to FIG. 7, a diagnostic data abnormality detection device 1 of this embodiment includes: an acquisition unit 11, a first matching unit 12, a second matching unit 13, a fusion unit 14, an identification unit 15 and a processing unit 16.

The acquiring unit 11 is used to acquire diagnostic data of the target patient.

Further, the acquiring unit 11 is configured to receive medical data of the target patient sent by the user terminal, the medical data including: basic information of the target patient, the target disease type, and multiple medical entities.

Among them, the basic information of the target patient may include: identification number of the target patient (such as ID card, medical insurance card number, etc.), age, gender, chief complaint, current medical history, family history and other information; the target disease type can be a disease classification number ; The medical entity can be the number of the test item, for example: blood test items (such as blood pressure, hemoglobin, platelets, etc.), urine test items (such as protein, ketone body, glucose), etc.

The acquiring unit 11 is also used to extract the medical entity in the medical data to generate the diagnostic data.

In this embodiment, the diagnostic data is composed of the ID of the medical entity, such as: [medical entity X1, medical entity X2, medical entity X3, ...], X in the medical entity represents the ID of the medical entity.

The first matching unit 12 is configured to match the diagnostic data with preset medical rules to obtain first candidate information.

Wherein, the preset medical rules are rules set in advance according to medical knowledge (doctors sorted out from medical knowledge), including multiple medical rules, each medical rule includes at least one medical entity, and each medical rule includes at least one medical entity. The medical rule corresponds to a type of disease.

Further, the first matching unit 12 is configured to match a plurality of the medical entities in the diagnostic data with each of the medical rules in the preset medical rules to obtain a match with the diagnostic data The degree of matching of the type of disease. The first matching unit 12 is also used to extract the matching degrees of all disease types that match the diagnosis data, and generate the first candidate information.

The second matching unit 13 is configured to match the diagnostic data with medical mining rules to obtain second candidate information.

Further, the second matching unit 13 is configured to match a plurality of the medical entities in the diagnostic data with each of the mining rules in the medical mining rules, to obtain the matching with the diagnostic data The matching degree of disease types; the second matching unit 13 is also used to extract the matching degrees of all disease types matching the diagnosis data, and generate the second candidate information.

In this embodiment, the medical mining rules are acquired based on historical sample data. The specific acquisition process is (see Figure 4):

A1. Classify the historical medical data in the historical sample data according to the disease type to generate a disease type set.

A2. Use a frequent set mining algorithm to filter the historical medical data in the disease type set, and generate mining rules corresponding to the disease type.

The fusion unit 14 is configured to fuse the first candidate information and the second candidate information to generate third candidate information.

Further, the fusion unit 14 may, according to the disease type in the first candidate information and the disease type in the second candidate information, determine the matching degree of the same disease type in the first candidate information and the matching degree in the first candidate information. The corresponding matching degree in the second candidate information calculates a matching average value, and generates the third candidate information including the disease type matching average value.

The identification unit 15 is configured to identify the diagnosis data using a disease identification model to obtain fourth candidate information.

Wherein, the fourth candidate information includes the matching value of the disease type.

In this embodiment, the disease recognition model adopts the BERT (Bidirectional Encoder Representations from Transformers) model, and the input of the BERT is diagnostic data, and the output is the fourth candidate information.

The processing unit 16 is configured to fuse the third candidate information and the fourth candidate information to obtain suspected disease information, and determine whether the diagnosis data is abnormal according to the suspected disease information.

Further, the processing unit 16 extracts the disease type whose suspect value meets the preset condition to generate the suspected disease information; compares the target disease type in the diagnostic data with the disease type in the suspected disease information Matching is performed. If it matches, it means that the diagnostic data is normal, and if it does not match, it means that the diagnostic data is abnormal.

In this embodiment, the diagnostic data abnormality detection device 1 matches the target patient's diagnostic data with preset medical rules through the first matching unit 12, and uses the second matching unit 13 to match the target patient's diagnostic data with medical mining rules. Two kinds of rules are matched to obtain two kinds of candidate information; the fusion unit 14 is used to fuse the two kinds of candidate information to obtain a multi-dimensional third candidate information that combines medical rules and medical mining rules; and the disease recognition in the recognition unit 15 The model recognizes the diagnostic data to obtain the fourth candidate information, which improves the flexibility of identifying the diagnostic data, and the recognition speed is fast; the processing unit 16 combines the fourth candidate information and the third candidate information to determine the suspected disease information of the target patient, so that To determine whether the diagnosis data is abnormal based on the suspected disease information, so as to achieve the purpose of quickly and effectively confirming the misdiagnosis.

