Classification method, computing device and storage medium for information and medical diagnosis

Abstract

The embodiment of the present application provides a classification method, computing device and storage medium for information and medical diagnosis. In the embodiment of the present application, the case information of the patient to be classified is obtained; the case information is input into a preset classification model, and the first feature of the time information in the case information is extracted through the first sub-neural network model of the preset classification model; the other information in the case information and the first feature are extracted through the common layer of the preset classification model, and the DRGs are classified. Since time information is very important for determining the classification of case information, by strengthening the attention and emphasis on time information, the case information can be classified more accurately.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information classification method, a model structure, a model generation method, a medical diagnosis classification method, a computing device, and a storage medium. Background Art

[0002] DRGs (Diagnosis Related Groups) are one of the world's most advanced payment methods. Their guiding principle is to standardize the utilization of medical resources through the development of unified fixed-rate payment standards based on disease diagnosis classifications. They help incentivize hospitals to strengthen medical quality management, compel them to proactively reduce costs for profit, and reduce induced medical expenses, ultimately promoting rational drug use across society.

[0003] Currently, the promotion of DRGs in China is still in its early stages. Faced with hundreds or even thousands of disease groups defined in DRGs, hospitals often assign cases to the wrong groups, resulting in irrational use of medical resources. Summary of the Invention

[0004] Multiple aspects of the present application provide an information classification method, a model structure, a model generation method, a medical diagnosis classification method computing device and a storage medium, so as to enable more accurate classification.

[0005] An embodiment of the present application provides an information classification method, including: obtaining case information of a patient to be classified; inputting the case information into a preset classification model, and performing a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model; performing a second feature extraction on other information in the case information and the first feature through a common layer of the preset classification model, and classifying DRGs.

[0006] An embodiment of the present application also provides a model structure, including an input layer, a hidden layer, an output layer and a classification layer connected in sequence; a first sub-neural network model and a second sub-neural network model are connected between part of the input layer and the hidden layer.

[0007] An embodiment of the present application also provides a model generation method, including: obtaining historical case information of a patient, the historical case information having a corresponding DRGs classification result; inputting the historical case information into a preset initial classification model, and performing a first historical feature extraction on the historical time information in the historical case information through the first initial sub-neural network model of the preset initial classification model; performing a second historical feature extraction on other historical information in the historical case information and the first historical feature through the common layer of the preset initial classification model, and performing DRGs classification to obtain a historical classification result; based on the historical classification result and the DRGs classification result corresponding to the historical case information, the training of the preset initial classification model is completed.

[0008] An embodiment of the present application also provides a medical diagnosis classification method, including: obtaining case information of a patient to be classified; inputting the case information into a preset classification model, and performing a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model; performing a second feature extraction on the attribute information, diagnosis and treatment information and the first feature in the case information through the common layer of the preset classification model, and classifying DRGs.

[0009] An embodiment of the present application also provides a computing device, comprising: a memory, a processor, and a communication component; the memory is used to store computer programs; the communication component is used to obtain case information of patients to be classified; the processor is used to execute the computer program, so as to: input the case information into a preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model; perform a second feature extraction on other information in the case information and the first feature through the common layer of the preset classification model, and classify DRGs.

[0010] An embodiment of the present application also provides a computing device, a memory, a processor, and a communication component; the memory is used to store a computer program; the communication component is used to obtain historical case information of a patient, and the historical case information has a corresponding DRGs classification result; the processor is used to execute the computer program to: input the historical case information into a preset initial classification model, and perform a first historical feature extraction on the historical time information in the historical case information through the first initial sub-neural network model of the preset initial classification model; perform a second historical feature extraction on other historical information in the historical case information and the first historical feature through the common layer of the preset initial classification model, and perform DRGs classification to obtain a historical classification result; based on the historical classification result and the DRGs classification result corresponding to the historical case information, the training of the preset initial classification model is completed.

[0011] An embodiment of the present application also provides a computing device, comprising: a memory, a processor, and a communication component; the memory is used to store computer programs; the communication component is used to obtain case information of patients to be classified; the processor is used to execute the computer program, so as to: input the case information into a preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model; perform a second feature extraction on the attribute information, diagnosis and treatment information, and the first feature in the case information through the common layer of the preset classification model, and classify DRGs.

[0012] An embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by one or more processors, the one or more processors implement the steps in the above-mentioned information classification method and model generation method.

[0013] An embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by one or more processors, the one or more processors implement the steps in the above-mentioned medical diagnosis classification method.

[0014] In an embodiment of the present application, case information of a patient to be classified is obtained; the case information is input into a preset classification model, and a first feature extraction is performed on the time information in the case information through the first sub-neural network model of the preset classification model; a second feature extraction is performed on other information in the case information and the first feature through the common layer of the preset classification model, and DRGs are classified. Among them, the first feature extraction of the time information through the first sub-neural network model of the preset classification model can strengthen the attention and emphasis on the time information. Since time information is very important for determining the classification of case information, by strengthening the attention and emphasis on time information, the case information can be classified more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0016] Figure 1 A schematic diagram of the structure of an information classification system according to an exemplary embodiment of the present application;

[0017] Figure 2 A flowchart of a method for classifying information according to an exemplary embodiment of the present application is provided;

[0018] Figure 3 A schematic diagram of a model structure of an exemplary embodiment of the present application;

[0019] Figure 4 A schematic flow chart of a method for generating a model according to an exemplary embodiment of the present application;

[0020] Figure 5 A schematic diagram of the structure of an information classification device provided in an exemplary embodiment of the present application;

[0021] Figure 6 A schematic structural diagram of a device for generating a model provided by an exemplary embodiment of the present application;

[0022] Figure 7 A schematic diagram of the structure of a computing device provided for an exemplary embodiment of the present application;

[0023] Figure 8 A schematic diagram of the structure of a computing device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0025] As described above, in order to improve the accuracy of grouping, the embodiments provided in this application provide more accurate classification.

[0026] Figure 1 This is a schematic diagram of a structural classification system for information provided by an exemplary embodiment of the present application. Figure 1 As shown, the system 100 may include: a first device 101 , a second device 102 , and a third device 103 .

[0027] First device 101 refers to a device that can provide computing and processing services in a network virtual environment, and can also refer to a device that uses the network for model training. Physically, first device 101 can be any device that can provide computing services, respond to service requests, and perform model training, such as a cloud server, cloud host, virtual center, or conventional server. The first device 101 primarily comprises a processor, hard drive, memory, and system bus, similar to a general-purpose computer architecture.

[0028] Second device 102 refers to a device that can provide computing and processing services in a network virtual environment, and can be a device that utilizes the network for model training. Physically, second device 102 can be any device capable of providing computing services, responding to service requests, and performing model training, such as a cloud server, cloud host, virtual center, or conventional server. Second device 102 primarily comprises a processor, hard drive, memory, and system bus, similar to a general-purpose computer architecture.

[0029] It should be noted that the first device 101 and the second device 102 can be trained on different parts of the same model. There can be multiple first devices 101, and each first device 101 can be trained on part of the training data from the same source (e.g., the same hospital).

[0030] The third device 103 can be a device with certain computing capabilities that can send data to the first device 101 and receive data sent by the first device 101. The basic structure of the third device 103 may include: at least one processor. The number of processors may depend on the configuration and type of the device with certain computing capabilities. The device with certain computing capabilities may also include memory, which can be volatile, such as RAM, or non-volatile, such as read-only memory (ROM), flash memory, etc., or may include both types. The memory typically stores an operating system (OS), one or more application programs, and may also store program data, etc. In addition to the processing unit and memory, the device with certain computing capabilities also includes some basic configurations, such as a network card chip, an I / O bus, a display component, and some peripheral devices. Optionally, some peripheral devices may include, for example, a keyboard, a stylus, etc. Other peripheral devices are well known in the art and are not described in detail here. Optionally, the third device 103 can be a smart terminal, such as a mobile phone, a desktop computer, a laptop, a tablet computer, etc.

[0031] Specifically, the first device 101 can obtain the case information of the patient to be classified from the third device 103; input the case information into the preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model; perform a second feature extraction on other information in the case information and the first feature through the common layer of the preset classification model, and classify the DRGs.

[0032] Among them, case information includes time information related to the case, attribute information of the patient to be classified, and diagnosis and treatment information.

[0033] Specifically, the first device 101 extracts the second feature of the attribute information, diagnosis and treatment information and the first feature in the case information through the hidden layer and the output layer of the preset classification model; and classifies the DRGs based on the second feature and the classification layer in the preset classification model.

[0034] Specifically, the first device 101 inputs the second feature into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and performs classification through the classification function to obtain a classification result.

