CN113838573B - Clinical assistant decision-making diagnosis self-learning method, device, equipment and storage medium - Google Patents

Abstract

The application discloses a clinical auxiliary decision-making diagnosis self-learning method, device, equipment and medium, and relates to the technical field of clinical decision-making in the self-learning field and AI medical field. The specific implementation scheme is as follows: structuring the medical record information to be processed to obtain the entity information of the disease state of the medical record information to be processed; acquiring at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information and the physical information of the medical record information to be processed; acquiring the region ordering characteristics of a current region where a clinical assistant decision-making diagnosis recommendation system is deployed; and ranking at least one candidate diagnosis based on the region ranking characteristics of the current region and the recall probability of each candidate diagnosis, and performing diagnosis self-learning according to the ranking result. According to the method and the device, the diagnosis recommendation result is more consistent with the diagnosis characteristics of the current region, the problem of region differentiation of diagnosis recommendation is solved to a certain extent, and the accuracy of clinical assistant decision diagnosis is improved.

Description

Self-learning method, device, equipment and storage medium for clinical assistant decision-making diagnosis

technical field

The present application relates to the field of clinical decision-making technology in the field of self-learning and AI medical treatment, and in particular, to a self-learning method, device, device and storage medium for clinical auxiliary decision-making diagnosis.

Background technique

With the continuous integration of big data and medical care, the Clinical Decision Support System (CDSS) has gradually become an important means to improve the quality of medical care. CDSS is based on the content of the patient's electronic medical records, such as admission records, outpatient records, disease course records, inspection results, doctor's orders, surgical records, nursing records, etc., using advanced artificial intelligence technology to learn high-quality medical records marked by medical experts, the algorithm will automatically Recommend diseases that the current patient may suffer from, thereby assisting doctors in making clinical diagnosis decisions and reducing the probability of misdiagnosis by doctors.

SUMMARY OF THE INVENTION

The present application provides a self-learning method, device, device and storage medium for clinical aided decision-making diagnosis.

According to the first aspect of the present application, a self-learning method for clinical auxiliary decision-making diagnosis is provided, including:

Perform structured processing on the medical record information to be processed, and obtain the medical condition entity information of the medical record information to be processed;

Obtain at least one candidate diagnosis and a recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the medical condition entity information;

Obtain the regional ranking features of the current region where the clinical auxiliary decision-making diagnosis recommendation system is deployed;

Based on the region ranking feature of the current region and the recall probability of each candidate diagnosis, the at least one candidate diagnosis is ranked, and diagnosis self-learning is performed according to the ranking result.

According to a second aspect of the present application, there is provided a self-learning device for clinical auxiliary decision-making diagnosis, including:

a structured processing module, configured to perform structured processing on the medical record information to be processed, and obtain the medical condition entity information of the medical record information to be processed;

a recall module, configured to acquire at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the medical condition entity information;

an acquisition module, used to acquire the regional ranking feature of the current region where the clinical assistant decision-making diagnosis recommendation system is deployed;

The self-learning module is configured to sort the at least one candidate diagnosis based on the area sorting feature of the current area and the recall probability of each candidate diagnosis, and perform self-diagnosis learning according to the sorting result.

According to a third aspect of the present application, an electronic device is provided, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the clinical aided decision diagnosis of the aforementioned first aspect self-learning method.

According to a fourth aspect of the present application, there is provided a non-transitory computer-readable storage medium storing computer instructions, the computer instructions being used to cause the computer to execute the self-learning method for clinical aided decision-making and diagnosis described in the first aspect. .

According to a fifth aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the self-learning method for clinical aided decision-making and diagnosis according to the aforementioned first aspect.

According to the technical solution of the present application, the medical record information to be processed that reflects the diagnostic characteristics of the current region is structured and processed, the medical condition entity information of the medical record information to be processed is obtained, and the deep learning is performed based on the text information and the medical condition entity information of the medical record information to be processed. Modeling to obtain at least one candidate diagnosis and the recall probability of each candidate diagnosis; based on the regional ranking features of the current region and the recall probability of each candidate diagnosis, rank at least one candidate diagnosis to obtain a candidate diagnosis suitable for the current region. Sorting the results, and performing self-diagnosis learning according to the sorting results, making the diagnostic recommendation results more in line with the diagnostic characteristics of the current region, solving the regional differentiation problem of diagnostic recommendation to a certain extent, and improving the accuracy of clinical auxiliary decision-making diagnosis.

It should be understood that the content described in this section is not intended to identify key or critical features of the embodiments of the application, nor is it intended to limit the scope of the application. Other features of the present application will become readily understood from the following description.

