Method and device for pushing medical condition information

Abstract

The present invention discloses a method and device for pushing medical condition information, which relates to the field of data processing technology and is applicable to the medical and health field. The main purpose is to solve the problem of poor effectiveness of existing medical condition information push. The method comprises: obtaining medical case texts collected from a digital medical platform; determining a text classification model that matches the classification task of the medical case texts, and classifying the medical case texts based on the text classification model to obtain case text classification results, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task; counting the number of condition classifications in the case text classification results at preset time intervals, generating medical condition information based on the number of condition classifications, and pushing the medical condition information to the doctor client.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, applicable to the field of medical health, and in particular to a method and device for pushing medical condition information. Background Art

[0002] The rapid development of digital medical technology has greatly increased the convenience with which doctors can process medical records. In particular, digital medical platforms can classify the conditions of massive amounts of electronic medical records, allowing doctors to treat patients according to their severity.

[0003] Currently, the push of medical condition information typically involves directly extracting word features from text and classifying them, thereby pushing information based on the condition. However, classification based solely on word features is inflexible and cannot categorize the severity of different diseases. This requires the deployment of numerous feature extraction models, significantly reducing the efficiency of model deployment within digital healthcare platforms and, consequently, reducing the effectiveness of pushing medical condition information. Summary of the Invention

[0004] In view of this, the present invention provides a method and device for pushing medical condition information, the main purpose of which is to solve the problem of poor effectiveness of the existing push of medical condition information.

[0005] According to one aspect of the present invention, a method for pushing medical condition information is provided, comprising:

[0006] Obtain medical case texts collected from digital medical platforms;

[0007] Determining a text classification model that matches the classification task of the medical case text, and classifying the medical case text based on the text classification model to obtain a case text classification result, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task;

[0008] The number of disease classifications of the case text classification results is counted at preset time intervals, and medical disease information is generated according to the number of disease classifications and pushed to the doctor client.

[0009] Furthermore, before determining the text classification model that matches the classification task of the medical case text, the method further includes:

[0010] Constructing an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster;

[0011] Determining different classification tasks corresponding to different medical case texts, wherein the classification tasks are used to represent the task content of classifying the medical case texts according to at least one of disease type, medical department type, and medical service type;

[0012] The learning parameters in the polynomial activation function of each parameter adjuster are adjusted according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks.

[0013] Furthermore, the method further comprises:

[0014] Obtaining a case training sample set corresponding to the classification task;

[0015] Model training is performed on the text classification model to be trained based on the case training sample set, and when the dimensionality increase operation or dimensionality reduction operation determined based on the learning parameters matches a preset dimensionality range, the model training of the text classification model is completed.

[0016] Furthermore, the determining of a text classification model that matches the classification task of the medical case text includes:

[0017] Extracting word features from the medical case text, and searching for the classification task corresponding to the word features from a task feature database, wherein the task feature database stores at least one word feature corresponding to different classification tasks;

[0018] The text classification model that matches the classification task is retrieved from a classification model pool, where all text classification models that have completed model training are stored.

[0019] Furthermore, adjusting the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks includes:

[0020] Initializing learning parameters in the polynomial activation function, and configuring the polynomial activation function with the learning parameters in the parameter adjuster;

[0021] receiving a first learning adjustment coefficient corresponding to the classification task, and adjusting the learning parameter based on the first learning adjustment coefficient to obtain the text classification model to be trained; or

[0022] Based on a preset classification adjustment parameter mapping relationship, a second learning adjustment coefficient matching the classification task is determined, and the learning parameter is adjusted based on the second learning adjustment coefficient to obtain the text classification model to be trained. The preset classification adjustment parameter mapping relationship is pre-configured with a correspondence between different classification tasks and different learning adjustment coefficients.

[0023] Furthermore, generating medical condition information according to the number of condition classifications and pushing it to the doctor client includes:

[0024] If the number of the condition classifications is less than the first push threshold, the patient information is marked as a first-level push, and when the first push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client;

[0025] If the number of the condition classifications is greater than the first push threshold and less than the second push threshold, the patient information is marked for secondary push, and when the second push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client;

[0026] If the condition classification data is greater than the second push threshold, the patient information is marked for third-level push, and when the third push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client;

[0027] Among them, the first push condition, the second push condition, and the third push condition are all different.

