Information identification method, device, computer equipment and storage medium

Abstract

The present application discloses an information identification method, device, computer equipment and storage medium, including: inputting N diagnosis and treatment information of a target patient into a first learning model in sequence, each diagnosis and treatment information of the N diagnosis and treatment information includes at least one information, and the first learning model is an unsupervised learning model; using the first learning model, obtaining N output vectors corresponding to the N diagnosis and treatment information, and the N output vectors are used to characterize the diagnosis and treatment result data of the N diagnosis and treatment information; using a preset algorithm, splicing the N diagnosis and treatment information and the N output vectors to obtain a splicing vector; inputting the splicing vector into a second learning model for processing, and the second learning model is a supervised learning model; when the splicing vector meets a predetermined condition, determining that the target patient is an abnormal patient. In this way, abnormal patients can be determined more accurately, and the efficiency of determining abnormal patients can be improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of information, and in particular to an information identification method, device, computer equipment and storage medium. Background Art

[0002] At present, with the rapid development of electronic equipment technology and big data technology, medical quality control is becoming more and more precise.

[0003] Since the process of patients' medical treatment can be recorded through the electronic medical treatment system, after collecting a certain amount of patient medical treatment data, the quality of the patients' medical treatment can be monitored in the subsequent patient medical treatment process. Among them, the identification of abnormal patients is one of the important tasks of medical quality control. In related technologies, the basic information of patients (for example, age, gender, height, weight, etc.) is mainly analyzed to identify abnormal patients from the statistical values ​​of these basic information. For example, some patients defraud insurance by falsifying medical treatment data, and these patients are abnormal patients.

[0004] However, since the information obtained is relatively limited, the recognition ability of the above-mentioned method of identifying abnormal patients is also limited, which in turn reduces the efficiency and accuracy of identifying abnormal patients. Summary of the invention

[0005] The embodiments of the present invention provide an information identification method, apparatus, computer equipment and storage medium capable of greatly improving the extensiveness, accuracy and timeliness of information identification.

[0006] In order to solve the above technical problems, a technical solution adopted by an embodiment created by the present invention is: providing an information identification method, comprising: inputting N medical information of a target patient into a first learning model in sequence, each medical information of the N medical information includes at least one information, and the first learning model is an unsupervised learning model; using the first learning model, obtaining N output vectors corresponding to the N medical information, and the N output vectors are used to represent the medical result data of the N medical information; using a preset algorithm, splicing the N medical information and the N output vectors to obtain a spliced ​​vector; inputting the spliced ​​vector into the second learning model for processing, and the second learning model is a supervised learning model; when the spliced ​​vector meets a predetermined condition, determining that the target patient is an abnormal patient.

[0007] Optionally, the above-mentioned N medical treatment information of the target patient is input into the first learning model in sequence. The information identification method provided by the embodiment of the present invention may include: inputting the medical treatment sub-information corresponding to each medical treatment information in the N medical treatment information of the above-mentioned target patient into the above-mentioned first learning model, and the medical treatment sub-information corresponding to each medical treatment information includes at least one information.

[0008] Optionally, the above diagnosis and treatment sub-information includes: project name information, project type information, and project price information.

[0009] Optionally, in the above-mentioned determination that the target patient is an abnormal patient when the above-mentioned splicing vector satisfies a predetermined condition, the information identification method provided by an embodiment of the present invention may include: using the above-mentioned second learning model to identify the above-mentioned splicing vector to obtain a target identification value; when the output value corresponding to the above-mentioned target identification value satisfies a predetermined threshold, determining that the above-mentioned target patient is an abnormal patient.

[0010] Optionally, the above-mentioned N medical treatment information of the target patient is input into the first learning model in sequence. The information identification method provided by the embodiment of the present invention may include: inputting the N medical treatment information of the target patient into the first learning model in sequence, and filling the placeholder mark for the first vector position in the above-mentioned first learning model.