[0035] Specifically, the first device 101 inputs the second features into multiple second sub-neural network models in the preset classification model respectively to obtain corresponding output features, and classifies them through corresponding classification functions respectively to obtain the same classification results.

[0036] In addition, the case information also includes the hospital's identification information; wherein, the first device 101, after inputting the case information into the preset classification model, performs a third feature extraction on the identification information through the third sub-neural network model of the preset classification model; performs a fourth feature extraction on other information, the first feature and the third feature through the common layer of the preset classification model, and classifies DRGs.

[0037] Specifically, the first device 101 extracts the fourth feature of the attribute information, diagnosis and treatment information, the first feature and the third feature through the hidden layer and the output layer of the preset classification model; and classifies DRGs based on the fourth feature and the classification layer in the preset classification model.

[0038] In addition, the first device 101 can obtain historical case information of the patient and use the historical case information as training data for the preset initial classification model to train the preset initial classification model.

[0039] It should be noted that the training process in the first device 101 can be a partial training process, such as the gradient determination process. Of course, the first device 101 can also train the entire process. When the first device 101 trains the entire process, the training methods provided in the embodiments of the present application, such as those described below, can all be executed by the first device 101. In addition, if there are multiple first devices 101, the model can be trained through joint training.

[0040] Among them, the historical case information includes historical time information related to the case, historical attribute information of the patient and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; wherein, the first device 101 inputs the historical case information into the preset initial classification model, and extracts the first historical feature of the historical time information through the first initial sub-neural network model of the preset initial classification model; through the common layer of the preset initial classification model, the historical attribute information, historical diagnosis and treatment information and the first historical feature are extracted as the second historical feature, and DRGs are classified to obtain historical classification results; based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

[0041] Specifically, the first device 101 extracts the second historical features of the attribute information, medical information and the first historical features through the initial hidden layer and the initial output layer of the preset initial classification model; and classifies DRGs based on the second historical features and the initial classification layer in the preset initial classification model.

[0042] Specifically, the first device 101 inputs the second historical features into multiple second initial sub-neural network models in the preset initial classification model to obtain corresponding output historical features, and classifies them through corresponding initial classification functions to obtain multiple historical classification results.

[0043] In addition, the first device 101 completes the training of the preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the differences between the multiple historical classification results.

[0044] Specifically, the first device 101 can obtain a corresponding gradient based on multiple historical classification results and the DRGs classification results corresponding to the historical case information, or based on the historical classification results and the DRGs classification results corresponding to the historical case information, and send the gradient to the second device 102 in an encrypted manner. The second device 102 adjusts the model parameters based on the gradient and returns the adjusted model parameters to the first device 101, so that the first device 101 continues to train the model through iteration to determine the gradient and returns it to the second device 102 until the adjusted parameters meet the training criteria. The second device 102 can then send the model, i.e., the trained model, to the first device 101.

[0045] It should be noted that when there are multiple first devices 101, the second device 102 can receive multiple gradients, thereby obtaining multiple model parameters, merging the corresponding model parameters to obtain a corresponding model parameter, and then issuing the model parameter.

[0046] In addition, the initial model parameters may also be sent by the second device 102 to each first device 101 , so that each first device 101 can be classified.

[0047] The second device 102 may also adjust the model parameters in a safe environment.

[0048] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the first device 101, after inputting the historical case information into the preset initial classification model, performs a third historical feature extraction on the historical identification through the third initial sub-neural network model of the preset initial classification model; performs a fourth feature extraction on the attribute information, diagnosis and treatment information, the first feature and the third feature through the initial common layer of the preset initial classification model, and classifies the DRGs to obtain the historical classification results.

[0049] Among them, the preset initial classification model includes an input layer, a hidden layer, an output layer and a classification layer connected in sequence; a first initial sub-neural network model and a third initial sub-neural network model are connected between part of the input layer and the hidden layer.

[0050] The classification layer includes multiple second initial sub-neural network models and corresponding initial classification functions.

[0051] It should be noted that for the first device 101, other devices can also be added based on the system 100. These devices have the same implementation as the first device 101 and will not be described in detail. However, it should be understood that these other devices should belong to the same computer room as the first device 101, or serve the same institution, such as servers belonging to the same hospital. These other devices can be used to store the medical records of the first device 101 or deploy trained models, and can be flexibly changed according to needs. This will not be described in detail.

[0052] In the classification scenario of DRGs in the embodiment of the present application, users, such as doctors, can record the patient's medical history, that is, medical record information (which can be when the patient first starts to see a doctor or when the patient recovers) through a third device 103, such as a computer. The medical record information can be sent to the first device 101, such as a hospital server, through a computer. The hospital server can store the medical record information and then automatically input the medical record information into the trained preset classification model for classification. In addition, the server can also transmit the medical record information to other servers of the hospital through an interface for classification. Regardless of which server obtains the medical record information and inputs it into the corresponding model as the input layer information of the model. The medical record information can include time information related to the case (such as hospitalization time, discharge time, medical treatment time, etc.), attribute information of the patient to be classified (such as age, occupation, etc.), and diagnosis and treatment information (type of diagnosed disease, type of surgery, etc.). This information needs to be converted into data first. For example, word vectors are extracted through the FastText (FastText is a fast text classifier that provides a simple and efficient method for text classification and representation learning) word vector representation learning method. Then other information can be simply mapped. The time information (which can be mapped or not) first passes through the first sub-neural network model in the model, such as a shallow neural network, to extract the first feature of the time information, and then inputs the hidden layer and output layer of the model together with the other mapped information to obtain the second feature, and inputs the second feature into the second sub-neural network model in the model respectively. For example, if there are 4 shallow neural network models in the model, the corresponding output features are obtained, and classification is performed through a classification function, such as softmax, to obtain a classification result, that is, a classification result. The server returns the classification result to the computer so that the computer can directly display the classification result.

[0053] In addition, the medical record information may also include the identification information of the hospital (hospital name or hospital ID, etc.). The identification information may also first be extracted through the shallow neural network model in the model, i.e., the third sub-neural network model, to extract the third feature, and then input it together with the first feature and other mapped information into the hidden layer and output layer of the model to obtain the fourth feature, and the fourth feature is respectively input into the second sub-neural network model in the model. If there are 4 shallow neural network models in the model, the corresponding output features are obtained, and classification is performed through a classification function, such as softmax, to obtain a classification result. The server returns the classification result to the computer so that the computer can directly display the classification result.

[0054] Furthermore, the model training process can be as follows: After obtaining medical records (i.e., historical medical records), the second device 102, such as a hospital server or other server in the hospital, can manually submit the medical records offline, determine the classification of the medical records, and obtain a classification result and corresponding compensation. The obtained classification result can be input into the server, and the historical medical records in the server will now have the corresponding classification result. Based on this, model training is performed.

[0055] The third device 103, such as a cloud server, can send an initial model to the servers of multiple hospitals (which can be the second device 102), that is, preset the initial classification model and the initial parameters of the model. This allows the servers of each hospital to train the initial model through historical medical record information, obtain gradients, and upload the gradients to the cloud server in an encrypted manner. After receiving each gradient, the cloud server can adjust the initial parameters of the model, merge the adjusted parameters, obtain updated parameters, and then send them to the servers of each hospital. The gradients are then determined until the cloud server determines the updated parameters to complete the model training. The cloud server can then send the trained model to the servers of each hospital.

[0056] In the above embodiment, the first device 101, the second device 102, and the third device 103 are connected to a network, which may be a wireless connection. If the first device 101, the second device 102, and the third device 103 are connected for communication, the network standard of the mobile network may be any one of 2G (GSM), 2.5G (GPRS), 3G (WCDMA, TD-SCDMA, CDMA2000, UTMS), 4G (LTE), 4G+ (LTE+), WiMax, 5G, etc.

[0057] The information classification process is described in detail below in conjunction with the method embodiment.

[0058] Figure 2This is a flow chart of an information classification method according to an exemplary embodiment of the present application. The method 200 provided in the embodiment of the present application is executed by a computing device, such as a server. The method 200 includes the following steps:

[0059] 201: Obtain case information of the patient to be classified.

[0060] 202: Input the case information into the preset classification model, and extract the first feature of the time information in the case information through the first sub-neural network model of the preset classification model. 203: Use the common layer of the preset classification model to extract the second feature of the other information in the case information and the first feature, and then classify the DRGs.

[0061] The following is a detailed explanation of the above steps:

[0062] 201: Obtain case information of the patient to be classified.