[0028] Furthermore, the method further comprises:

[0029] Receive medical operations performed by the doctor client on the patient information, and count the number of operations of the medical operations within a preset time interval, wherein the medical operations include at least one of consultation, prescription, examination item push, and alarm;

[0030] If the number of operations is greater than a preset operation threshold, abnormal condition information is generated, and a condition warning is issued according to the abnormal condition information.

[0031] According to another aspect of the present invention, a device for pushing medical condition information is provided, comprising:

[0032] The acquisition module is used to obtain the medical case texts counted in the digital medical platform;

[0033] A classification module, configured to determine a text classification model that matches the classification task of the medical case text, and classify the medical case text based on the text classification model to obtain a case text classification result, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task;

[0034] The push module is used to count the number of disease classifications of the case text classification results according to a preset time interval, generate medical disease information according to the number of disease classifications, and push it to the doctor client.

[0035] Furthermore, the device further comprises:

[0036] A construction module is used to construct an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster;

[0037] a determination module for determining different classification tasks corresponding to different medical case texts, wherein the classification tasks are used to represent the task content of classifying the medical case texts according to at least one of disease type, medical department type, and medical service type;

[0038] An adjustment module is used to adjust the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks.

[0039] Furthermore, the device further comprises: a training module,

[0040] The acquisition module is further used to acquire a case training sample set corresponding to the classification task;

[0041] The training module is used to perform model training on the text classification model to be trained based on the case training sample set, and complete the model training of the text classification model when the dimensionality increase operation or dimensionality reduction operation determined based on the learning parameters matches the preset dimensionality range.

[0042] Furthermore,

[0043] The determination module is specifically used to extract word features from the medical case text and search for the classification task corresponding to the word features from a task feature database, where the task feature database stores at least one word feature corresponding to different classification tasks; and retrieve the text classification model that matches the classification task from a classification model pool, where the classification model pool stores all text classification models that have completed model training.

[0044] Furthermore,

[0045] The adjustment module is specifically further used to initialize the learning parameters in the polynomial activation function, and configure the polynomial activation function with the learning parameters in the parameter adjuster; receive a first learning adjustment coefficient corresponding to the classification task, and adjust the learning parameters based on the first learning adjustment coefficient to obtain the text classification model to be trained; or, determine a second learning adjustment coefficient matching the classification task based on a preset classification adjustment parameter mapping relationship, and adjust the learning parameters based on the second learning adjustment coefficient to obtain the text classification model to be trained, wherein the preset classification adjustment parameter mapping relationship is pre-configured with a correspondence between different classification tasks and different learning adjustment coefficients.

[0046] Furthermore,

[0047] The push module is specifically further used to, if the number of the disease classifications is less than a first push threshold, mark the patient information for first-level push, and when the first push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; if the number of the disease classifications is greater than the first push threshold and less than the second push threshold, mark the patient information for second-level push, and when the second push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; if the disease classification data is greater than the second push threshold, mark the patient information for third-level push, and when the third push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; wherein, the first push condition, the second push condition, and the third push condition are all different.

[0048] Furthermore, the device further comprises:

[0049] a receiving module, configured to receive medical operations performed by the doctor client on the patient information and to count the number of operations of the medical operations within a preset time interval, wherein the medical operations include at least one of: consultation, prescription, examination item push, and alarm;

[0050] The generating module is used to generate abnormal condition information if the number of operations is greater than a preset operation threshold, and to issue a condition warning according to the abnormal condition information.

[0051] According to another aspect of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for pushing medical condition information.

[0052] According to another aspect of the present invention, there is provided a computer device comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0053] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for pushing medical condition information.

[0054] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0055] The present invention provides a method and device for pushing medical condition information. Compared with the existing technology, the embodiment of the present invention obtains medical case texts counted in a digital medical platform; determines a text classification model that matches the classification task of the medical case text, and classifies the medical case text based on the text classification model to obtain a case text classification result. The text classification model is obtained by completing model training after determining a parameter adjuster connected to each model connection layer based on the classification task; counts the number of condition classifications of the case text classification results according to a preset time interval, generates medical condition information according to the number of condition classifications, and pushes it to a doctor client, thereby achieving the purpose of classifying different disease conditions, and improving the classification of disease characteristics based on an efficient model deployment method to determine different condition information, thereby improving the efficiency and accuracy of pushing medical condition information.