Description of drawings

The accompanying drawings are used for better understanding of the present solution, and do not constitute a limitation to the present application. in:

Fig. 1 is the flow chart of a kind of clinical assistant decision-making diagnosis self-learning method provided according to the implementation of the present application;

Fig. 2 is a flow chart of obtaining the region ranking feature of the current region in which the clinical assistant decision-making diagnosis recommendation system is deployed according to the implementation of the present application;

3 is a flow chart of obtaining the co-occurrence probability of disease incidence, condition and diagnosis in medical records according to the implementation of the present application;

4 is a flowchart of sorting at least one candidate diagnosis according to the implementation of the present application, and performing self-learning of the diagnosis according to the sorting result;

5 is a flowchart of another self-learning method for clinical assistant decision-making diagnosis provided according to the implementation of the present application;

6 is a structural block diagram of a self-learning device for clinical auxiliary decision-making diagnosis provided according to the implementation of the present application;

7 is a structural block diagram of another self-learning device for clinical assistant decision-making diagnosis provided according to the implementation of the present application;

FIG. 8 is a block diagram of an electronic device for implementing a self-learning method for clinical aided decision-making diagnosis provided according to the implementation of the present application.

Detailed ways

Exemplary embodiments of the present application are described below with reference to the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and should be considered as exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted from the following description for clarity and conciseness.

With the continuous integration of big data and medical care, the Clinical Decision Support System (CDSS) has gradually become an important means to improve the quality of medical care. At present, most clinical auxiliary decision support systems build a general clinical auxiliary diagnosis recommendation system. However, due to regional differences in disease incidence and differences in doctors' diagnostic habits in different regions, the same medical records are expected in different regions. The recommended diagnosis may vary. For this reason, when landing in the actual area, the system needs to establish the regional self-learning capability of recommended diagnosis, so that the diagnosis recommended by the algorithm can conform to the diagnosis characteristics of the current area.

In the prior art, the technical solutions for self-learning of the diagnostic recommendation area mainly include two types: using a traditional machine learning model and using a complex machine learning model. Among them, for clinical auxiliary decision support systems using traditional machine learning models, the self-learning strategy directly updates the model recommendation results by counting the distribution of medical records in the current region; for clinical auxiliary decision-making using complex machine learning models (such as convolutional neural networks) Most of the support systems need to re-analyze the electronic medical record data in the corresponding area, such as chief complaint, current illness history, auxiliary examination, physical examination, past history, etc., and manually label the medical record data, and then retrain the model based on the labelled data. To achieve the purpose of self-learning in the diagnostic area.

However, for clinical aided decision support systems that use traditional machine learning models, such as decision trees, ideally, faster regional self-learning can be achieved, but traditional models have poor diagnostic results due to model complexity and other reasons. Second, for clinical auxiliary decision support systems using complex machine learning models (such as convolutional neural networks), it is necessary to re-label the data and then train and update the model, which is costly, has a long cycle, and lacks flexibility.

To this end, the present application provides a self-learning method, system, device and storage medium for clinical assistant decision-making diagnosis. Specifically, the self-learning method, system, device, and storage medium for clinical assistant decision-making diagnosis according to the embodiments of the present application are described below with reference to the accompanying drawings.

FIG. 1 is a flowchart of a self-learning method for clinical auxiliary decision-making diagnosis provided according to an embodiment of the present application. It should be noted that the self-learning method for clinical auxiliary decision-making diagnosis in the embodiment of the present application can be applied to the self-learning device for clinical auxiliary decision-making diagnosis in the embodiment of the present application, which can be configured on an electronic device.

As shown in FIG. 1, the self-learning method for clinical decision-making and diagnosis assistance may include the following steps:

Step 101: Perform structured processing on the medical record information to be processed, and obtain the medical condition entity information of the medical record information to be processed.

It should be noted that the information to be processed may include main complaint, history of present illness, auxiliary examination, physical examination, past history, and the like. Since the medical record information to be processed is basically in the form of plain text, in order to better analyze and understand the medical record information to be processed, the plain text in the medical record information to be processed can be entity parsed and extracted based on natural language processing technology to obtain the information to be processed. Condition entity information for processing medical record information. The entity information of the disease condition may include information such as disease name, symptoms, signs, test results, surgery, medication, and test results.

Step 102: Acquire at least one candidate diagnosis and a recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the medical condition entity information.

Optionally, in some embodiments of the present application, the candidate diagnoses and the recall probability of each candidate diagnosis may be obtained through deep learning modeling according to the text information and the condition entity information of the medical record information to be processed. As an example, deep learning modeling can be performed based on the text information of the medical record information to be processed and the entity information of the disease condition, and the modeled features can be fused to perform diagnosis and derivation, so as to return the possible diseases of the patient to the greatest possible extent. Among them, for the modeling of the text information of the medical record information to be processed, the text information of the medical record information to be processed can be divided into words, and each word can be encoded, and the structure of a convolutional neural network can be used to generate a word vector to Obtain text feature representation; for the modeling of medical condition entity information, a medical knowledge graph can be constructed through the co-occurrence relationship between medical condition entity information and diagnosis in a large number of medical records, and on the basis of the medical knowledge graph, an export corresponding to each medical record is generated The derived graph is composed of several sub-graphs, each sub-graph describing a collection of a type of condition of the patient for this visit. Then, based on the hierarchical attention mechanism, the subgraphs are encoded to obtain the condition entity feature representation of the medical record. Thus, based on the aforementioned text feature representation and condition entity information feature representation, at least one candidate diagnosis and the recall probability of each candidate diagnosis can be obtained, and the recall probability of each candidate diagnosis can be denoted as P _r .