[0063] Among them, case information includes time information related to the case, attribute information of the patient to be classified, and diagnosis and treatment information (i.e. other information). The time information can be hospitalization time, discharge time, consultation time, and cure time, etc. Attribute information can be the patient's age, occupation, height, weight, etc. Diagnosis and treatment information can be the type of diagnosed disease and the type of surgery, such as "main diagnosis name" and "surgery name", and can also be other diagnosis names, such as accompanying diseases, for example, diabetes (main diagnosis name) accompanied by fundus complications (other diagnosis names). In addition, the medical record information can also include department name, doctor number, etc.

[0064] For example, as described above, a doctor may see a patient and enter the patient's medical records into a hospital computer, which may serve as the patient's medical records. In response to the doctor's save or send operation, the computer may send this information to the hospital's server for storage.

[0065] During the patient's treatment, the doctor can continuously update the patient's medical records and update them to the server via computer.

[0066] It should be noted that after receiving the medical records, the server can perform classification. For example, classification can be performed when the medical records are first received, or after the patient's treatment is completed. Classification can also be performed based on a doctor's trigger instruction.

[0067] In addition, step 201 may also include: obtaining case information of the patient to be classified. The case information may include attribute information, diagnosis and treatment information, and hospital identification information of the patient to be classified. Classification is performed based on this, and will not be described in detail.

[0068] 202: Input the case information into a preset classification model, and perform a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model.

[0069] The preset classification model may be a neural network model, and the neural network model may be a DNN (Deep Neural Networks).

[0070] The first sub-neural network model can be a common neural network model, such as a shallow neural network model, or other types of neural network models, such as a CNN (Convolutional Neural Network) neural network model. However, in order to increase training speed and simplify training difficulty, a shallow neural network model can be used. The shallow neural network model includes an input layer, a hidden layer, and an output layer.

[0071] For example, as mentioned above, after obtaining the medical record information, the server can input the medical record information into the input layer of the model for classification. The time information can be first feature extracted and extracted through the shallow neural network model in the model. The shallow neural network model is connected to the input layer of the model to receive the time information of the input layer, and other information is not received.

[0072] It should be noted that since the final grouping of DRGs is formulated by the Medical Insurance Bureau, the standards for DRG grouping vary from place to place, and the grouping standards are updated over time. Therefore, the closer the time information, such as the "admission time", is to the current medical record information, the higher the training "weight" should be. Therefore, during the training process, the first sub-neural network model is added to obtain a higher training "weight". Accordingly, when performing classification, the attention and emphasis on time information can be strengthened based on this trained first sub-neural network model, which can more accurately classify case information.

[0073] Furthermore, for medical records, each piece of information can be mapped first. Based on pre-set mapping rules, the corresponding information is mapped to the corresponding data, meaning that the pre-set data represents the corresponding information. For example, if the patient attribute information is "female," "female" can be mapped to 0, and "male" can be mapped to 1. This mapping rule is pre-set. For time information, mapping is optional, as time information can be directly represented as numeric information.

[0074] It should be noted that the above mapping can be implemented directly by the server or by the model.

[0075] Medical information can be expressed differently in different hospitals. For example, the same "diagnosis" may be written differently. We can use methods such as FastText to transform medical information into high-dimensional data for effective neural network learning.

[0076] Specifically, the method 200 also includes: determining the vector features of the diagnosis and treatment information in other information; inputting the case information into the preset classification model, including: inputting the vector features into the preset classification model so that the second feature extraction of the vector features is performed through the common layer of the preset classification model.

[0077] The common layer may include the hidden layer and the output layer of the model.

[0078] For example, as described above, the model can use FastText to extract word vector features for "primary diagnosis name" and "surgery name." Then, based on these vector features and the model's hidden and output layers, the corresponding second features can be derived.

[0079] Alternatively, the server may first extract the vector features and then input them into the model.

[0080] In addition, the above-mentioned case information also includes the identification information of the hospital; wherein, the method 200 also includes: after the case information is input into the preset classification model, the third feature of the identification information is extracted through the third sub-neural network model of the preset classification model.

[0081] The identification information may be a hospital ID.

[0082] The third sub-neural network model can be a common neural network model, such as a shallow neural network model, or other types of neural network models, such as a CNN (Convolutional Neural Network) neural network model. However, in order to increase the training speed and simplify the training difficulty, a shallow neural network model can be used. The shallow neural network model includes an input layer, a hidden layer, and an output layer. This shallow neural network model is also partially connected to the input layer of the model, and is used to receive the hospital identification information in the input layer. Other information is not received.

[0083] It should be noted that the extraction process of the third feature is similar to the extraction process of the first feature described above, and will not be repeated here.

[0084] Furthermore, training data from different medical institutions is distributed across their servers, and these institutions can differ significantly. For example, large, comprehensive hospitals may cover a wide range of diseases, while data from specialized hospitals is more concentrated within a specific specialty. Therefore, a third neural network model is added to learn the weights associated with hospital distinctions. Accordingly, during classification, this trained third neural network model can be used to prioritize hospital identification information, enabling more accurate case classification.

[0085] In addition, for step 202, the case information may be input into a preset classification model, and the hospital identification information may be subjected to a third feature extraction through the second sub-neural network model of the preset classification model.

[0086] 203: Through the common layer of the preset classification model, the second feature is extracted from other information in the case information and the first feature, and the DRGs are classified.

[0087] Specifically, through the common layer of the preset classification model, the second feature is extracted from other information in the case information and the first feature, and DRGs are classified, including: through the hidden layer and output layer of the preset classification model, the second feature is extracted from the attribute information, diagnosis and treatment information and the first feature in the case information; and DRGs are classified based on the second feature and the classification layer in the preset classification model.

[0088] Among them, the classification of DRGs is performed based on the second feature and the classification layer in the preset classification model, including: inputting the second feature into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and classifying through the classification function to obtain the classification result.

[0089] The above classification layer can include an ordinary neural network model, that is, the second sub-neural network model, such as a shallow neural network model, or other types of neural network models, such as CNN (Convolutional Neural Networks) neural network model, etc., as well as classification functions, such as softmax. However, in order to increase the training speed and simplify the training difficulty, a shallow neural network model can be used. The shallow neural network model includes an input layer, a hidden layer, and an output layer. The shallow neural network model is also partially connected to the input layer of the model to receive the hospital identification information of the input layer.

[0090] The classification layer may include one or more groups of second sub-neural network models + classification functions. The second sub-neural network model is connected to the output layer of the model and receives the second feature output by the output layer.

[0091] For example, as described above, the server can use the model to input the mapped information corresponding to the attribute information, the mapped information corresponding to the diagnosis and treatment information, and the first feature into the hidden layer and output layer of the model to obtain the second feature. For a set of second sub-neural network models + classification functions, for example, the shallow neural network model receives the second feature output by the model output layer, extracts the corresponding output feature, and then performs softmax classification to obtain the corresponding classification result.

[0092] For multiple groups of second sub-neural network models + classification functions, the above classification process can be:

[0093] Specifically, DRGs are classified based on the second feature and the classification layer in the preset classification model, including: inputting the second feature into multiple second sub-neural network models in the preset classification model respectively to obtain corresponding output features, and classifying them respectively through corresponding classification functions to obtain the same classification results.

[0094] The classification process of multiple groups of second sub-neural network models + classification function is similar to the classification process of one group of second sub-neural network models + classification function described above, and will not be repeated here. It is only explained that for multiple groups (e.g., 4 groups), each second sub-neural network model is connected to the output layer of the model, that is, the second feature is input to each second sub-neural network model, and then the corresponding output feature is obtained by each second sub-neural network model and input into the corresponding classification function for classification, and the same classification result is obtained.

[0095] It should be noted that the network structures of the multiple second sub-neural network models can all be, for example, sub-shallow neural network models, but the specific structures can be different. For example, the structures of the hidden layers or output layers in the sub-shallow neural network models are different, such as the number of neurons therein, so that their model parameters are different, but the same classification results can be obtained through the classification function. In addition, it is also possible to train second sub-neural network models with the same structure but different model parameters.

[0096] For the hospital identification information mentioned above, the third feature is obtained through the third sub-neural network model. The third feature can also be added to the above basis for classification:

[0097] Specifically, the fourth feature is extracted from other information, the first feature and the third feature through the common layer of the preset classification model, and the DRGs are classified.

[0098] Since a similar classification process has been described previously, I will not repeat it here. Simply put, the hidden layer in the common layer receives the mapped information corresponding to the attribute information, the mapped information corresponding to the diagnosis and treatment information, the first feature, and the third feature. The fourth feature is obtained through the hidden layer and the output layer. Subsequently, classification can be performed using one or more sets of second sub-neural network models combined with a classification function.