[0011] Optionally, in the information identification method provided by the embodiment of the present invention, the information identification method provided by the embodiment of the present invention may include: the above-mentioned first learning model includes M vector positions and X-layer vector processing encoders, and the M vector positions include random vector values, each vector position in the above-mentioned M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information in the above-mentioned N times of diagnosis and treatment information, and the above-mentioned X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the above-mentioned M vector positions, and M and X are positive integers.

[0012] Optionally, the above-mentioned N times of diagnosis and treatment information of the target patient is input into the first learning model in sequence. The information identification method provided by the embodiment of the present invention may include: starting from the second vector position of the above-mentioned first learning model, the N times of diagnosis and treatment information of the target patient is input into the first learning model in sequence.

[0013] Optionally, the above-mentioned use of the first learning model to obtain N output vectors corresponding to the above-mentioned N times of medical treatment information includes: using the above-mentioned X-layer vector processing encoder of the above-mentioned first learning model to sequentially learn the above-mentioned N times of medical treatment information, and sequentially obtain the output vector corresponding to each time of the above-mentioned N times of medical treatment information.

[0014] To solve the above technical problems, an embodiment of the present invention further provides an information identification device, comprising: an input module, used to input N medical information of a target patient into a first learning model in sequence, each medical information of the N medical information includes at least one information, and the first learning model is an unsupervised learning model; an acquisition module, used to use the first learning model to obtain N output vectors corresponding to the N medical information, and the N output vectors are used to represent the medical result data of the N medical information; a splicing module, used to use a preset algorithm to splice the N medical information and the N output vectors to obtain a spliced ​​vector; the input module is also used to input the spliced ​​vector into the second learning model for processing, and the second learning model is a supervised learning model; a determination module, used to determine that the target patient is an abnormal patient when the spliced ​​vector meets a predetermined condition.

[0015] Optionally, the input module is specifically used to input the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information of the target patient's N times of diagnosis and treatment information into the first learning model, and the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information includes at least one information.

[0016] Optionally, the above diagnosis and treatment sub-information includes: project name information, project type information, and project price information.

[0017] Optionally, the above-mentioned information identification device also includes: an acquisition module, used to use the above-mentioned second learning model to identify the above-mentioned splicing vector and obtain a target identification value; the above-mentioned determination module is specifically used to determine that the above-mentioned target patient is an abnormal patient when the output value corresponding to the above-mentioned target identification value meets a predetermined threshold.

[0018] Optionally, the input module is specifically used to input the target patient's N diagnosis and treatment information into the first learning model in sequence, and fill the first vector position in the first learning model with a placeholder mark.

[0019] Optionally, the first learning model includes M vector positions and X-layer vector processing encoders, and the M vector positions include random vector values, each of the M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each of the N diagnosis and treatment information, and the X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the M vector positions, and M and X are positive integers.

[0020] Optionally, the input module is specifically used to input the N diagnosis and treatment information of the above-mentioned target patient into the first learning model in sequence starting from the second vector position of the above-mentioned first learning model.

[0021] Optionally, the acquisition module is specifically used to use the X-layer vector processing encoder of the first learning model to sequentially learn the N medical information, and sequentially acquire the output vector corresponding to each medical information in the N medical information.

[0022] To solve the above technical problem, an embodiment of the present invention further provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the above-mentioned information identification method.

[0023] In order to solve the above technical problem, an embodiment of the present invention further provides a storage medium storing computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the above-mentioned information identification method.