Step 103: Obtain the region ranking feature of the current region where the clinical assistant decision-making diagnosis recommendation system is deployed.

Optionally, in some embodiments of the present application, the region ranking feature of the current region may be obtained from the region ranking feature library by determining the current region information for deploying the clinical aided decision diagnosis and recommendation system. For the specific implementation, reference may be made to the descriptions of the subsequent embodiments.

Step 104: Rank at least one candidate diagnosis based on the area ranking feature of the current area and the recall probability of each candidate diagnosis, and perform self-diagnosis learning according to the ranking result.

Optionally, in some embodiments of the present application, at least one candidate diagnosis may be ranked based on the regional ranking feature of the current area and the recall probability of each candidate diagnosis; a diagnosis recommendation result for the medical record information to be processed is obtained according to the ranking result. ; Obtain the real diagnosis results for the medical record information to be processed; carry out diagnosis self-learning according to the diagnosis recommendation results and the real diagnosis results. For the specific implementation, reference may be made to the descriptions of the subsequent embodiments.

According to the self-learning method for clinical auxiliary decision-making diagnosis according to the embodiment of the present application, the medical record information to be processed that reflects the diagnostic characteristics of the current region is structured and processed, the entity information of the medical condition of the medical record information to be processed is obtained, and the text information based on the medical record information to be processed is obtained. Perform deep learning modeling with disease entity information to obtain at least one candidate diagnosis and the recall probability of each candidate diagnosis; based on the regional ranking features of the current region and the recall probability of each candidate diagnosis, rank at least one candidate diagnosis and get Based on the ranking results of the candidate diagnoses in the current region, and carry out self-diagnosis learning according to the ranking results, the diagnostic recommendation results are more in line with the diagnostic characteristics of the current region, to a certain extent, solve the regional differentiation problem of diagnostic recommendations, and improve the accuracy of clinical auxiliary decision-making diagnosis. Spend.

It should be noted that, since the disease incidence and the co-occurrence probability of the disease and diagnosis in the medical records are important indicators of regional localization ranking, in some embodiments of the present application, the regional ranking features may include the disease incidence and the disease state in the medical records and the relationship between them. Co-occurrence probability of diagnosis. As an example, as shown in FIG. 2 , the implementation process of obtaining the region ranking feature of the current region in which the clinical assistant decision-making diagnosis recommendation system is deployed provided by the embodiment of the present application may include the following steps:

Step 201, obtaining medical records in the current area.

In step 202, self-learning is performed on the medical record log based on Bayesian statistics, so as to obtain the co-occurrence probability of disease incidence, disease condition and diagnosis in the medical record.

It should be noted that in different regions, the incidence probability of each disease may be different, so the disease incidence rate can be used as an important indicator for regional ranking, including incidence characteristics such as population and gender. In addition, the co-occurrence probability of the condition of the medical record and the diagnosis also reflects the regionalized characteristics of the diagnosis, such as symptoms, signs, and examinations. Therefore, the incidence of disease and the co-occurrence probability of condition and diagnosis are two important indicators in regional ranking. As an example, as shown in FIG. 3 , the implementation process of self-learning the medical record log based on Bayesian statistics provided by the embodiment of the present application to obtain the co-occurrence probability of disease incidence, disease condition and diagnosis in the medical record may include the following step:

Step 301 , perform structured and structured field statistics on the medical record log, and obtain medical condition entity information and structured field statistical information of the medical record log.

Optionally, in some embodiments of the present application, the CDSS central control may perform structured and structured field statistics on the primary CDSS medical record log to obtain the medical condition entity information and structured field statistical information of the medical record log. The entity information of the disease condition may include: diagnosis (standard name), past medical history, positive symptoms, abnormal signs, patient gender, patient age, pregnancy status, and other information.

Step 302 , based on the disease diagnosis basis relationship constructed by the medical knowledge graph and the disease condition entity information of the medical record log, screen the correlation between the diagnosis and the disease condition to obtain statistical relationship information between the diagnosis and the disease condition.

It should be noted that, in some embodiments of the present application, the condition may include information such as symptoms, abnormal signs, examinations, and tests, and the correlation between the diagnosis and the condition is screened, and the irrelevant association is filtered to obtain the diagnosis and the condition by statistics. relationship information.

Step 303 , based on Bayesian statistics, perform voting statistics on the statistical information of the structured field and the relationship information between the diagnosis and the condition, and obtain the co-occurrence probability of the disease incidence, the condition and the diagnosis in the medical record.

In the embodiment of the present application, the disease incidence rate in the medical record may be recorded as P ₁ , and the co-occurrence probability of the disease condition and diagnosis may be recorded as P ₂ .