[0099] Specifically, through the common layer of the preset classification model, the fourth feature is extracted from other information, the first feature and the third feature, and DRGs are classified, including: through the hidden layer and the output layer of the preset classification model, the fourth feature is extracted from the attribute information, the diagnosis and treatment information, the first feature and the third feature; and DRGs are classified based on the fourth feature and the classification layer in the preset classification model.

[0100] Since a similar process has been described in the previous article, it will not be repeated here.

[0101] Specifically, DRGs are classified based on the fourth feature and the classification layer in the preset classification model, including: inputting the fourth feature into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and classifying through the classification function to obtain the classification result.

[0102] Since a similar process has been described in the previous article, it will not be repeated here.

[0103] Specifically, DRGs are classified based on the fourth feature and the classification layer in the preset classification model, including: inputting the fourth feature into multiple second sub-neural network models in the preset classification model respectively to obtain corresponding output features, and classifying them respectively through corresponding classification functions to obtain the same classification results.

[0104] Since a similar process has been described in the previous article, it will not be repeated here.

[0105] In addition, for step 203, it is also possible to extract the fourth feature of the attribute information, diagnosis and treatment information (i.e., other information) and the third feature through the common layer of the preset classification model, and classify the DRGs. I will not go into details here. Based on this, for steps 201-203 that focus on the hospital's identification information, the subsequent other implementation steps can be adjusted based on the other implementation steps of this method 200, which will not be repeated here. The principle of adjustment is that the hospital's identification information can be swapped with the time information related to the case, and the technical features related to the two, such as the sub-neural network model, can also be swapped, which will not be repeated here.

[0106] After obtaining the classification results, the server can return the classification results to the computer for the doctor to view, so that the doctor can treat the patient reasonably based on the classification results. It can also provide analysis data for the doctor so that the doctor can conduct self-examination.

[0107] The training process of the above preset classification model is as follows:

[0108] Specifically, the method 200 further includes: obtaining historical case information of the patient; and using the historical case information as training data for a preset initial classification model to train the preset initial classification model.

[0109] Among them, historical medical record information refers to medical record information within a historical period, and the medical record information has the corresponding DRGs classification results determined by the Medical Insurance Bureau. Historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information. The historical case information has the corresponding DRGs classification results. It should be understood that historical medical record information is similar to the aforementioned medical record information, but historical medical record information belongs to medical record information within a historical period and has the corresponding DRGs classification results determined by the Medical Insurance Bureau. Therefore, the historical time information in the historical medical record information can be hospitalization time, discharge time, consultation time, and cure time, etc. Historical attribute information can be the patient's age, occupation, height, weight, etc. (it should be understood that the historical attribute information can change over time or remain unchanged over time). Historical diagnosis and treatment information can be the type of diagnosed disease and the type of surgery, such as "main diagnosis name" and "surgery name", and can also be other diagnosis names, such as accompanying symptoms, for example, diabetes (main diagnosis name) accompanied by fundus complications (other diagnosis names). In addition, the historical medical record information can also include department name, doctor number, etc.

[0110] The structure of the preset initial classification model is the same as that of the preset classification model, and the data processing process within the model is also the same or similar, so it will not be described here. Figure 3 As shown, the preset initial classification model includes an initial input layer 301, an initial hidden layer 304, an initial output layer 305, and an initial classification layer connected in sequence; a first initial sub-neural network model 303 and a third initial sub-neural network model 302 are connected between the partial input layer and the initial hidden layer 304. However, the model parameters of the preset initial classification model are not set, and are initial model parameters that need to be determined through training. The cloud server can send the preset initial classification model, including its initial model parameters, to the servers of each hospital so that the hospital servers can perform training and determine the gradient.

[0111] Among them, historical case information is used as training data for a preset initial classification model, and the preset initial classification model is trained, including: inputting historical case information into the preset initial classification model, and extracting the first historical feature of the historical time information through the first initial sub-neural network model of the preset initial classification model; extracting the second historical feature of the historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial common layer of the preset initial classification model, and classifying DRGs to obtain historical classification results; based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

[0112] The first initial sub-neural network model refers to the first sub-neural network model with initial parameters mentioned above. After the parameters are set, the first initial sub-neural network model becomes the first sub-neural network model, and no further details will be given. Figure 3 As shown, the preset initial classification model includes an initial input layer 301 and a first initial sub-neural network model 303 connected thereto. The first initial sub-neural network model 303 is used to receive time information (or mapped information) and extract the first historical feature, which will not be described in detail. After the data features corresponding to the historical time information (such as hospitalization time and discharge time, etc.), a first initial sub-neural network model, such as a shallow neural network model, is added to learn time-related weights to improve the accuracy of the model.

[0113] The process of extracting the second historical feature and classifying DRGs to obtain the historical classification results is the same as described above and will not be repeated here. However, during training, the model is determined during training, not after training. Furthermore, during training, the model can be updated as model parameters are updated until training is complete.

[0114] Among them, the training of the preset initial classification model is completed based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the difference between multiple historical classification results.

[0115] In addition, the server (which can be a server of multiple hospitals) determines the gradient through the gap between the historical classification results and the DRGs classification results corresponding to the historical case information. The gradient is then transmitted in an encrypted manner to the secure computing area (SGX) of the cloud server for decryption, and then the corresponding model parameters are determined by each gradient, and the model parameters are merged to obtain the updated model parameters. For the updated model parameters, the cloud server can determine whether the conditions for stopping training have been met, such as testing through test set data. If not, the cloud server can continue to send the updated model parameters to the servers of each hospital, continue training, and repeat the above process until the cloud server determines that the updated model parameters can complete the model training, and then send the final trained model with the final model parameters to the servers of each hospital for subsequent classification. Thereby, the preset initial classification model can be trained in a safe environment.

[0116] It should be noted that, in addition to being applicable to the field of DRGs, the embodiments of the present application can also be used in other application scenarios, such as in the scenario of financial institutions, where the data of different financial institutions can also be independently isolated with data security requirements; for example, in the government scenario, different government data are also independent, and in the e-commerce scenario, the data of different stores and manufacturers are independent, etc., and the above-mentioned isolation training model method can be used to perform model training according to different needs.

[0117] For multi-classification models, the neural network structure is used for classification, which can efficiently transmit gradients.

[0118] Specifically, through the initial common layer of the preset initial classification model, the second historical feature is extracted from the historical attribute information, historical diagnosis and treatment information and the first historical feature, and DRGs are classified, including: through the initial hidden layer and the initial output layer of the preset initial classification model, the second historical feature is extracted from the historical attribute information, historical diagnosis and treatment information and the first historical feature; and DRGs are classified based on the second historical feature and the initial classification layer in the preset initial classification model.

[0119] The initial hidden layer and the initial output layer of the preset initial classification model are the same as the hidden layer and the output layer of the preset classification model described above. For the initial hidden layer and the initial output layer, their model parameters are the initial parameters and are not described in detail. Figure 3 As shown, the output of the initial input layer 301 and the first initial sub-neural network model 302 is sent to the initial hidden layer 304, and the second historical feature extraction is performed through the output layer 305, which will not be repeated here.

[0120] The classification process of DRGs is similar to the classification process described above and will not be repeated here.

[0121] Specifically, DRGs are classified based on the second historical features and the initial classification layer in the preset initial classification model, including: inputting the second historical features into multiple second initial sub-neural network models in the preset initial classification model to obtain corresponding output historical features, and classifying them through corresponding initial classification functions to obtain multiple historical classification results.

[0122] Since a similar process has been described in the previous article, I will not repeat it here. I will only explain: Figure 3 As shown, the initial classification layer may include multiple second initial sub-neural network models 306 and corresponding initial classification functions Softmax 307. The server processes the second historical features through multiple second initial sub-neural network models 306 (the number may be 4, such as a shallow neural network model) in the preset initial classification model to obtain corresponding output historical features, and classifies them respectively through the corresponding initial classification functions Softmax 307 to obtain multiple historical classification results.

[0123] In the embodiment of the present application, multiple random "second initial sub-neural network models" can be integrated to further enhance the generalization ability of the model. In principle, the data input into the second initial sub-neural network model should obtain the same result regardless of which "second initial sub-neural network model" is used for the final classification result. In fact, since the four "second initial sub-neural network models" use different network models (the specific structures of the hidden layer or output layer in the second initial sub-neural network model may be different, such as the number of neurons therein, etc., or the second initial sub-neural network model with the same structure but different model parameters may be trained), the scores corresponding to the classification results will be different, but the classification results are the same, just like a classification result. The back propagation of the neural network with the preset initial classification model will minimize the difference as much as possible, thereby achieving the effect of preventing overfitting.