[0056] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0058] Figure 1 A flow chart of a method for pushing medical condition information provided by an embodiment of the present invention is shown;

[0059] Figure 2 A block diagram showing the composition of a device for pushing medical condition information provided by an embodiment of the present invention is shown;

[0060] Figure 3A schematic structural diagram of a computer device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0061] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0062] Embodiments of the present invention can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0063] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0064] Based on this, in one embodiment, an embodiment of the present invention provides a method for pushing medical condition information, which is illustrated by taking the application of this method to a computer device such as a server as an example, wherein the server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, such as smart medical systems, digital medical platforms, etc.

[0065] The embodiment of the present invention provides a method for pushing medical condition information, such as Figure 1 As shown, the method includes:

[0066] 101. Obtain the medical case text statistics from the digital medical platform.

[0067] In an embodiment of the present invention, the current execution end is a cloud server or terminal server of a digital medical platform, so as to execute the method for pushing medical condition information in the digital medical platform. Specifically, the current execution end can obtain the medical case texts entered or stored for all medical patients through the digital medical platform. The medical case texts are electronic texts of the cases entered by doctors based on the patient's medical situation, including but not limited to the patient's symptoms, the patient's examination information, the patient's treatment plan, etc., which are not specifically limited in this embodiment of the present invention.

[0068] It should be noted that, for the current scenario of condition analysis, the execution terminal can trigger the acquisition of medical case text statistics from the digital medical platform after a certain time interval to execute the condition push in steps 102-103. Furthermore, for the statistical analysis of medical case text, the case content entered by the doctor can be extracted from formatted files such as electronic case reports, and the entire case content can be used as the statistical medical case text.

[0069] 102. Determine a text classification model that matches the classification task of the medical case text, and classify the medical case text based on the text classification model to obtain a case text classification result.

[0070] In an embodiment of the present invention, the current execution end pre-trains a matching text classification model for different classification tasks, and classifies the medical case text based on the text classification model of the corresponding classification task to obtain a case text classification result. The text classification model is obtained by completing model training after determining the parameter adjuster connected to each model connection layer based on the classification task. That is, when the current execution end creates a text classification model that needs to be trained, it connects the matching parameter adjuster to each model connection layer, so as to pre-train the matching text classification model for different classification tasks, and adjust the parameters of the model through the parameter adjuster to improve the applicable scenarios of the model, thereby improving the processing efficiency of different classification tasks based on the text classification model.

[0071] It should be noted that, in an embodiment of the present invention, the case text classification result may include the intended case text classification label pre-marked in the training sample set as the result of model classification. In a specific case classification scenario, the case classification result includes at least one case classification, for example, epidemic disease a, epidemic disease b, or classification for liver disease or brain disease. The embodiment of the present invention does not make specific limitations to determine the number of disease classifications in step 103.

[0072] 103. Count the number of disease classifications of the case text classification results at preset time intervals, generate medical disease information according to the number of disease classifications, and push it to the doctor client.

[0073] In an embodiment of the present invention, in order to determine medical disease information based on the text classification results of the medical condition, the current execution end counts the number of disease classifications corresponding to each case text classification result according to a preset time interval to generate medical disease information and push it to the doctor client. The preset time interval can be a time set based on the needs of the medical condition analysis, such as 1 week, 10 days, etc. At the same time, the statistics of the number of disease classifications corresponding to each case text classification result can be accumulated within this preset time interval or can be reset and accumulated, which is not specifically limited in the embodiment of the present invention.

[0074] It should be noted that when the current execution end generates medical disease information according to the number of disease classifications, a quantity threshold is pre-set. If the quantity threshold is greater than this, it means that the disease is more serious. Therefore, medical disease information is generated, including but not limited to the number of patients with infectious disease a being x, the number of cases of disease c being y, etc., so as to be pushed to the doctor client for the doctor to process. The embodiment of the present invention does not make specific limitations.

[0075] In another embodiment of the present invention, for further definition and explanation, before the step of determining a text classification model that matches the classification task of the medical case text, the method further includes:

[0076] Constructing an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster;

[0077] Determine the different classification tasks corresponding to different medical case texts;

[0078] The learning parameters in the polynomial activation function of each parameter adjuster are adjusted according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks.