Therefore, through the above steps 301 to 303, the medical record log can be self-learned based on Bayesian statistics, so as to continuously update the regional ranking feature of the co-occurrence probability of disease incidence, disease condition and diagnosis in the regional medical records learned in the self-learning stage In order to make the features in the regional sorting feature library more able to play a positive role in the sorting stage.

Step 203: Update the co-occurrence probability of disease incidence, disease condition and diagnosis in the medical record to the regional ranking feature database.

Step 204: Determine the current area information for deploying the clinical assistant decision-making diagnosis recommendation system.

Step 205 , according to the current region information, obtain the region ranking feature of the current region from the region ranking feature library.

In some embodiments of the present application, the ranking probability of each candidate diagnosis can be calculated based on the ranking probability calculation model, and self-learning of the diagnosis is performed according to the ranking result, and the parameters in the ranking probability calculation model are adjusted. As an example, as shown in FIG. 4 , based on the region ranking feature of the current region and the recall probability of each candidate diagnosis provided by the embodiment of the present application, at least one candidate diagnosis is ranked, and the diagnosis self-learning is performed according to the ranking result. The implementation process can include the following steps:

Step 401 , based on the regional ranking feature of the current region and the recall probability of each candidate diagnosis, use a ranking probability calculation model to calculate the ranking probability of each candidate diagnosis.

Optionally, in this embodiment of the present application, the following ranking probability calculation model may be used to calculate the ranking probability P _final of each candidate diagnosis:

P _final =σ _r P _r +σ ₁ P ₁ +σ ₁ P ₂

where P _r is the recall probability of the candidate diagnosis, σ _r is the weight of the recall probability P _r of the candidate diagnosis, P ₁ is the disease incidence rate of the candidate diagnosis, σ ₁ is the weight of the disease incidence rate P ₁ of the candidate diagnosis, P ₂ is the co-occurrence probability of the condition of the candidate diagnosis and the diagnosis, and σ ₂ is the weight of the co-occurrence probability P ₂ of the condition of the candidate diagnosis and the diagnosis.

Step 402: Rank at least one candidate diagnosis according to the ranking probability of each candidate diagnosis.

Step 403: Obtain a diagnosis recommendation result for the medical record information to be processed according to the sorting result.

Optionally, in some embodiments of the present application, the candidate diagnosis with the largest ranking probability P _final value in the ranking result may be used as the diagnosis recommendation result of the medical record information to be processed.

Step 404: Obtain the real diagnosis result for the medical record information to be processed.

As an example, in some embodiments of the present application, the doctor will refer to the diagnosis recommendation result obtained according to the ranking result, and make a correct diagnosis in combination with the actual condition and the examination result.

Step 405: Calculate the loss value according to the diagnosis recommendation result and the actual diagnosis result.

Step 406, according to the loss value, adjust the model parameters in the sorting probability calculation model.

That is to say, according to the loss value, adjust the weight σ _r of the recall probability P _r of the candidate diagnosis in the ranking probability calculation model, the weight σ ₁ of the disease incidence P ₁ of the candidate diagnosis, and the co-occurrence of the condition of the candidate diagnosis and the diagnosis The weight σ _{2 of the probability P 2 .} The self-learning of diagnosis based on the ranking results can improve the coincidence rate between the diagnosis recommendation results given by the ranking probability calculation model and the real diagnosis results given by experts. The coincidence rate means that the diagnostic recommendation results given by the ranking probability calculation model are consistent with the real diagnostic results given by experts, and they are aliases to each other or have a superordinate relationship with each other.

In this embodiment of the present application, it may be selected according to the actual situation of each region whether it is necessary to perform regional sorting on the candidate diagnosis. As an example, as shown in FIG. 5 , the self-learning method for clinical auxiliary decision-making diagnosis provided by the embodiment of the present application may include the following steps:

Step 501 , perform structured processing on the medical record information to be processed, and obtain the medical condition entity information of the medical record information to be processed.

Step 502: Acquire at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the medical condition entity information.

Step 503, select whether to perform regional self-learning. If regional self-learning is performed, go to step 504; if it is selected not to perform regional self-learning, go to step 505.

As an example, in some embodiments of the present application, a button may be set in the model application to select whether the region self-learning needs to be performed according to the actual situation of the region.

Step 504, determine whether the regional self-learning requirement is met. If the regional self-learning requirements are met, go to step 507; if the regional self-learning requirements are not met, go to step 505.

As an example, in some embodiments of the present application, the area self-learning requirement may be that the case samples reach a certain preset threshold. If the case sample is too small, regional self-learning cannot be performed based on historical cases.

Step 505: Rank at least one candidate diagnosis according to the recall probability of each candidate diagnosis.

Step 506: Obtain a diagnosis recommendation result for the medical record information to be processed according to the sorting result.

That is to say, if regional self-learning is not required or the conditions for regional self-learning are not met, sorting can be performed based on the recall probability, thereby obtaining a diagnosis recommendation result.