[0124] In addition, the second historical features can also be input into the second initial sub-neural network model (which can be a second initial sub-neural network model) in the preset initial classification model to obtain corresponding output historical features, and then classified using the corresponding initial classification function to obtain multiple historical classification results. This will not be repeated here.

[0125] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the method 200 also includes: after inputting the historical case information into the preset initial classification model, the third historical feature extraction of the historical identification is performed through the third initial sub-neural network model of the preset initial classification model; through the initial common layer of the preset initial classification model, the fourth feature extraction of the historical attribute information, historical diagnosis and treatment information, the first feature and the third feature is performed, and DRGs are classified to obtain the historical classification results.

[0126] The historical identification information is similar to the identification information mentioned above, and the historical identification information refers to the identification information within a historical period.

[0127] Since a similar process has been described in the previous article, I will not repeat it here. I will only explain: Figure 3 As shown, the server inputs the information mapped from the historical identification information into the third initial sub-neural network model 302 through the initial input layer of the preset initial classification model to perform third historical feature extraction. The fourth feature extraction is performed on the attribute information (mapped information), diagnosis and treatment information (mapped information), the first feature, and the third feature through the initial hidden layer 304 and output layer 305 of the preset initial classification model, and the DRGs are classified to obtain the historical classification results, which will not be repeated here.

[0128] Adding a sub-neural network model, such as a shallow neural network, after the data features of the historical identification information (e.g., the information after the mapping of the historical identification information) to learn the relevant weights for hospital distinction can improve the accuracy of the model.

[0129] Based on the similar inventive concepts described above, another exemplary embodiment of the present application provides a model structure. The structure provided in the embodiment of the present application can be deployed in a server, and includes an input layer, a hidden layer, an output layer, and a classification layer connected in sequence; a first sub-neural network model and a second sub-neural network model are connected between part of the input layer and the hidden layer.

[0130] Since the structure of this model has been explained in detail above, it will not be repeated here. It is only explained that the second sub-neural network model is the third sub-neural network model (or the third initial sub-neural network model) described above. The input layer, hidden layer, output layer, and classification layer can also be the initial input layer, initial hidden layer, initial output layer, and initial classification layer described above. The first sub-neural network model can also be the first initial sub-neural network model described above.

[0131] Specifically, the classification layer includes multiple third sub-neural network models and corresponding initial classification functions. Since the structure of the model has been explained in detail above, it will not be repeated here. It is only explained that the third sub-neural network model is the second sub-neural network model (or the second initial sub-neural network model) described above. It should be understood that the above model structure can be expressed as the structure of the initial model or the structure of the trained model.

[0132] In addition, the patient's historical case information is obtained; the historical case information is used as training data for the initial model to train the initial model.

[0133] Since this has been explained in detail in the previous article, I will not repeat it here.

[0134] Specifically, historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; wherein, the historical case information is used as the training data of the initial model to train the initial model, including: inputting the historical case information into the initial input layer, and performing a first historical feature extraction on the historical time information through the initial first sub-neural network model; performing a second historical feature extraction on the historical attribute information, historical diagnosis and treatment information, and the first historical feature through the initial common layer, and performing DRGs classification to obtain a historical classification result; based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the initial model is completed.

[0135] Since this has been explained in detail above, it will not be repeated here. It is only explained that the initial input layer, the initial first sub-neural network, and the initial common layer can be the initial input layer, the first initial sub-neural network model, and the initial common layer described above.

[0136] Specifically, through the initial common layer, the second historical feature is extracted from the historical attribute information, historical diagnosis and treatment information and the first historical feature, and DRGs are classified, including: through the initial hidden layer and the initial output layer, the second historical feature is extracted from the attribute information, diagnosis and treatment information and the first historical feature; and DRGs are classified based on the second historical feature and the initial classification layer.

[0137] Since this has been explained in detail above, it will not be repeated here. It is only explained that the initial hidden layer, the initial output layer, and the initial classification layer can be the initial hidden layer, the initial output layer, and the initial classification layer described above.

[0138] Specifically, DRGs are classified based on the second historical features and the initial classification layer, including: inputting the second historical features into multiple initial third sub-neural network models respectively to obtain corresponding output historical features, and classifying them respectively through the corresponding initial classification functions to obtain multiple historical classification results.

[0139] Since it has been explained in detail above, it will not be repeated here. It is only explained that the initial third sub-neural network model can be the second initial sub-neural network model mentioned above.

[0140] In addition, the initial model training is completed based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the differences between multiple historical classification results.

[0141] Since this has been explained in detail in the previous article, I will not repeat it here.

[0142] In addition, historical case information also includes the historical identification information of the hospital; after the historical case information is input into the initial model, the third historical feature of the historical identification information is extracted through the initial second sub-neural network model; through the initial common layer, the fourth feature of the attribute information, diagnosis and treatment information, the first feature and the third feature is extracted, and the DRGs are classified to obtain the historical classification results.

[0143] Since it has been explained in the previous article, it will not be repeated here. It will only be explained that the initial second sub-neural network model can be the second initial sub-neural network model described in the previous article.

[0144] In addition, for the contents not described in detail in this structure, reference may also be made to the contents in the above method 200.

[0145] Based on the above similar inventive concept, Figure 4 A flow chart of a method for generating a model provided by another exemplary embodiment of the present application is shown. The method 400 provided in the embodiment of the present application is executed by a server and includes the following steps:

[0146] 401: Get the patient's historical medical information.

[0147] Among them, historical case information has corresponding DRGs classification results.

[0148] 402: Input the historical case information into a preset initial classification model, and extract a first historical feature of the historical time information in the historical case information using a first initial sub-neural network model of the preset initial classification model.

[0149] 403: By presetting the common layer of the initial classification model, extracting the second historical feature from other historical information and the first historical feature in the historical case information, and classifying the DRGs to obtain a historical classification result.

[0150] 404: Based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

[0151] Since the specific implementation of steps 401-404 has been described in detail above, they will not be repeated here.

[0152] Among them, historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information.

[0153] Specifically, through the common layer of the preset initial classification model, the second historical feature is extracted from other historical information in the historical case information and the first historical feature, and DRGs are classified, including: through the initial hidden layer and the initial output layer of the preset initial classification model, the second historical feature is extracted from the historical attribute information, historical diagnosis and treatment information and the first historical feature; and DRGs are classified based on the second historical feature and the initial classification layer in the preset initial classification model.

[0154] Specifically, DRGs are classified based on the second historical features and the initial classification layer in the preset initial classification model, including: inputting the second historical features into multiple second initial sub-neural network models in the preset initial classification model to obtain corresponding output historical features, and classifying them through corresponding initial classification functions to obtain multiple historical classification results.

[0155] In addition, the method 400 also includes: completing the training of the preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the difference between the multiple historical classification results.

[0156] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the method 400 also includes: after inputting the historical case information into the preset initial classification model, the third historical feature extraction is performed on the historical identification information through the third initial sub-neural network model of the preset initial classification model; through the initial common layer of the preset initial classification model, the fourth feature extraction is performed on other historical information (such as attribute information, diagnosis and treatment information), the first feature and the third feature, and DRGs are classified to obtain historical classification results.

[0157] In addition, for the contents not described in detail in the present method 400 , reference may also be made to the steps in the above-mentioned method 200 .

[0158] Figure 5 This is a schematic diagram of the structural framework of an information classification device provided in an exemplary embodiment of the present application. The device 500 can be applied to a server. The device 500 includes: an acquisition module 501, an extraction module 502, and a classification module 503. The functions of each module are described in detail below:

[0159] The acquisition module 501 is used to acquire case information of the patient to be classified.

[0160] The extraction module 502 is used to input the case information into the preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model.

[0161] The classification module 503 is used to extract the second feature from other information in the case information and the first feature through the common layer of the preset classification model, and classify the DRGs.

[0162] Among them, case information includes time information related to the case, attribute information of the patient to be classified, and diagnosis and treatment information.

[0163] Specifically, the classification module 503 includes: a first extraction unit, used to extract the second feature of the attribute information, diagnosis and treatment information and the first feature in the case information through the hidden layer and output layer of the preset classification model; a first classification unit, used to classify DRGs based on the second feature and the classification layer in the preset classification model.

[0164] Specifically, the first classification unit is used to: input the second feature into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and classify through the classification function to obtain the classification result.

[0165] Specifically, the first classification unit is used to: input the second features into multiple second sub-neural network models in the preset classification model respectively to obtain corresponding output features, and classify them respectively through corresponding classification functions to obtain the same classification results.