[0079] In order to classify medical case texts based on a text classification model, and thus to count the patient's illness, the current execution end first determines the classification tasks corresponding to different medical case texts to match the text classification model to be trained. The classification task is used to characterize the task content of classifying the medical case text according to at least one of the disease type, medical department type, and medical service type. When determining the classification tasks of different medical case texts, it can be determined based on the received task instructions, or it can be matched in sequence based on the preset classification task execution order, etc. The embodiment of the present invention does not make specific limitations. Furthermore, the current execution end adjusts the learning parameters in the polynomial activation function of the parameter adjuster according to the determined classification task, and has obtained a text classification model to be trained that is suitable for different classification tasks.

[0080] It should be noted that in the embodiment of the present invention, since matching parameter adjusters are connected to each model connection layer when creating a text classification model to be trained, the learning parameters in the polynomial activation function of each parameter adjuster are adjusted. Specifically, the polynomial activation function in each parameter adjuster adapter is expressed as: f(x) = a0 + b1x -1 +b2x -2 ...+b m x -m +a1x+a2x 2 ...+a n x n , where a0, b1, b2, ..., b m , a1, a2, ..., a n To learn parameters, they can be modified and updated in different classification tasks to suit different implementation scenarios, that is, different parameter adjusters and adapters, learning parameters a0, b1, b2, ..., b m , a1, a2, ..., a n Different model connection layers are connected to different parameter adjuster adapters. The current execution end can pre-configure corresponding learning parameters for different classification tasks, or can randomly generate parameter values ​​between 0 and 1 based on a random function as flexible learning parameters. This is not specifically limited in the present embodiment. In addition, m and n are the number of coefficients, which can be configured through initialization. Preferably, m = 4 and n = 4 are selected to improve model optimization efficiency.

[0081] In another embodiment of the present invention, for further definition and explanation, the steps further include:

[0082] Obtaining a case training sample set corresponding to the classification task;

[0083] Model training is performed on the text classification model to be trained based on the case training sample set, and when the dimensionality increase operation or dimensionality reduction operation determined based on the learning parameters matches a preset dimensionality range, the model training of the text classification model is completed.

[0084] To implement case classification based on text classification models, improve model deployment efficiency within digital healthcare platforms, and thus enhance the effectiveness of pushing medical information, the current execution end first obtains a case training sample set corresponding to the classification task and then trains the text classification model based on this training sample set. During model training, since the learning parameters have been updated, a dimensionality increase or reduction operation is calculated for each model connection layer during training to determine whether the data dimensions after the increase or reduction match the preset dimensionality range.

[0085] It should be noted that during the training process, the main parameters of each model connection layer remain unchanged, and the learning parameters in the adapter only need to be updated. For example, if the model connection layer is 12 transformer layers, the current execution end adds a parameter adjuster adapter to each transformer layer, and the parameter adjuster adapter is added after the linear connection layer of the transformer layer. At this time, the intermediate vector representation H of a sample x undergoes a dimensionality reduction operation W. down , and get H (1) , and then get H through the polynomial activation function (2) , and then through the dimension-raising operation W up , and get H (3) , that is, after the adapter adjustment, the sample representation is H+H (3) , and based on sample H+H (3) The dimension of the data is determined to determine whether it matches the preset dimension range. If it matches, the training of the intent classification model is completed.

[0086] In another embodiment of the present invention, for further definition and explanation, the step of determining a text classification model that matches the classification task of the medical case text includes:

[0087] Extracting word features from the medical case text, and searching for the classification task corresponding to the word features from a task feature database;

[0088] The text classification model that matches the classification task is retrieved from the classification model pool.

[0089] To ensure classification accuracy for different classification tasks and improve model deployment efficiency, the current execution end, when determining a text classification model that matches a classification task, first extracts word features from the medical case text to match the classification task based on a task feature database. The task feature database stores at least one word feature corresponding to each classification task. Then, based on a one-to-one comparison of the word features from the medical case text with the word features in the task feature database, a matching classification task is determined. This is not specifically limited in the present embodiment. Furthermore, the word features extracted from the medical case text can be extracted based on a natural language processing algorithm or a natural language dictionary. This is not specifically limited in the present embodiment. After determining the classification task, the current execution end retrieves a text classification model from a classification model pool that stores multiple trained text classification models. The classification model pool stores all trained text classification models. Since the text classification models are trained based on the case training sample set corresponding to the classification task, each text classification model stored in the classification model pool can be associated with a corresponding classification task, allowing for direct search and retrieval. This is not specifically limited in the present embodiment.