[0024] The beneficial effect of the embodiment of the present invention is: after the target patient's N times of diagnosis and treatment information is sequentially input into the first learning model (each diagnosis and treatment information of the N times of diagnosis and treatment information includes at least one information, and the first learning model is an unsupervised learning model), the first learning model is used to obtain the N output vectors corresponding to the above N times of diagnosis and treatment information (the N output vectors represent the diagnosis and treatment result data of the N times of diagnosis and treatment information), and then, the preset algorithm is used to splice the above N times of diagnosis and treatment information and the N output vectors to obtain the spliced ​​vector, and finally, the above spliced ​​vector is input into the second learning model for processing, and when the predetermined conditions are met, the target patient is determined to be an abnormal patient. In this way, the limitation of determining abnormal patients through the patient's personal basic information is broken, and by obtaining the N times of diagnosis and treatment information of the patient's diagnosis and treatment information, that is, obtaining the patient's long-term diagnosis and treatment information, the authenticity of the patient's illness can be comprehensively judged using long-term data, and then abnormal patients can be determined more accurately, and the efficiency of determining abnormal patients can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0026] Figure 1 A basic flow chart of an information identification method according to a specific embodiment of the present invention;

[0027] Figure 2 A schematic diagram of an information identification method according to an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of the basic structure of an information identification device according to an embodiment of the present invention;

[0029] Figure 4 The basic structural block diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0031] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0032] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0033] It will be understood by those skilled in the art that the term "terminal" as used herein includes both devices with wireless signal receivers, which are devices with only wireless signal receivers without transmission capabilities, and devices with receiving and transmitting hardware, which are devices with receiving and transmitting hardware capable of performing two-way communications over a two-way communication link. Such devices may include: cellular or other communication devices, which have a single-line display or a multi-line display or a cellular or other communication device without a multi-line display; PCS (Personal Communications Service), which may combine voice, data processing, fax and / or data communication capabilities; PDA (Personal Digital Assistant), which may include a radio frequency receiver, a pager, Internet / intranet access, a web browser, a notepad, a calendar and / or a GPS (Global Positioning System) receiver; conventional laptop and / or palmtop computers or other devices, which have and / or include a conventional laptop and / or palmtop computer or other device with and / or including a radio frequency receiver. The "terminal" used here can be portable, transportable, installed in a vehicle (air, sea and / or land), or suitable for and / or configured to run locally, and / or in a distributed form, running at any other location on the earth and / or in space. The "terminal" used here can also be a communication terminal, an Internet terminal, a music / video playing terminal, for example, a PDA, a MID (Mobile Internet Device) and / or a mobile phone with a music / video playing function, or a smart TV, a set-top box and other devices.

[0034] See also Figure 1 , Figure 1 The basic flow chart of the information identification method provided in this embodiment includes steps 201 to 205:

[0035] Step 201: Input N diagnosis and treatment information of the target patient into the first learning model in sequence.

[0036] In the embodiment of the present invention, each of the N medical treatment information includes at least one piece of information, and the first learning model is an unsupervised learning model.

[0037] In the embodiment of the present invention, the target patient may be any patient, or a patient whose medical information may have problems and whose medical information needs to be checked, for example, a patient who falsely reports his medical condition to an insurance company.

[0038] In the embodiment of the present invention, the above-mentioned N times of medical treatment information is information about the user's multiple visits to the hospital for medical treatment, and each time of medical treatment information corresponds to a medical treatment process.

[0039] Furthermore, each consultation process may include multiple consultation items, that is, each diagnosis and treatment information may include more than one information. For example, a consultation process may include: ultrasound examination items, blood routine examination items, X-ray examination items, etc.

[0040] In the embodiment of the present invention, the first learning model may be a pre-set learning model.

[0041] In one example, the first learning model is a learning model obtained through learning and training based on the medical information of known abnormal patients. The first learning model can be used to represent the medical behavior of each medical information of the patient, for example, after the first learning model is used to count the medical behavior of a patient's single medical visit, the behavior vector corresponding to the medical behavior of the medical visit is represented.

[0042] It can be understood that the N times of medical treatment information of the above-mentioned target patient can be counted according to the statistical method required by the first learning model, and then the N times of medical treatment information can be input into the first learning model in sequence. For example, after the medical treatment information is counted using a table in the target format, N tables containing the medical treatment information are generated, and then the above N tables are input into the first learning model in sequence.