[0166] In addition, the case information also includes the hospital's identification information; wherein, the extraction module 502 is also used to: after the case information is input into the preset classification model, the third feature of the identification information is extracted through the third sub-neural network model of the preset classification model; through the common layer of the preset classification model, the fourth feature of other information, the first feature and the third feature is extracted, and DRGs are classified.

[0167] Specifically, the extraction module 502 includes: a second extraction unit for performing a fourth feature extraction on the attribute information, the diagnosis and treatment information, the first feature, and the third feature through the hidden layer and the output layer of the preset classification model; a second classification unit for classifying DRGs based on the fourth feature and the classification layer in the preset classification model. In addition, the device 500 also includes: a determination module for determining the vector features of the diagnosis and treatment information in other information; and the extraction module 502 for inputting the vector features into the preset classification model so that the second feature extraction is performed on the vector features through the common layer of the preset classification model.

[0168] In addition, the acquisition module 501 is also used to obtain the patient's historical case information; the device 500 also includes a training module for using the historical case information as training data for the preset initial classification model to train the preset initial classification model.

[0169] In addition, the historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; wherein, the training module includes: a third extraction unit, used to input the historical case information into the preset initial classification model, and extract the first historical feature of the historical time information through the first initial sub-neural network model of the preset initial classification model; the third extraction unit, used to extract the second historical feature of the historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial common layer of the preset initial classification model, and classify DRGs to obtain historical classification results; the training unit, used to complete the training of the preset initial classification model based on the historical classification results and the DRGs classification results corresponding to the historical case information.

[0170] Specifically, the third extraction unit is used to extract the second historical feature of the historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial hidden layer and the initial output layer of the preset initial classification model; and classify DRGs based on the second historical feature and the initial classification layer in the preset initial classification model.

[0171] Specifically, the third extraction unit is used to input the second historical features into multiple second initial sub-neural network models in the preset initial classification model respectively to obtain corresponding output historical features, and classify them through corresponding initial classification functions respectively to obtain multiple historical classification results.

[0172] In addition, the device 500 also includes: a training module for completing the training of a preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the difference between multiple historical classification results.

[0173] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the extraction module 502 is also used to: after inputting the historical case information into the preset initial classification model, perform a third historical feature extraction on the historical identification through the third initial sub-neural network model of the preset initial classification model; the classification module 503 is also used to perform a fourth feature extraction on the historical attribute information, historical diagnosis and treatment information, the first feature and the third feature through the initial common layer of the preset initial classification model, and perform DRGs classification to obtain a historical classification result.

[0174] Among them, the preset initial classification model includes an initial input layer, an initial hidden layer, an initial output layer and an initial classification layer connected in sequence; a first initial sub-neural network model and a third initial sub-neural network model are connected between part of the initial input layer and the initial hidden layer.

[0175] The initial classification layer includes multiple second initial sub-neural network models and corresponding initial classification functions.

[0176] In addition, the training module is also used to train the preset initial classification model in a safe environment.

[0177] It should be noted that the device can be transformed according to different schemes of the method 200 described above. For example, the technical scheme of steps 201-203 focusing on the identification information of the hospital will not be described in detail here.

[0178] Figure 6 A schematic diagram of the structural framework of a model generation device provided by another exemplary embodiment of the present application is shown. The device 600 can be applied to a server. The device 600 includes: an acquisition module 601, an extraction module 602, a classification module 603, and a training module 604. The functions of each module are described in detail below:

[0179] The acquisition module 601 is used to acquire the patient's historical case information.

[0180] The historical case information has corresponding DRGs classification results. The extraction module 602 is used to input the historical case information into a preset initial classification model, and perform a first historical feature extraction on the historical time information in the historical case information through the first initial sub-neural network model of the preset initial classification model.

[0181] The classification module 603 is used to extract the second historical feature from other historical information and the first historical feature in the historical case information through the common layer of the preset initial classification model, and classify the DRGs to obtain the historical classification result.

[0182] The training module 604 is used to complete the training of the preset initial classification model based on the historical classification results and the DRGs classification results corresponding to the historical case information.

[0183] Among them, historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information. The historical case information has corresponding DRGs classification results.

[0184] Specifically, the classification module 603 includes: an extraction unit, used to extract the second historical feature of the historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial hidden layer and the initial output layer of the preset initial classification model; a classification unit, used to classify DRGs based on the second historical feature and the initial classification layer in the preset initial classification model.

[0185] Specifically, the classification unit is used to: input the second historical features into multiple second initial sub-neural network models in the preset initial classification model respectively, obtain corresponding output historical features, and classify them respectively through the corresponding initial classification functions to obtain multiple historical classification results.

[0186] In addition, the training module 604 is used to complete the training of the preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the differences between the multiple historical classification results.

[0187] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the extraction module 602 is also used to: after the historical case information is input into the preset initial classification model, the historical identification information is subjected to a third historical feature extraction through the third initial sub-neural network model of the preset initial classification model; the classification module 603 is also used to perform a fourth feature extraction on other historical information, the first feature and the third feature through the initial common layer of the preset initial classification model, and classify the DRGs to obtain a historical classification result.

[0188] It should be noted that for some contents not mentioned in the apparatus 600 , reference may be made to the contents of the above-mentioned apparatus 500 .

[0189] The above describes Figure 5 The internal functions and structure of the device 500 shown, in one possible design, Figure 5 The structure of the apparatus 500 shown can be implemented as a computing device, such as a server. Figure 7 As shown, the device 700 may include: a memory 701, a processor 702, and a communication component 703;

[0190] The memory 701 is used to store computer programs.

[0191] Processor 702 is used to execute a computer program to: input case information into a preset classification model, and extract a first feature of the time information in the case information through the first sub-neural network model of the preset classification model; extract a second feature of other information in the case information and the first feature through the common layer of the preset classification model, and classify DRGs.

[0192] The communication component 703 is used to obtain case information of the patient to be classified.

[0193] Specifically, the processor 702 is specifically used to: extract the second feature of the attribute information, diagnosis and treatment information and the first feature in the case information through the hidden layer and output layer of the preset classification model; the first classification unit is used to classify DRGs based on the second feature and the classification layer in the preset classification model.

[0194] Among them, case information includes time information related to the case, attribute information of the patient to be classified, and diagnosis and treatment information.

[0195] Specifically, the processor 702 is specifically used to: input the second feature into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and classify through the classification function to obtain the classification result.

[0196] Specifically, the processor 702 is specifically used to: input the second features into multiple second sub-neural network models in the preset classification model respectively to obtain corresponding output features, and classify them respectively through corresponding classification functions to obtain the same classification results.

[0197] In addition, the case information also includes the hospital's identification information; wherein, the processor 702 is also used to: after inputting the case information into the preset classification model, perform a third feature extraction on the identification information through the third sub-neural network model of the preset classification model; perform a fourth feature extraction on other information, the first feature and the third feature through the common layer of the preset classification model, and classify DRGs.

[0198] Specifically, processor 702 is specifically used to: extract the fourth feature of attribute information, diagnosis and treatment information, the first feature and the third feature through the hidden layer and output layer of the preset classification model; and classify DRGs based on the fourth feature and the classification layer in the preset classification model.

[0199] In addition, the processor 702 is further used to: determine the vector features of the diagnosis and treatment information in other information; input the vector features into the preset classification model so that the second feature extraction is performed on the vector features through the common layer of the preset classification model.

[0200] In addition, the processor 702 is further configured to: obtain historical case information of the patient; and use the historical case information as training data for a preset initial classification model to train the preset initial classification model.

[0201] In addition, the historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; wherein, the processor 702 is specifically used to: input the historical case information into the preset initial classification model, and extract the first historical feature of the historical time information through the first initial sub-neural network model of the preset initial classification model; extract the second historical feature of the historical attribute information, historical diagnosis and treatment information, and the first historical feature through the initial common layer of the preset initial classification model, and classify DRGs to obtain historical classification results; based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

[0202] Specifically, processor 702 is specifically used to: extract the second historical feature of historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial hidden layer and the initial output layer of the preset initial classification model; classify DRGs based on the second historical feature and the initial classification layer in the preset initial classification model.

[0203] Specifically, the processor 702 is specifically used to: input the second historical features into multiple second initial sub-neural network models in the preset initial classification model respectively, obtain corresponding output historical features, and classify them respectively through the corresponding initial classification functions to obtain multiple historical classification results.

[0204] In addition, the processor 702 is also used to complete the training of the preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the difference between multiple historical classification results.