Step 507: Obtain the medical records in the current area.

Step 508 , perform structured and structured field statistics on the medical record log, and obtain the medical condition entity information and structured field statistical information of the medical record log.

Step 509 , based on the disease diagnosis basis relationship constructed by the medical knowledge graph and the disease condition entity information of the medical record log, screen the correlation between the diagnosis and the disease condition to obtain statistical relationship information between the diagnosis and the disease condition.

Step 510 , based on Bayesian statistics, perform voting statistics on the statistical information of the structured field and the relationship information between the diagnosis and the condition, and obtain the co-occurrence probability of the disease incidence, the condition and the diagnosis in the medical record.

Step 511: Update the co-occurrence probability of disease incidence, disease condition and diagnosis in the medical record to the regional ranking feature database.

Step 512: Determine the current area information for deploying the clinical assistant decision-making diagnosis recommendation system.

Step 513 , according to the current region information, obtain the region ranking feature of the current region from the region ranking feature library.

Step 514 , based on the regional ranking feature of the current region and the recall probability of each candidate diagnosis, use a ranking probability calculation model to calculate the ranking probability of each candidate diagnosis.

Step 515: Rank at least one candidate diagnosis according to the ranking probability of each candidate diagnosis.

Step 516: Obtain a diagnosis recommendation result for the medical record information to be processed according to the sorting result.

Step 517: Obtain the real diagnosis result for the medical record information to be processed.

Step 518: Calculate the loss value according to the diagnosis recommendation result and the actual diagnosis result.

Step 519, according to the loss value, adjust the model parameters in the sorting probability calculation model.

In the embodiments of the present application, steps 501 to 502 and steps 507 to 519 may be implemented in any of the embodiments of the present application, which are not specifically limited in the present application, and will not be described again.

According to the self-learning method for clinical auxiliary decision-making diagnosis according to the embodiment of the present application, the medical record information to be processed that reflects the diagnostic characteristics of the current region is structured and processed, the entity information of the medical condition of the medical record information to be processed is obtained, and the text information based on the medical record information to be processed is obtained. Perform deep learning modeling with disease entity information to obtain at least one candidate diagnosis and the recall probability of each candidate diagnosis; choose whether to perform regional self-learning according to the actual situation of the current area, if not, according to each candidate diagnosis The recall probability of at least one candidate diagnosis is sorted, and the diagnostic recommendation result of the medical record information to be processed is obtained according to the sorting result; if the regional self-learning is performed, it is judged whether the regional self-learning requirements are met, and if the regional self-learning requirements are not met, such as medical records If the number of samples is small, and regional self-learning cannot be performed, the ranking is based on the recall probability, and the diagnostic recommendation results of the medical record information to be processed are obtained; if the regional self-learning requirements are met, based on the regional sorting characteristics of the current region and the recall probability of each candidate diagnosis , using the sorting probability calculation model to calculate the sorting probability of each candidate diagnosis, and sorting at least one candidate diagnosis to obtain the sorting result of the candidate diagnosis suitable for the current area, and perform self-diagnosis learning according to the sorting result, and adjust the sorting probability calculation model The model parameters in the model make the diagnostic recommendation results more in line with the diagnostic characteristics of the current region, solve the problem of regional differentiation of diagnostic recommendations to a certain extent, and improve the accuracy of clinical auxiliary decision-making diagnosis.

FIG. 6 is a structural block diagram of a self-learning device for clinical decision-making assistance provided according to an embodiment of the present application. As shown in FIG. 6 , the self-learning device for clinical assistant decision-making and diagnosis may include a structured processing module 601 , a recall module 602 , an acquisition module 603 and a self-learning module 604 .

Specifically, the structured processing module 601 is configured to perform structured processing on the medical record information to be processed, and obtain the medical condition entity information of the medical record information to be processed.

The recall module 602 is configured to acquire at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information and the condition entity information of the medical record information to be processed.

The obtaining module 603 is configured to obtain the region ranking feature of the current region where the clinical assistant decision-making diagnosis recommendation system is deployed.

In some embodiments of the present application, the obtaining module 603 is specifically configured to: determine the current region information for deploying the clinical aided decision-making diagnosis recommendation system; and obtain the region ranking feature of the current region from the region ranking feature library according to the current region information.

The self-learning module 604 is configured to sort at least one candidate diagnosis based on the region sorting feature of the current region and the recall probability of each candidate diagnosis, and perform self-diagnosis learning according to the sorting result.

In some embodiments of the present application, the self-learning module 604 is specifically configured to: rank at least one candidate diagnosis based on the regional ranking feature of the current region and the recall probability of each candidate diagnosis; obtain information about the medical records to be processed according to the ranking result The diagnosis recommendation results are obtained; the real diagnosis results for the medical record information to be processed are obtained; the diagnosis self-learning is performed according to the diagnosis recommendation results and the real diagnosis results.