[0090] In another embodiment of the present invention, for further definition and explanation, the step of adjusting the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks includes:

[0091] Initializing learning parameters in the polynomial activation function, and configuring the polynomial activation function with the learning parameters in the parameter adjuster;

[0092] receiving a first learning adjustment coefficient corresponding to the classification task, and adjusting the learning parameter based on the first learning adjustment coefficient to obtain the text classification model to be trained; or

[0093] A second learning adjustment coefficient matching the classification task is determined based on a preset classification adjustment parameter mapping relationship, and the learning parameter is adjusted based on the second learning adjustment coefficient to obtain the text classification model to be trained.

[0094] In order to improve the deployment efficiency of the model and thus improve the push efficiency of medical condition information, when the current execution end matches the text classification model to be trained for different classification tasks, specifically, first initialize the learning parameters in the polynomial activation function, that is, a0, b1, b2, ..., b m , a1, a2, ..., a nPerform initialization configuration, and configure the polynomial activation function with learning parameters in the parameter adjuster to achieve the purpose of initializing and constructing all parameter adjusters. Furthermore, in an implementation scenario, the current execution end receives the first learning adjustment coefficient originally configured by the developer for the classification task, so that the current execution end adjusts the learning parameters based on the first learning adjustment coefficient to obtain the text classification model to be trained. In an implementation scenario, the current execution end can determine the second learning adjustment coefficient that matches the classification task based on the preset classification adjustment parameter mapping relationship, and adjust the learning parameters based on the second learning adjustment coefficient to obtain the text classification model to be trained, thereby achieving the purpose of automatically and efficiently adjusting the parameters of all parameter adjusters. At this time, the preset classification adjustment parameter mapping relationship is pre-configured with the correspondence between different classification tasks and different learning adjustment coefficients, so as to automatically and flexibly adjust the learning parameters and achieve the purpose of flexible model deployment.

[0095] In another embodiment of the present invention, for further definition and explanation, the step of generating medical condition information according to the number of condition classifications and pushing the information to the doctor client includes:

[0096] If the number of the condition classifications is less than the first push threshold, the patient information is marked as a first-level push, and when the first push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client;

[0097] If the number of the condition classifications is greater than the first push threshold and less than the second push threshold, the patient information is marked for secondary push, and when the second push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client;

[0098] If the condition classification data is greater than the second push threshold, the patient information will be marked as a third-level push, and when the third push condition is triggered, the medical condition information with the medical case text and patient information will be pushed to the doctor client.

[0099] In order to improve the effectiveness and accuracy of medical condition push notifications, the current execution end compares the number of medical condition classifications counted with the first push condition, the second push condition, and the third push condition when generating medical condition information. If the number of condition classifications is less than the first push threshold, the patient information is marked for first-level push, that is, the patient classified into one condition classification is marked for first-level push, so that when the first push condition is triggered, the medical condition information with medical case text and patient information is pushed to the doctor's client, so that the doctor can determine the severity of the condition according to the number of conditions and the first-level label, and determine whether to contact such patients. If the number of condition classifications is greater than the first push threshold and less than the second push threshold, the patient information is marked for second-level push, that is, the patient classified into one condition classification is marked for second-level push, and when the second push condition is triggered, the medical condition information with medical case text and patient information is pushed to the doctor's client, so that the doctor can determine the severity of the condition according to the number of conditions and the second-level label, and determine whether to contact such patients. If the condition classification data is greater than the second push threshold, the patient information will be marked for three-level push, that is, the patient classified into one condition classification will be marked for three-level push, and when the third push condition is triggered, the medical condition information with medical case text and patient information will be pushed to the doctor client, so that the doctor can determine the severity of the condition according to the number of conditions and the three-level label, and determine whether to contact such patients.