Example three

In order to achieve the above objective, the present application also provides a computer device 2 which includes a plurality of computer devices 2. The components of the diagnostic data abnormality detection device 1 of the second embodiment can be dispersed in different computer devices 2. Device 2 can be a smart phone, a tablet, a laptop, a desktop computer, a rack server, a blade server, a tower server, or a rack server (including an independent server, or a server cluster composed of multiple servers) that executes the program )Wait. The computer device 2 of this embodiment at least includes but is not limited to: a memory 21, a processor 23, a network interface 22, and a diagnostic data abnormality detection device 1 (refer to FIG. 8) that can be communicably connected to each other through a system bus. It should be pointed out that FIG. 8 only shows the computer device 2 with components, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

In this embodiment, the memory 21 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access Memory (RAM), Static Random Access Memory (SRAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Programmable Read Only Memory (PROM), Magnetic Memory, Magnetic Disk, Optical Disk, etc. In some embodiments, the memory 21 may be an internal storage unit of the computer device 2, for example, a hard disk or a memory of the computer device 2. In other embodiments, the memory 21 may also be an external storage device of the computer device 2, such as a plug-in hard disk, a smart media card (SMC), and a secure digital (Secure Digital, SMC) equipped on the computer device 2. SD) card, flash card (Flash Card), etc. Of course, the memory 21 may also include both the internal storage unit of the computer device 2 and its external storage device. In this embodiment, the memory 21 is generally used to store the operating system and various application software installed in the computer device 2, such as the program code of the diagnostic data abnormality detection method in the first embodiment. In addition, the memory 21 can also be used to temporarily store various types of data that have been output or will be output.

In some embodiments, the processor 23 may be a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 23 is generally used to control the overall operation of the computer device 2, for example, to perform data interaction or communication-related control and processing with the computer device 2. In this embodiment, the processor 23 is used to run the program code or processing data stored in the memory 21, for example, to run the diagnostic data abnormality detection device 1 and the like.

The network interface 22 may include a wireless network interface or a wired network interface, and the network interface 22 is generally used to establish a communication connection between the computer device 2 and other computer devices 2. For example, the network interface 22 is used to connect the computer device 2 with an external terminal through a network, and establish a data transmission channel and a communication connection between the computer device 2 and the external terminal. The network may be Intranet, Internet, Global System of Mobile Communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G Network, Bluetooth (Bluetooth), Wi-Fi and other wireless or wired networks.

It should be pointed out that FIG. 8 only shows the computer device 2 with components 21-23, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

In this embodiment, the diagnostic data abnormality detection device 1 stored in the memory 21 may also be divided into one or more program modules, and the one or more program modules are stored in the memory 21 and consist of one Or executed by multiple processors (in this embodiment, the processor 23) to complete the application.

Example four

In order to achieve the above objective, the present application also provides a computer-readable storage medium. The computer-readable storage medium may be non-volatile or volatile, and includes multiple storage media, such as flash memory, hard disk, and multimedia. Card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), Programmable read-only memory (PROM), magnetic memory, magnetic disks, optical disks, servers, App application malls, etc., have computer programs stored thereon, and corresponding functions are realized when the programs are executed by the processor 23. The computer-readable storage medium of this embodiment is used to store the diagnostic data abnormality detection device 1, and when executed by the processor 23, the diagnostic data abnormality detection method of the first embodiment is implemented.

The serial numbers of the foregoing embodiments of the present application are for description only, and do not represent the superiority or inferiority of the embodiments.

Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。

The above are only the preferred embodiments of the application, and do not limit the scope of the patent for this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of the application, or directly or indirectly applied to other related technical fields , The same reason is included in the scope of patent protection of this application.

Claims (20)

1. A method for detecting abnormal diagnostic data, which includes:

   Obtain the diagnostic data of the target patient;

   Matching the diagnostic data with preset medical rules to obtain first candidate information;

   Matching the diagnostic data with medical mining rules to obtain second candidate information;

   Fusing the first candidate information and the second candidate information to generate third candidate information;

   Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

   The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.
2. The diagnostic data abnormality detection method according to claim 1, wherein acquiring the diagnostic data of the target patient comprises:

   Receiving medical data of the target patient sent by the user terminal, the medical data including: basic information of the target patient, the target disease type, and multiple medical entities;

   The medical entity in the medical data is extracted to generate the diagnostic data.
3. The diagnostic data abnormality detection method according to claim 2, wherein the preset medical rule is a rule set in advance according to medical knowledge, including a plurality of medical rules, and each medical rule includes at least one of the medical entities , Each of the medical rules corresponds to a type of disease;

   Matching the diagnostic data with preset medical rules to obtain first candidate information includes:

   Matching the multiple medical entities in the diagnostic data with each of the medical rules in the preset medical rules to obtain the matching degree of the disease type matching the diagnostic data;

   The matching degrees of all disease types matching the diagnosis data are extracted, and the first candidate information is generated.
4. The diagnostic data abnormality detection method according to claim 3, wherein matching the diagnostic data with medical mining rules to obtain the second candidate information further comprises:

   Generate the medical mining rule based on historical sample data:

   The historical sample data includes multiple pieces of historical medical data, and each piece of historical medical data includes the disease type of the historical patient and multiple medical entities;

   The medical mining rule includes a plurality of mining rules, each of the mining rules includes at least one of the medical entities, and each of the mining rules corresponds to a type of disease;

   Classify the historical medical data in the historical sample data according to the disease type to generate a disease type set;

   A frequent set mining algorithm is used to screen the historical medical data in the disease type set, and a mining rule corresponding to the disease type is generated.
5. The diagnostic data abnormality detection method according to claim 4, wherein matching the diagnostic data with medical mining rules to obtain the second candidate information comprises:

   Matching a plurality of the medical entities in the diagnostic data with each of the mining rules in the medical mining rules to obtain the matching degree of the disease type matching the diagnostic data;

   Extract the matching degrees of all disease types that match the diagnostic data, and generate the second candidate information.
6. The diagnostic data abnormality detection method according to claim 5, wherein the fusion of the first candidate information and the second candidate information to generate the third candidate information comprises:

   According to the disease type in the first candidate information and the disease type in the second candidate information, the matching degree for the same disease type in the first candidate information and the corresponding degree in the second candidate information The matching degree calculates the matching average value, and generates the third candidate information including the disease type matching average value.
7. The diagnostic data abnormality detection method according to claim 5, wherein the fourth candidate information includes a matching value of a disease type;

   The fusion of the third candidate information and the fourth candidate information to obtain suspected disease information, and judging whether the diagnosis data is abnormal according to the suspected disease information includes:

   Calculating a suspicion value for the matching average value corresponding to the same disease type in the third candidate information and the corresponding matching value in the fourth candidate information;

   Extracting the type of disease whose suspected value meets the preset condition, and generating the suspected disease information;

   The target disease type in the diagnostic data is matched with the disease type in the suspected disease information. If it matches, it means that the diagnostic data is normal, and if it does not match, it means that the diagnostic data is abnormal.
8. A diagnostic data abnormality detection device, which includes:

   The acquiring unit is used to acquire the diagnostic data of the target patient;

   The first matching unit is configured to match the diagnostic data with preset medical rules to obtain first candidate information;

   The second matching unit is used to match the diagnostic data with medical mining rules to obtain second candidate information;

   A fusion unit for fusing the first candidate information and the second candidate information to generate third candidate information;

   An identification unit, configured to use a disease identification model to identify the diagnostic data to obtain fourth candidate information;

   The processing unit is configured to fuse the third candidate information and the fourth candidate information to obtain suspected disease information, and determine whether the diagnosis data is abnormal according to the suspected disease information.
9. A diagnostic data abnormality detection device, wherein the diagnostic data abnormality detection device includes a memory and at least one processor;

   The at least one processor invokes the instructions in the memory, so that the diagnostic data abnormality detection device executes the steps of the diagnostic data abnormality detection method as described below:

   Obtain the diagnostic data of the target patient;

   Matching the diagnostic data with preset medical rules to obtain first candidate information;

   Matching the diagnostic data with medical mining rules to obtain second candidate information;

   Fusing the first candidate information and the second candidate information to generate third candidate information;

   Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

   The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.
10. The diagnostic data abnormality detection device according to claim 9, wherein, when the diagnostic data abnormality detection device is executed by the processor, the step of acquiring the diagnostic data of the target patient comprises:

    Receiving medical data of the target patient sent by the user terminal, the medical data including: basic information of the target patient, the target disease type, and multiple medical entities;

    The medical entity in the medical data is extracted to generate the diagnostic data.
11. The diagnostic data abnormality detection device according to claim 10, wherein the preset medical rule is a rule set in advance based on medical knowledge, including a plurality of medical rules, and each of the medical rules includes at least one of the medical entities , Each of the medical rules corresponds to a type of disease;

    When the diagnostic data abnormality detection device is executed by the processor, the step of matching the diagnostic data with preset medical rules to obtain first candidate information includes:

    Matching the multiple medical entities in the diagnostic data with each of the medical rules in the preset medical rules to obtain the matching degree of the disease type matching the diagnostic data;

    The matching degrees of all disease types matching the diagnosis data are extracted, and the first candidate information is generated.
12. The diagnostic data anomaly detection device according to claim 11, wherein the diagnostic data anomaly detection device is executed by the processor before the step of matching the diagnostic data with medical mining rules to obtain second candidate information ,Also includes:

    Generate the medical mining rule based on historical sample data:

    The historical sample data includes multiple pieces of historical medical data, and each piece of historical medical data includes the disease type of the historical patient and multiple medical entities;

    The medical mining rule includes a plurality of mining rules, each of the mining rules includes at least one of the medical entities, and each of the mining rules corresponds to a type of disease;