[0043] Step 202: Using the first learning model, obtain N output vectors corresponding to the N medical treatment information.

[0044] In the embodiment of the present invention, the above-mentioned N output vectors are used to represent the diagnosis and treatment result data of N diagnosis and treatment information.

[0045] In the embodiment of the present invention, N pieces of diagnosis and treatment information are processed in sequence by the first learning model to obtain N output vectors.

[0046] Optionally, in an embodiment of the present invention, the first learning model includes M vector positions and X layers of vector processing encoders, and the M vector positions include random vector values, and M and X are positive integers.

[0047] In an embodiment of the present invention, each of the M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information in the N times of diagnosis and treatment information, and the X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the M vector positions.

[0048] In the embodiment of the present invention, the vector positions of the first learning model are preset in advance. For example, the first learning model is preset in advance to include 4 types of vectors, and each type of vector includes 512 vector positions.

[0049] Furthermore, the above-mentioned M vector positions are embedding layers of the input diagnosis and treatment information. When the first learning model includes multiple types of vectors, each type of vector is an embedding layer.

[0050] It should be noted that the above M vector positions include random vector values ​​before information filling. After information filling, the filled vector positions become the actual filled information, while the unfilled vector positions still contain the previous vector random values.

[0051] In the embodiment of the present invention, the above-mentioned X-layer vector processing encoder is preset in advance. For example, it is preset in advance that the above-mentioned first learning model includes 12 layers of vector processing encoders, and each layer of vector processors is arranged in sequence.

[0052] It should be noted that the order of the above-mentioned M vector positions and the above-mentioned X-layer vector processing encoder in the first learning model is M vector positions in front and the above-mentioned X-layer vector processing encoder in the back, that is, the processing process of the above-mentioned first learning model is: first input the medical information into the M vector positions, and then input the medical information passing through the vector positions into the M-layer encoder for processing, and finally obtain the output vector corresponding to the above-mentioned medical information.

[0053] It should be noted that in the above-mentioned first learning model, for each piece of medical information, the above-mentioned M vector positions correspond to multiple types of vectors. When the final output vector is multiple output vectors, only the first vector is taken as the final actual output vector.

[0054] Step 203: Using a preset algorithm, concatenate the N diagnosis and treatment information and the N output vectors to obtain a concatenated vector.

[0055] In the embodiment of the present invention, the above-mentioned preset algorithm may be preset in advance or may be user-defined, which is not limited in the embodiment of the present invention.

[0056] In the embodiment of the present invention, the above-mentioned splicing vector is used to represent the overall diagnosis and treatment process of the above-mentioned N times of diagnosis and treatment information.

[0057] In one example, the above preset algorithm can refer to the following formula 1, where v1, v2 to v n Embedding TF (v1), Embedding TF (v2), Embedding TF (v n ) are N output vectors, E v is the concatenated vector:

[0058] E v =EmbeddingTF (v1)||Embedding TF (v2)...||Embedding TF (v n ) (Formula 1)

[0059] Step 204: Input the concatenated vector into the second learning model for processing.

[0060] In an embodiment of the present invention, the second learning model is a supervised learning model.

[0061] In an embodiment of the present invention, the second learning model is a learning model obtained after multiple trainings using diagnosis and treatment information of abnormal patients. The second model can be used to detect abnormal conditions of the splicing vector.

[0062] Step 205: When the splicing vector satisfies a predetermined condition, the target patient is determined to be an abnormal patient.

[0063] In the embodiment of the present invention, the above-mentioned predetermined condition may be preset or may be a custom setting, and the embodiment of the present invention does not limit this.

[0064] In an example, satisfying the preset condition means that the second learning model detects an abnormality in the splicing vector and outputs a numerical value for characterizing the abnormality.

[0065] In the embodiment of the present invention, when the above-mentioned splicing vector does not meet the predetermined condition, it can be determined that the target patient is not an abnormal patient.