[0100] It should be noted that in the embodiments of the present invention, the first push threshold is less than the second push threshold, which is less than the third push threshold. These can be configured based on the condition classification requirements, which are not specifically limited in the embodiments of the present invention. Furthermore, the first push condition, the second push condition, and the third push condition are all different. Push conditions are used to represent the triggering conditions for pushing medical condition information, including but not limited to time conditions and data storage conditions, which are not specifically limited in the embodiments of the present invention.

[0101] In another embodiment of the present invention, for further definition and explanation, the steps further include:

[0102] receiving the medical operation performed by the doctor client on the patient information, and counting the number of operations of the medical operation within a preset time interval;

[0103] If the number of operations is greater than a preset operation threshold, abnormal condition information is generated, and a condition warning is issued according to the abnormal condition information.

[0104] To ensure the effectiveness of pushing medical condition information, the current execution end can also receive the medical operation behavior of the doctor client on the patient information, thereby counting the number of medical operation behaviors. Among them, the medical operation behavior includes at least one of the following: consultation behavior, prescription behavior, examination item push behavior, and alarm behavior. This can indicate that the doctor user has taken further measures after viewing the medical condition information. For example, after viewing the medical condition information, the doctor finds that there are too many patients with the same disease, so he or she works overtime to see patients, etc. This is not specifically limited in the embodiment of the present invention. When the doctor generates a medical operation behavior, the current execution end counts the number of medical operation behaviors, such as the number of consultations, at a preset time interval (such as 1 week, 10 days, etc.). This is not specifically limited in the embodiment of the present invention. When the number of operations exceeds a preset operation threshold, the current execution end automatically generates abnormal condition information and issues a condition warning based on the abnormal condition information to improve the efficiency of condition treatment. Among them, the preset operation threshold can be configured based on the doctor's needs for treating the condition, which is not specifically limited in the embodiment of the present invention.

[0105] An embodiment of the present invention provides a method for pushing medical condition information. Compared with the existing technology, the embodiment of the present invention obtains medical case texts counted in a digital medical platform; determines a text classification model that matches the classification task of the medical case text, and classifies the medical case text based on the text classification model to obtain a case text classification result. The text classification model is obtained by completing model training after determining a parameter adjuster connected to each model connection layer based on the classification task; counts the number of condition classifications of the case text classification results according to a preset time interval, generates medical condition information according to the number of condition classifications, and pushes it to the doctor client, thereby achieving the purpose of classifying different disease conditions, and improving the classification of disease characteristics based on an efficient model deployment method to determine different condition information and improve the efficiency and accuracy of pushing medical condition information.

[0106] Furthermore, as a response to the above Figure 1 The embodiment of the present invention provides a device for pushing medical condition information, such as Figure 2 As shown, the device includes:

[0107] An acquisition module 21 is used to acquire medical case texts collected in the digital medical platform;

[0108] A classification module 22 is configured to determine a text classification model that matches the classification task of the medical case text, and classify the medical case text based on the text classification model to obtain a case text classification result. The text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task;

[0109] The push module 23 is used to count the number of disease classifications of the case text classification results according to a preset time interval, generate medical disease information according to the number of disease classifications, and push it to the doctor client.

[0110] Furthermore, the device further comprises:

[0111] A construction module is used to construct an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster;

[0112] a determination module, configured to determine different classification tasks corresponding to different medical case texts, wherein the classification tasks are configured to represent the task content of classifying the medical case texts according to at least one of disease type, medical department type, and medical service type;

[0113] An adjustment module is used to adjust the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks.

[0114] Furthermore, the device further comprises: a training module,

[0115] The acquisition module is further used to acquire a case training sample set corresponding to the classification task;

[0116] The training module is used to perform model training on the text classification model to be trained based on the case training sample set, and complete the model training of the text classification model when the dimensionality increase operation or dimensionality reduction operation determined based on the learning parameters matches the preset dimensionality range.

[0117] Further,

[0118] The determination module is specifically used to extract word features from the medical case text and search for the classification task corresponding to the word features from a task feature database, where the task feature database stores at least one word feature corresponding to different classification tasks; and retrieve the text classification model that matches the classification task from a classification model pool, where the classification model pool stores all text classification models that have completed model training.