    Classify the historical medical data in the historical sample data according to the disease type to generate a disease type set;

    A frequent set mining algorithm is used to screen the historical medical data in the disease type set, and a mining rule corresponding to the disease type is generated.
13. The diagnostic data anomaly detection device according to claim 12, wherein the diagnostic data anomaly detection device is executed by the processor when the step of matching the diagnostic data with medical mining rules to obtain second candidate information ,include:

    Matching a plurality of the medical entities in the diagnostic data with each of the mining rules in the medical mining rules to obtain the matching degree of the disease type matching the diagnostic data;

    Extract the matching degrees of all disease types that match the diagnostic data, and generate the second candidate information.
14. The diagnostic data anomaly detection device according to claim 13, wherein the diagnostic data anomaly detection device is executed by the processor to fuse the first candidate information and the second candidate information to generate a third candidate The information steps include:

    According to the disease type in the first candidate information and the disease type in the second candidate information, the matching degree for the same disease type in the first candidate information and the corresponding degree in the second candidate information The matching degree calculates the matching average value, and generates the third candidate information including the disease type matching average value.
15. The diagnostic data abnormality detection device according to claim 13, wherein the fourth candidate information includes a matching value of a disease type;

    The diagnostic data abnormality detection device is executed by the processor to fuse the third candidate information and the fourth candidate information to obtain suspected disease information, and determine whether the diagnostic data is based on the suspected disease information The abnormal steps include:

    Calculating a suspicion value for the matching average value corresponding to the same disease type in the third candidate information and the corresponding matching value in the fourth candidate information;

    Extracting the type of disease whose suspected value meets the preset condition, and generating the suspected disease information;

    The target disease type in the diagnostic data is matched with the disease type in the suspected disease information. If it matches, it means that the diagnostic data is normal, and if it does not match, it means that the diagnostic data is abnormal.
16. A computer-readable storage medium having instructions stored on the computer-readable storage medium, wherein, when the instructions are executed by a processor, the steps of the diagnostic data abnormality detection method described below are implemented:

    Obtain the diagnostic data of the target patient;

    Matching the diagnostic data with preset medical rules to obtain first candidate information;

    Matching the diagnostic data with medical mining rules to obtain second candidate information;

    Fusing the first candidate information and the second candidate information to generate third candidate information;

    Recognizing the diagnosis data using a disease recognition model to obtain fourth candidate information;

    The third candidate information and the fourth candidate information are fused to obtain suspected disease information, and whether the diagnosis data is abnormal is determined according to the suspected disease information.
17. The computer-readable storage medium according to claim 16, wherein, when the instruction for detecting abnormality of the diagnostic data is executed by the processor, the step of acquiring the diagnostic data of the target patient comprises:

    Receiving medical data of the target patient sent by the user terminal, the medical data including: basic information of the target patient, the target disease type, and multiple medical entities;

    The medical entity in the medical data is extracted to generate the diagnostic data.
18. The computer-readable storage medium according to claim 17, wherein the preset medical rule is a rule set in advance based on medical knowledge, including a plurality of medical rules, and each medical rule includes at least one of the medical entities , Each of the medical rules corresponds to a type of disease;

    When the instruction for detecting abnormality of the diagnostic data is executed by the processor, the step of matching the diagnostic data with preset medical rules and obtaining first candidate information includes:

    Matching the multiple medical entities in the diagnostic data with each of the medical rules in the preset medical rules to obtain the matching degree of the disease type matching the diagnostic data;

    The matching degrees of all disease types matching the diagnosis data are extracted, and the first candidate information is generated.
19. 18. The computer-readable storage medium according to claim 18, wherein the instructions for detecting abnormality of the diagnostic data are executed by the processor, and the step of matching the diagnostic data with medical mining rules to obtain second candidate information Before, it also included:

    Generate the medical mining rule based on historical sample data:

    The historical sample data includes multiple pieces of historical medical data, and each piece of historical medical data includes the disease type of the historical patient and multiple medical entities;

    The medical mining rule includes a plurality of mining rules, each of the mining rules includes at least one of the medical entities, and each of the mining rules corresponds to a type of disease;

    Classify the historical medical data in the historical sample data according to the disease type to generate a disease type set;

    A frequent set mining algorithm is used to screen the historical medical data in the disease type set, and a mining rule corresponding to the disease type is generated.
20. The computer-readable storage medium according to claim 19, wherein the instructions for detecting abnormality of the diagnostic data are executed by the processor, and the step of matching the diagnostic data with medical mining rules to obtain second candidate information When, including:

    Matching a plurality of the medical entities in the diagnostic data with each of the mining rules in the medical mining rules to obtain the matching degree of the disease type matching the diagnostic data;

    Extract the matching degrees of all disease types that match the diagnostic data, and generate the second candidate information.

Images