In some embodiments of the present application, the self-learning module 604 is further specifically configured to: calculate the ranking probability of each candidate diagnosis by adopting a ranking probability calculation model based on the regional ranking feature of the current region and the recall probability of each candidate diagnosis; The ranking probability of each candidate diagnosis, ranking at least one candidate diagnosis.

In some embodiments of the present application, the self-learning module 604 is further specifically configured to: calculate the loss value according to the diagnosis recommendation result and the real diagnosis result; and adjust the model parameters in the sorting probability calculation model according to the loss value.

Optionally, in some embodiments of the present application, the regional ranking feature includes the disease incidence rate, the co-occurrence probability of the disease condition and the diagnosis in the medical record, as shown in FIG. 705. Among them, the feature update module 705 is specifically used to: obtain the medical record log in the current area, and perform self-learning on the medical record log based on Bayesian statistics, so as to obtain the co-occurrence probability of disease incidence, medical condition and diagnosis in the medical record, and convert the medical record The co-occurrence probability of disease incidence, condition and diagnosis is updated to the regional ranking feature database.

In some embodiments of the present application, the feature update module 705 is specifically configured to: perform structured and structured field statistics on the medical record log, and obtain the medical condition entity information and structured field statistical information of the medical record log; The disease diagnosis is based on the relationship and the disease entity information of the medical record log, and the correlation between the diagnosis and the disease condition is screened to obtain the relationship information between the diagnosis and the disease state. Voting statistics were performed to obtain the co-occurrence probability of disease incidence, condition and diagnosis in the medical records.

Among them, 701-704 in FIG. 7 and 601-604 in FIG. 6 have the same function and structure.

Regarding the apparatus in the above-mentioned embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be described in detail here.

According to the self-learning device for clinical auxiliary decision-making and diagnosis according to the embodiment of the present application, the medical record information to be processed that reflects the diagnostic characteristics of the current area is structured and processed to obtain the medical condition entity information of the medical record information to be processed, and the text information based on the medical record information to be processed is obtained. Perform deep learning modeling with disease entity information to obtain at least one candidate diagnosis and the recall probability of each candidate diagnosis; choose whether to perform regional self-learning according to the actual situation of the current area, if not, according to each candidate diagnosis The recall probability of at least one candidate diagnosis is sorted, and the diagnostic recommendation result of the medical record information to be processed is obtained according to the sorting result; if the regional self-learning is performed, it is judged whether the regional self-learning requirements are met, and if the regional self-learning requirements are not met, such as medical records If the number of samples is small, and regional self-learning cannot be performed, the ranking is based on the recall probability, and the diagnostic recommendation results of the medical record information to be processed are obtained; if the regional self-learning requirements are met, based on the regional sorting characteristics of the current region and the recall probability of each candidate diagnosis , using the sorting probability calculation model to calculate the sorting probability of each candidate diagnosis, and sorting at least one candidate diagnosis to obtain the sorting result of the candidate diagnosis suitable for the current area, and perform self-diagnosis learning according to the sorting result, and adjust the sorting probability calculation model The model parameters in the model make the diagnostic recommendation results more in line with the diagnostic characteristics of the current region, solve the problem of regional differentiation of diagnostic recommendations to a certain extent, and improve the accuracy of clinical auxiliary decision-making diagnosis.

According to the embodiments of the present application, the present application further provides an electronic device, a readable storage medium, and a computer program product.

As shown in FIG. 8 , it is a block diagram of an electronic device for implementing a self-learning method for clinical decision-making assistance and diagnosis according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the application described and/or claimed herein.

As shown in FIG. 8, the electronic device includes: one or more processors 801, a memory 802, and interfaces for connecting various components, including a high-speed interface and a low-speed interface. The various components are interconnected using different buses and may be mounted on a common motherboard or otherwise as desired. The processor may process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of the GUI on an external input/output device, such as a display device coupled to the interface. In other embodiments, multiple processors and/or multiple buses may be used with multiple memories and multiple memories, if desired. Likewise, multiple electronic devices may be connected, each providing some of the necessary operations (eg, as a server array, a group of blade servers, or a multiprocessor system). In FIG. 8, a processor 801 is used as an example.

The memory 802 is the non-transitory computer-readable storage medium provided by the present application. Wherein, the memory stores instructions executable by at least one processor, so that the at least one processor executes the self-learning method for clinical assistant decision-making diagnosis provided by the present application. The non-transitory computer-readable storage medium of the present application stores computer instructions, and the computer instructions are used to cause the computer to execute the self-learning method for clinical assistant decision-making diagnosis provided by the present application.

As a non-transitory computer-readable storage medium, the memory 802 can be used to store non-transitory software programs, non-transitory computer-executable programs and modules, such as program instructions/modules corresponding to the self-learning method for clinical decision-making assistance in the embodiments of the present application (For example, the structured processing module 701, the recall module 702, the acquisition module 703, the self-learning module 704, and the feature update module 705 shown in FIG. 7). The processor 801 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions and modules stored in the memory 802 , that is, to implement the self-learning method for clinical aided decision-making diagnosis in the above method embodiments.