[0119] Further,

[0120] The adjustment module is specifically further used to initialize the learning parameters in the polynomial activation function, and configure the polynomial activation function with the learning parameters in the parameter adjuster; receive a first learning adjustment coefficient corresponding to the classification task, and adjust the learning parameters based on the first learning adjustment coefficient to obtain the text classification model to be trained; or, determine a second learning adjustment coefficient matching the classification task based on a preset classification adjustment parameter mapping relationship, and adjust the learning parameters based on the second learning adjustment coefficient to obtain the text classification model to be trained, wherein the preset classification adjustment parameter mapping relationship is pre-configured with a correspondence between different classification tasks and different learning adjustment coefficients.

[0121] Furthermore,

[0122] The push module is specifically further used to, if the number of the disease classifications is less than a first push threshold, mark the patient information for first-level push, and when the first push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; if the number of the disease classifications is greater than the first push threshold and less than the second push threshold, mark the patient information for second-level push, and when the second push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; if the disease classification data is greater than the second push threshold, mark the patient information for third-level push, and when the third push condition is triggered, push the medical disease information with the medical case text and patient information to the doctor client; wherein, the first push condition, the second push condition, and the third push condition are all different.

[0123] Furthermore, the device further comprises:

[0124] a receiving module, configured to receive medical operations performed by the doctor client on the patient information and to count the number of operations of the medical operations within a preset time interval, wherein the medical operations include at least one of: consultation, prescription, examination item push, and alarm;

[0125] The generating module is used to generate abnormal condition information if the number of operations is greater than a preset operation threshold, and to issue a condition warning according to the abnormal condition information.

[0126] An embodiment of the present invention provides a device for pushing medical condition information. Compared with the existing technology, the embodiment of the present invention obtains medical case texts counted in a digital medical platform; determines a text classification model that matches the classification task of the medical case text, and classifies the medical case text based on the text classification model to obtain a case text classification result. The text classification model is obtained by completing model training after determining a parameter adjuster connected to each model connection layer based on the classification task; counts the number of condition classifications of the case text classification results according to a preset time interval, generates medical condition information according to the number of condition classifications, and pushes it to a doctor client, thereby achieving the purpose of classifying different disease conditions, and improving the classification of disease characteristics based on an efficient model deployment method to determine different condition information, thereby improving the efficiency and accuracy of pushing medical condition information.

[0127] According to one embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the method for pushing medical condition information in any of the above method embodiments.

[0128] Figure 3 A schematic structural diagram of a computer device provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computer device.

[0129] like Figure 3 As shown, the computer device may include: a processor 302 , a communication interface 304 , a memory 306 , and a communication bus 308 .

[0130] The processor 302 , the communication interface 304 , and the memory 306 communicate with each other via a communication bus 308 .

[0131] The communication interface 304 is used to communicate with other devices such as clients or other servers.

[0132] The processor 302 is configured to execute the program 310 , and specifically to execute the relevant steps in the above-mentioned embodiment of the method for pushing medical condition information.

[0133] Specifically, the program 310 may include program codes, which include computer operation instructions.

[0134] Processor 302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in a computer device may be of the same type, such as one or more CPUs, or may be of different types, such as one or more CPUs and one or more ASICs.

[0135] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0136] The program 310 may be specifically configured to cause the processor 302 to perform the following operations:

[0137] Obtain medical case texts collected from digital medical platforms;

[0138] Determining a text classification model that matches the classification task of the medical case text, and classifying the medical case text based on the text classification model to obtain a case text classification result, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task;

[0139] The number of disease classifications of the case text classification results is counted at preset time intervals, and medical disease information is generated according to the number of disease classifications and pushed to the doctor client.