[0205] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the processor 702 is also used to: after inputting the historical case information into the preset initial classification model, perform a third historical feature extraction on the historical identification through the third initial sub-neural network model of the preset initial classification model; perform a fourth feature extraction on the historical attribute information, historical diagnosis and treatment information, the first feature and the third feature through the initial common layer of the preset initial classification model, and perform DRGs classification to obtain a historical classification result.

[0206] Among them, the preset initial classification model includes an initial input layer, an initial hidden layer, an initial output layer and an initial classification layer connected in sequence; a first initial sub-neural network model and a third initial sub-neural network model are connected between part of the initial input layer and the initial hidden layer.

[0207] The initial classification layer includes multiple second initial sub-neural network models and corresponding initial classification functions.

[0208] In addition, the processor 702 is further configured to: train a preset initial classification model in a safe environment.

[0209] In addition, an embodiment of the present invention provides a computer storage medium, which, when a computer program is executed by one or more processors, causes the one or more processors to implement Figure 1-Figure 3 The present invention provides steps of an information classification method in a method embodiment.

[0210] It should be noted that the device can also be transformed according to different schemes of method 200 described above, such as the technical scheme for steps 201-203 focusing on the hospital's identification information, which will not be described in detail here.

[0211] The above describes Figure 6 The internal functions and structure of the device 600 shown, in one possible design, Figure 6 The structure of the apparatus 600 shown can be implemented as a computing device, such as a server. Figure 8 As shown, the device 800 may include: a memory 801, a processor 802, and a communication component 803;

[0212] The memory 801 is used to store computer programs.

[0213] The communication component 803 is used to obtain the patient's historical case information, which has corresponding DRGs classification results.

[0214] Processor 802 is used to execute the computer program to: input historical case information into a preset initial classification model, and extract a first historical feature of the historical time information in the historical case information through the first initial sub-neural network model of the preset initial classification model; extract a second historical feature of other historical information and the first historical feature in the historical case information through the common layer of the preset initial classification model, and classify DRGs to obtain a historical classification result; based on the historical classification result and the DRGs classification result corresponding to the historical case information, complete the training of the preset initial classification model.

[0215] Among them, historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information.

[0216] Specifically, processor 802 is specifically used to: extract the second historical feature of historical attribute information, historical diagnosis and treatment information and the first historical feature through the initial hidden layer and the initial output layer of the preset initial classification model; and classify DRGs based on the second historical feature and the initial classification layer in the preset initial classification model.

[0217] Specifically, the processor 802 is specifically used to: input the second historical features into multiple second initial sub-neural network models in the preset initial classification model, obtain corresponding output historical features, and classify them through corresponding initial classification functions to obtain multiple historical classification results.

[0218] In addition, the processor 802 is also used to complete the training of the preset initial classification model based on multiple historical classification results and DRGs classification results corresponding to historical case information, as well as the difference between the multiple historical classification results.

[0219] In addition, the historical case information also includes the historical identification information of the hospital; wherein, the processor 802 is also used to: after inputting the historical case information into the preset initial classification model, perform a third historical feature extraction on the historical identification information through the third initial sub-neural network model of the preset initial classification model; perform a fourth feature extraction on other historical information, the first feature and the third feature through the initial common layer of the preset initial classification model, and classify the DRGs to obtain a historical classification result.

[0220] It should be noted that for some contents not mentioned in the device 800, reference may be made to the contents of the above-mentioned device 700.

[0221] In addition, an embodiment of the present invention provides a computer storage medium, which, when a computer program is executed by one or more processors, causes the one or more processors to implement Figure 4 The present invention provides steps of a method for generating a model in a method embodiment.

[0222] Based on the similar inventive concept described above, another exemplary embodiment of the present application provides a medical diagnosis classification method. The method 900 provided in the embodiment of the present application is executed by a server and includes the following steps:

[0223] 901: Obtain case information of the patient to be classified.

[0224] 902: Input the case information into a preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model.

[0225] 903: Through the common layer of the preset classification model, the second feature is extracted from the attribute information, diagnosis and treatment information and the first feature in the case information, and the DRGs are classified.

[0226] Since the specific implementation of steps 901-903 has been described in detail above, they will not be repeated here.

[0227] It should be noted that for some contents not mentioned in the method 900 , reference may be made to the contents of the above-mentioned method 200 .

[0228] Another exemplary embodiment of the present application provides a medical diagnosis classification device. The device 1000 can be applied to a server. The device 1000 includes: an acquisition module 1001, an extraction module 1002, and a classification module 1003. The functions of each module are described in detail below:

[0229] Acquisition module 1001: Acquisition of case information of the patient to be classified.

[0230] Extraction module 1002: Input the case information into the preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model.

[0231] Classification module 1003: Through the common layer of the preset classification model, the second feature is extracted from the attribute information, diagnosis and treatment information and the first feature in the case information, and the DRGs are classified.

[0232] It should be noted that for some contents not mentioned in the apparatus 1000 , reference may be made to the contents of the above-mentioned apparatus 500 .

[0233] The above describes the internal functions and structure of the device 1000. In one possible design, the structure of the device 1000 can be implemented as a computing device, such as a server. The device 1100 may include: a memory 1101, a processor 1102, and a communication component 1103;

[0234] The memory 1101 is used to store computer programs.

[0235] The communication component 1103 is used to obtain case information of the patient to be classified.

[0236] Processor 1102 is used to execute the computer program to: input case information into a preset classification model, and perform a first feature extraction on the time information in the case information through the first sub-neural network model of the preset classification model; perform a second feature extraction on the attribute information, diagnosis and treatment information and the first feature in the case information through the common layer of the preset classification model, and classify DRGs.

[0237] It should be noted that for some contents not mentioned in the device 1100, reference may be made to the contents of the above-mentioned device 700.

[0238] In addition, an embodiment of the present invention provides a computer storage medium, which, when a computer program is executed by one or more processors, causes the one or more processors to implement the steps of a medical diagnosis classification method in the embodiment of method 900.

[0239] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The sequence numbers of the operations, such as 201, 202, 203, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0240] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0241] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by adding a necessary general hardware platform, and of course can also be implemented by a combination of hardware and software. Based on this understanding, the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a computer product. The present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0242] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable multimedia data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable multimedia data processing device generate instructions for implementing the process in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0243] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable multimedia data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0244] These computer program instructions can also be loaded onto a computer or other programmable multimedia data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0245] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0246] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0247] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0248] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for classifying information, characterized in that: include: Obtain case information of patients to be classified; Inputting the case information into a preset classification model, and performing a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model; Performing a second feature extraction on other information in the case information and the first feature through the common layer of the preset classification model, and classifying DRGs, including: performing a second feature extraction on attribute information, diagnosis and treatment information in the case information and the first feature through the hidden layer and the output layer of the preset classification model; DRGs are classified based on the second feature and the classification layer in the preset classification model.

2. The method according to claim 1, characterized in that The classifying of DRGs based on the second feature and the classification layer in the preset classification model includes: The second feature is input into the second sub-neural network model in the preset classification model to obtain the corresponding output feature, and classification is performed through the classification function to obtain the classification result.

3. The method according to claim 1, characterized in that The classifying of DRGs based on the second feature and the classification layer in the preset classification model includes: The second features are respectively input into multiple second sub-neural network models in the preset classification model to obtain corresponding output features, and are respectively classified by corresponding classification functions to obtain the same classification results.

4. The method according to claim 1, wherein The case information also includes hospital identification information; The method further comprises: After inputting the case information into a preset classification model, performing a third feature extraction on the identification information through a third sub-neural network model of the preset classification model; Through the common layer of the preset classification model, the fourth feature is extracted from the other information, the first feature and the third feature, and the DRGs are classified.

5. The method according to claim 4, characterized in that The extracting a fourth feature from the other information, the first feature, and the third feature through the common layer of the preset classification model, and classifying DRGs, includes: Performing a fourth feature extraction on the attribute information, the diagnosis and treatment information, the first feature, and the third feature through the hidden layer and the output layer of the preset classification model; DRGs are classified based on the fourth feature and the classification layer in the preset classification model.

6. The method according to claim 1, characterized in that The method further comprises: Determining vector features of diagnosis and treatment information in the other information; The step of inputting the case information into a preset classification model includes: inputting the vector features into the preset classification model, so as to perform a second feature extraction on the vector features through a common layer of the preset classification model.

7. The method according to claim 1, characterized in that The method further comprises: Obtain patient medical history information; The historical case information is used as training data for a preset initial classification model to train the preset initial classification model.