The memory 802 can include a stored program area and a stored data area, wherein the stored program area can store an operating system and an application program required by at least one function; Use the created data, etc. Additionally, memory 802 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 802 may optionally include a memory disposed remotely relative to the processor 801, and the remote memory may be connected to an electronic device for implementing the self-learning method for clinical decision-making assistance diagnosis through a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

The electronic device used to implement the self-learning method for clinical decision-aided diagnosis and diagnosis may further include: an input device 803 and an output device 804 . The processor 801 , the memory 802 , the input device 803 and the output device 804 may be connected by a bus or in other ways, and the connection by a bus is taken as an example in FIG. 8 .

The input device 803 can receive input numerical or character information, and generate key signal input related to user settings and function control of the electronic equipment used to implement the self-learning method for clinical aided decision-making diagnosis, such as touch screen, keypad, mouse, trackpad , touchpad, pointing stick, one or more mouse buttons, trackball, joystick and other input devices. Output devices 804 may include display devices, auxiliary lighting devices (eg, LEDs), haptic feedback devices (eg, vibration motors), and the like. The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some implementations, the display device may be a touch screen.

Various implementations of the systems and techniques described herein can be implemented in digital electronic circuitry, integrated circuit systems, application specific ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various implementations may include: being implemented in one or more computer programs, that is, the present application also proposes a computer program that, when executed by a processor, implements the clinical decision-aided diagnosis described in the above embodiments A self-learning approach, the one or more computer programs may be executed and/or interpreted on a programmable system comprising at least one programmable processor, which may be a special purpose or general purpose programmable processor, which may be retrieved from a memory system , at least one input device, and at least one output device receive data and instructions, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

These computational programs (also referred to as programs, software, software applications, or codes) include machine instructions for programmable processors, and may be implemented using high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages calculation program. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or apparatus for providing machine instructions and/or data to a programmable processor ( For example, magnetic disks, optical disks, memories, programmable logic devices (PLDs), including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or trackball) through which a user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); and can be in any form (including acoustic input, voice input, or tactile input) to receive input from the user.

The systems and techniques described herein may be implemented on a computing system that includes back-end components (eg, as a data server), or a computing system that includes middleware components (eg, an application server), or a computing system that includes front-end components (eg, a user computer having a graphical user interface or web browser through which a user may interact with implementations of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include: Local Area Networks (LANs), Wide Area Networks (WANs), the Internet, and blockchain networks.

A computer system can include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or a cloud host. It is a host product in the cloud computing service system to solve the traditional physical host and VPS service ("Virtual Private Server", or "VPS" for short). , there are the defects of difficult management and weak business expansion. The server can also be a server of a distributed system, or a server combined with a blockchain.

It should be understood that steps may be reordered, added or deleted using the various forms of flow shown above. For example, the steps described in the present application can be performed in parallel, sequentially or in different orders, and as long as the desired results of the technical solutions disclosed in the present application can be achieved, no limitation is imposed herein.

The above-mentioned specific embodiments do not constitute a limitation on the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application shall be included within the protection scope of this application.

Claims (17)

1. A clinical assistant decision making diagnostic self-learning method comprising:

structuring medical record information to be processed to obtain entity information of the medical conditions of the medical record information to be processed;

acquiring at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the entity information of the disease state;

acquiring a regional ranking characteristic of a current region where a clinical assistant decision-making diagnosis recommendation system is deployed, wherein the current region is a region where the clinical assistant decision-making diagnosis recommendation system is deployed, and the regional ranking characteristic comprises the co-occurrence probability of disease morbidity, state of illness and diagnosis in medical records;

and ranking the at least one candidate diagnosis based on the region ranking features of the current region and the recall probability of each candidate diagnosis, and performing diagnosis self-learning according to the ranking result.

2. The method of claim 1, wherein the obtaining a region ordering feature of a current region in which the clinical assistant decision diagnostic recommendation system is deployed comprises:

determining current region information for deploying a clinical assistant decision-making diagnosis recommendation system;

and acquiring the region sorting feature of the current region from a region sorting feature library according to the current region information.

3. The method of claim 2, wherein the method further comprises:

acquiring a medical record log in the current area;

self-learning the medical record log based on Bayesian statistics to obtain disease incidence and co-occurrence probability of the disease state and diagnosis in the medical record;

and updating the disease incidence in the medical record and the co-occurrence probability of the disease state and diagnosis to the regional sequencing feature library.

4. The method of claim 3, wherein the self-learning of the medical record log based on Bayesian statistics to derive a co-occurrence probability of disease incidence, the condition, and diagnosis in the medical record comprises:

carrying out structured and structured field statistics on the medical record log to obtain the entity information of the state of illness and the structured field statistical information of the medical record log;

screening the correlation between diagnosis and the disease state according to the disease diagnosis relationship constructed based on the medical knowledge map and the entity information of the disease state in the medical record log so as to obtain the correlation information between diagnosis and the disease state by statistics;

and voting and counting the structured field statistical information and the relationship information between the diagnosis and the disease state based on Bayesian statistics to obtain the disease incidence and the co-occurrence probability between the disease state and the diagnosis in the medical record.