[0140] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing device, centralized on a single computing device, or distributed across a network of multiple computing devices. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than that shown, or can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0141] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for pushing medical condition information, characterized in that: include: Obtain medical case texts collected from digital medical platforms; Determining a text classification model that matches the classification task of the medical case text, and classifying the medical case text based on the text classification model to obtain a case text classification result, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task; Counting the number of disease classifications of the case text classification results at preset time intervals, generating medical disease information according to the number of disease classifications, and pushing the information to the doctor client; Before determining the text classification model that matches the classification task of the medical case text, the method further includes: Constructing an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster; Determining different classification tasks corresponding to different medical case texts, wherein the classification tasks are used to represent the task content of classifying the medical case texts according to at least one of disease type, medical department type, and medical service type; Adjusting the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks; The step of adjusting the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks to obtain a text classification model to be trained that matches different classification tasks includes: Initializing learning parameters in the polynomial activation function, and configuring the polynomial activation function with the learning parameters in the parameter adjuster; receiving a first learning adjustment coefficient corresponding to the classification task, and adjusting the learning parameter based on the first learning adjustment coefficient to obtain the text classification model to be trained; or Based on a preset classification adjustment parameter mapping relationship, a second learning adjustment coefficient matching the classification task is determined, and the learning parameter is adjusted based on the second learning adjustment coefficient to obtain the text classification model to be trained. The preset classification adjustment parameter mapping relationship is pre-configured with a correspondence between different classification tasks and different learning adjustment coefficients.

2. The method according to claim 1, characterized in that The method further comprises: Obtaining a case training sample set corresponding to the classification task; Model training is performed on the text classification model to be trained based on the case training sample set, and when the dimensionality increase operation or dimensionality reduction operation determined based on the learning parameters matches a preset dimensionality range, the model training of the text classification model is completed.

3. The method according to claim 1, characterized in that The determining of a text classification model that matches the classification task of the medical case text includes: Extracting word features from the medical case text, and searching for the classification task corresponding to the word features from a task feature database, wherein the task feature database stores at least one word feature corresponding to different classification tasks; The text classification model that matches the classification task is retrieved from a classification model pool, where all text classification models that have completed model training are stored.

4. The method according to claim 1, wherein Generating medical condition information according to the number of condition classifications and pushing it to the doctor client includes: If the number of the condition classifications is less than the first push threshold, the patient information is marked as a first-level push, and when the first push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client; If the number of the condition classifications is greater than the first push threshold and less than the second push threshold, the patient information is marked for secondary push, and when the second push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client; If the number of the condition classifications is greater than the second push threshold, the patient information is marked for third-level push, and when the third push condition is triggered, the medical condition information with the medical case text and patient information is pushed to the doctor client; Among them, the first push condition, the second push condition, and the third push condition are all different.

5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Receive medical operations performed by the doctor client on the patient information, and count the number of operations of the medical operations within a preset time interval, wherein the medical operations include at least one of: consultation, prescription, examination item push, and alarm; If the number of operations is greater than a preset operation threshold, abnormal condition information is generated, and a condition warning is issued according to the abnormal condition information.

6. A device for pushing medical condition information, characterized in that: include: The acquisition module is used to obtain the medical case texts counted in the digital medical platform; A classification module, configured to determine a text classification model that matches the classification task of the medical case text, and classify the medical case text based on the text classification model to obtain a case text classification result, wherein the text classification model is obtained by completing model training after determining parameter adjusters connected to each model connection layer based on the classification task; A push module is used to count the number of disease classifications of the case text classification results at preset time intervals, generate medical disease information according to the number of disease classifications, and push it to the doctor client; The device further comprises: A construction module is used to construct an initial text classification model, wherein each model connection layer in the initial text classification model is respectively connected to a parameter adjuster; a determination module for determining different classification tasks corresponding to different medical case texts, wherein the classification tasks are used to represent the task content of classifying the medical case texts according to at least one of disease type, medical department type, and medical service type; An adjustment module, configured to adjust the learning parameters in the polynomial activation function of each parameter adjuster according to different classification tasks, so as to obtain a text classification model to be trained that matches different classification tasks; The adjustment module is specifically further used to initialize the learning parameters in the polynomial activation function, and configure the polynomial activation function with the learning parameters in the parameter adjuster; receive a first learning adjustment coefficient corresponding to the classification task, and adjust the learning parameters based on the first learning adjustment coefficient to obtain the text classification model to be trained; or, determine a second learning adjustment coefficient matching the classification task based on a preset classification adjustment parameter mapping relationship, and adjust the learning parameters based on the second learning adjustment coefficient to obtain the text classification model to be trained, wherein the preset classification adjustment parameter mapping relationship is pre-configured with a correspondence between different classification tasks and different learning adjustment coefficients.

7. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute an operation corresponding to the method for pushing medical condition information according to any one of claims 1 to 5.

8. A computer device comprising: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute an operation corresponding to the method for pushing medical condition information according to any one of claims 1 to 5.

Images