8. The method according to claim 7, characterized in that The historical case information includes historical time information related to the case, historical attribute information of the patient, and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; The method of using the historical case information as training data for a preset initial classification model to train the preset initial classification model includes: Inputting the historical case information into a preset initial classification model, and performing a first historical feature extraction on the historical time information through a first initial sub-neural network model of the preset initial classification model; Through the initial common layer of the preset initial classification model, the second historical feature extraction is performed on the historical attribute information, the historical diagnosis and treatment information, and the first historical feature, and DRGs are classified to obtain a historical classification result; Based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

9. The method according to claim 8, characterized in that The extracting of a second historical feature from the historical attribute information, the historical diagnosis and treatment information, and the first historical feature through the initial common layer of the preset initial classification model, and classifying DRGs, includes: Performing second historical feature extraction on the historical attribute information, historical diagnosis and treatment information, and the first historical feature through the initial hidden layer and the initial output layer of the preset initial classification model; DRGs are classified based on the second historical features and the initial classification layer in the preset initial classification model.

10. The method according to claim 9, characterized in that The classifying of DRGs based on the second historical feature and the initial classification layer in the preset initial classification model includes: The second historical features are respectively input into multiple second initial sub-neural network models in the preset initial classification model to obtain corresponding output historical features, and are classified by corresponding initial classification functions to obtain multiple historical classification results.

11. The method according to claim 10, characterized in that The method further comprises: Based on the multiple historical classification results and the DRGs classification results corresponding to the historical case information, as well as the differences between the multiple historical classification results, the training of the preset initial classification model is completed.

12. The method according to claim 8, characterized in that The historical case information also includes historical identification information of the hospital; The method further comprises: After inputting the historical case information into a preset initial classification model, performing a third historical feature extraction on the historical identification through a third initial sub-neural network model of the preset initial classification model; Through the initial common layer of the preset initial classification model, the fourth feature extraction is performed on the historical attribute information, historical diagnosis and treatment information, the first feature and the third historical feature, and DRGs are classified to obtain a historical classification result.

13. The method according to claim 8, characterized in that The preset initial classification model includes an initial input layer, an initial hidden layer, an initial output layer and an initial classification layer connected in sequence; A first initial sub-neural network model and a third initial sub-neural network model are connected between part of the initial input layer and the initial hidden layer.

14. The method according to claim 13, wherein: The initial classification layer includes multiple second initial sub-neural network models and corresponding initial classification functions.

15. The method according to claim 7, characterized in that The method further comprises: The preset initial classification model is trained in a safe environment.

16. A structure of an information classification model, characterized in that: It includes an input layer and a common layer connected in sequence, wherein the common layer includes a hidden layer, an output layer, and a classification layer; A first sub-neural network model and a second sub-neural network model are connected between part of the input layer and the hidden layer; Obtain the patient's historical case information; the historical case information includes historical time information related to the case, the patient's historical attribute information and historical diagnosis and treatment information, and the historical case information has corresponding DRGs classification results; Inputting the historical case information into an initial input layer, and performing a first historical feature extraction on the historical time information through an initial first sub-neural network model; performing a second historical feature extraction on the historical attribute information, historical diagnosis and treatment information, and the first historical feature through an initial common layer, and performing DRGs classification, including: performing a second historical feature extraction on the attribute information, diagnosis and treatment information, and the first historical feature through an initial hidden layer and an initial output layer; and performing DRGs classification based on the second historical feature and the initial classification layer; Based on the historical classification results and the DRGs classification results corresponding to the historical case information, the initial model training is completed.

17. The structure according to claim 16, characterized in that The classification layer includes multiple third sub-neural network models and corresponding initial classification functions.

18. The structure according to claim 16, characterized in that The extracting of a second historical feature from the historical attribute information, the historical diagnosis and treatment information, and the first historical feature through the initial common layer, and classifying DRGs, includes: Extracting a second historical feature from the attribute information, the diagnosis and treatment information, and the first historical feature through an initial hidden layer and an initial output layer; DRGs are classified based on the second historical features and the initial classification layer.

19. The structure according to claim 18, characterized in that The classifying of DRGs based on the second historical feature and the initial classification layer includes: The second historical features are respectively input into multiple initial third sub-neural network models to obtain corresponding output historical features, and are respectively classified by corresponding initial classification functions to obtain multiple historical classification results.

20. The structure according to claim 19, characterized in that The training of the initial model is completed based on the multiple historical classification results and the DRGs classification results corresponding to the historical case information, as well as the differences between the multiple historical classification results.

21. The structure according to claim 16, characterized in that The historical case information also includes historical identification information of the hospital; After the historical case information is input into the initial model, the initial second sub-neural network model is used to extract the third historical features of the historical identification information; Through the initial common layer, the fourth feature extraction is performed on the attribute information, diagnosis and treatment information, the first historical feature and the third historical feature, and DRGs are classified to obtain a historical classification result.

22. A method for generating a classification model of information, characterized in that: include: Obtaining historical medical information of the patient, wherein the historical medical information has corresponding DRGs classification results; Inputting the historical case information into a preset initial classification model, and performing a first historical feature extraction on the historical time information in the historical case information through a first initial sub-neural network model of the preset initial classification model; Through the common layer of the preset initial classification model, other historical information in the historical case information and the first historical feature are subjected to second historical feature extraction, and DRGs are classified, including: through the initial hidden layer and the initial output layer of the preset initial classification model, the historical attribute information, the historical diagnosis and treatment information, and the first historical feature are subjected to second historical feature extraction; based on the second historical feature and the initial classification layer in the preset initial classification model, the DRGs are classified to obtain a historical classification result; Based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

23. The method according to claim 22, characterized in that The classifying of DRGs based on the second historical feature and the initial classification layer in the preset initial classification model includes: The second historical features are respectively input into multiple second initial sub-neural network models in the preset initial classification model to obtain corresponding output historical features, and are classified by corresponding initial classification functions to obtain multiple historical classification results.

24. The method according to claim 23, wherein The method further comprises: Based on the multiple historical classification results and the DRGs classification results corresponding to the historical case information, as well as the differences between the multiple historical classification results, the training of the preset initial classification model is completed.

25. The method according to claim 22, wherein The historical case information also includes historical identification information of the hospital; The method further comprises: After inputting the historical case information into a preset initial classification model, performing a third historical feature extraction on the historical identification information through a third initial sub-neural network model of the preset initial classification model; Through the initial common layer of the preset initial classification model, the fourth feature extraction is performed on the other historical information, the first historical feature and the third historical feature, and DRGs are classified to obtain a historical classification result.

26. A computing device comprising: memory, processors, and communication components; The memory is used to store computer programs; The communication component is used to obtain case information of the patient to be classified; The processor is configured to execute the computer program to: Inputting the case information into a preset classification model, and performing a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model; Performing a second feature extraction on other information in the case information and the first feature through the common layer of the preset classification model, and classifying DRGs, including: performing a second feature extraction on attribute information, diagnosis and treatment information in the case information and the first feature through the hidden layer and the output layer of the preset classification model; DRGs are classified based on the second feature and the classification layer in the preset classification model.

27. A computing device comprising: memory, processors, and communication components; The memory is used to store computer programs; The communication component is used to obtain the patient's historical case information, wherein the historical case information has corresponding DRGs classification results; The processor is configured to execute the computer program to: Inputting the historical case information into a preset initial classification model, and performing a first historical feature extraction on the historical time information in the historical case information through a first initial sub-neural network model of the preset initial classification model; Through the common layer of the preset initial classification model, other historical information in the historical case information and the first historical feature are subjected to second historical feature extraction, and DRGs are classified, including: through the initial hidden layer and the initial output layer of the preset initial classification model, the historical attribute information, the historical diagnosis and treatment information, and the first historical feature are subjected to second historical feature extraction; based on the second historical feature and the initial classification layer in the preset initial classification model, the DRGs are classified to obtain a historical classification result; Based on the historical classification results and the DRGs classification results corresponding to the historical case information, the training of the preset initial classification model is completed.

28. A computing device comprising: memory, processors, and communication components; The memory is used to store computer programs; The communication component is used to obtain case information of the patient to be classified; The processor is configured to execute the computer program to: Inputting the case information into a preset classification model, and performing a first feature extraction on the time information in the case information through a first sub-neural network model of the preset classification model; Performing a second feature extraction on the attribute information, diagnosis and treatment information, and the first feature in the case information through the common layer of the preset classification model, and classifying DRGs, including: performing a second historical feature extraction on the attribute information, diagnosis and treatment information, and the first historical feature through an initial hidden layer and an initial output layer; DRGs are classified based on the second historical features and the initial classification layer.

29. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by one or more processors, the one or more processors are caused to implement the steps of the method according to any one of claims 1 to 15 and 22 to 25.

Images