5. The method of claim 1, wherein ranking the at least one candidate diagnosis based on the region ranking features of the current region and the recall probability of each of the candidate diagnoses and performing diagnostic self-learning according to the ranking results comprises:

ranking the at least one candidate diagnosis based on a region ranking feature of the current region and a recall probability of each of the candidate diagnoses;

acquiring a diagnosis recommendation result aiming at the medical record information to be processed according to the sorting result;

acquiring a real diagnosis result aiming at the medical record information to be processed;

and carrying out diagnosis self-learning according to the diagnosis recommendation result and the real diagnosis result.

6. The method of claim 5, wherein said ranking said at least one candidate diagnosis based on region-ranking features of said current region and recall probability of each said candidate diagnosis comprises:

calculating a ranking probability of each candidate diagnosis by adopting a ranking probability calculation model based on the region ranking characteristics of the current region and the recall probability of each candidate diagnosis;

ranking the at least one candidate diagnosis according to the ranking probability of each candidate diagnosis.

7. The method of claim 6, wherein said performing diagnostic self-learning based on said diagnostic recommendation and said true diagnostic result comprises:

calculating a loss value according to the diagnosis recommendation result and the real diagnosis result;

and adjusting model parameters in the sequencing probability calculation model according to the loss value.

8. A clinical assisted decision making diagnostic self-learning device comprising:

the structured processing module is used for carrying out structured processing on medical record information to be processed to obtain entity information of the medical conditions of the medical record information to be processed;

the recall module is used for acquiring at least one candidate diagnosis and the recall probability of each candidate diagnosis according to the text information of the medical record information to be processed and the entity information of the disease state;

the system comprises an acquisition module, a judgment module and a processing module, wherein the acquisition module is used for acquiring the region ranking characteristics of a current region where a clinical assistant decision-making diagnosis recommendation system is deployed, the current region is the region where the clinical assistant decision-making diagnosis recommendation system is deployed, and the ranking characteristics comprise the disease incidence, the state of illness and the co-occurrence probability of diagnosis in medical records;

and the self-learning module is used for sequencing the at least one candidate diagnosis based on the region sequencing characteristics of the current region and the recall probability of each candidate diagnosis and performing diagnosis self-learning according to the sequencing result.

9. The apparatus of claim 8, wherein the acquisition module is specifically configured to:

determining current region information for deploying a clinical assistant decision-making diagnosis recommendation system;

and acquiring the region sorting feature of the current region from a region sorting feature library according to the current region information.

10. The apparatus of claim 9, wherein the apparatus further comprises:

and the characteristic updating module is used for acquiring the medical record logs in the current region, self-learning the medical record logs based on Bayesian statistics to obtain the disease incidence in the medical record and the co-occurrence probability of the disease state and diagnosis, and updating the disease incidence in the medical record and the co-occurrence probability of the disease state and diagnosis to the regional ranking characteristic library.

11. The apparatus of claim 10, wherein the feature update module is specifically configured to:

carrying out structured and structured field statistics on the medical record log to obtain the entity information of the state of illness and the structured field statistical information of the medical record log;

screening the correlation between diagnosis and the disease state according to the disease diagnosis relationship constructed based on the medical knowledge map and the entity information of the disease state in the medical record log so as to obtain the correlation information between diagnosis and the disease state by statistics;

and voting and counting the structured field statistical information and the relationship information between the diagnosis and the disease state based on Bayesian statistics to obtain the disease incidence and the co-occurrence probability between the disease state and the diagnosis in the medical record.

12. The apparatus of claim 8, wherein the self-learning module is specifically configured to:

ranking the at least one candidate diagnosis based on a region ranking feature of the current region and a recall probability of each of the candidate diagnoses;

acquiring a diagnosis recommendation result aiming at the medical record information to be processed according to the sorting result;

acquiring a real diagnosis result aiming at the medical record information to be processed;

and performing diagnosis self-learning according to the diagnosis recommendation result and the real diagnosis result.

13. The apparatus of claim 12, wherein the self-learning module is specifically configured to:

calculating a ranking probability of each candidate diagnosis by adopting a ranking probability calculation model based on the region ranking characteristics of the current region and the recall probability of each candidate diagnosis;

ranking the at least one candidate diagnosis according to the ranking probability of each candidate diagnosis.

14. The apparatus according to claim 13, wherein the self-learning module is specifically configured to:

calculating a loss value according to the diagnosis recommendation result and the real diagnosis result;

and adjusting model parameters in the sequencing probability calculation model according to the loss value.

15. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

16. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-7.

17. A computer program product comprising a computer program which, when executed by a processor, carries out the steps of the method according to any one of claims 1-7.

Images