[0066] In the information identification method provided by the embodiment of the present invention, after the target patient's N times of diagnosis and treatment information are sequentially input into the first learning model (each diagnosis and treatment information of the N times of diagnosis and treatment information includes at least one information, and the first learning model is an unsupervised learning model), the first learning model is used to obtain the N output vectors corresponding to the above N times of diagnosis and treatment information (the N output vectors represent the diagnosis and treatment result data of the N times of diagnosis and treatment information), and then, the preset algorithm is used to splice the above N times of diagnosis and treatment information and the N output vectors to obtain the spliced ​​vector, and finally, the above spliced ​​vector is input into the second learning model for processing, and when the predetermined conditions are met, the target patient is determined to be an abnormal patient. In this way, the limitation of determining abnormal patients by the patient's personal basic information is broken, and by obtaining the N times of diagnosis and treatment information of the patient's diagnosis and treatment information, that is, obtaining the patient's long-term diagnosis and treatment information, the authenticity of the patient's illness can be comprehensively judged using long-term data, and then the abnormal patients can be determined more accurately, and the efficiency of determining abnormal patients can be improved.

[0067] Optionally, in the embodiment of the present invention, in the above step 201, the information identification method provided by the embodiment of the present invention further includes the following step A:

[0068] Step A: Input the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information of the target patient's N diagnosis and treatment information into the first learning model.

[0069] Exemplarily, the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information includes at least one piece of information.

[0070] Optionally, in an embodiment of the present invention, the above-mentioned diagnosis and treatment sub-information includes: project name information, project type information, and project price information.

[0071] Exemplarily, as described above, each of the N medical treatment information may include multiple types of vectors. For example, in one medical treatment information, multiple items are included, and each item corresponds to its own item name information, item type information, and item price information.

[0072] Further, in combination with the above content, when the diagnosis and treatment sub-information includes: project name information, project type information, and project price information, the first learning model includes 4 types of vectors, namely: project name vector, project type vector, project price vector, and comprehensive input vector. Among them, the input method of the diagnosis and treatment sub-information is: input in sequence starting from the second vector position.

[0073] It should be noted that each of the above four types of vectors corresponds to an embedding layer of the first learning model.

[0074] In one example, when the diagnosis and treatment sub-information is the item name information, correspondingly, the sub-information is located in the name embedding layer of the first learning model and is the item name vector. The name embedding layer is used to vectorize the name of the charging item.

[0075] In one example, when the diagnosis and treatment sub-information is item type information, correspondingly, the sub-information is located in the type embedding layer of the first learning model, which is an item type vector. The type embedding layer is used to vectorize the type of the charging item, for example, bed care fee, examination fee, and inspection fee.

[0076] In one example, when the diagnosis and treatment sub-information is the project price information, the corresponding sub-information is located in the price embedding layer of the first learning model, which is the project price vector. The price embedding layer is used to vectorize the price range of the charging items. For example, multiple charging intervals (0-100 yuan, 100-200 yuan...) are divided in advance in the price embedding layer. The diagnosis and treatment sub-information in actual application is 150 yuan, and the corresponding price range is 100-200 yuan. The project price vector of the final price embedding layer is the above-mentioned price range vector of 100-200 yuan.

[0077] Furthermore, after obtaining the above-mentioned diagnosis and treatment sub-information, an input vector can be obtained through processing by the first learning model, and the input vector exists in the input embedding layer.

[0078] It can be understood that in each of the above embedding layers, vectors corresponding to the same type have the same representation.

[0079] Example 1: Figure 2 As shown, in Figure 2 In the above, N0, T0, C0, and E0 correspond to the first vector of the name embedding layer, type embedding layer, price embedding layer, and input embedding layer, respectively. The filling vector is a vector random value. Starting from N1, T1, C1, and the subsequent vectors (for example, N2, T2, C2), it can be used to fill in the diagnosis and treatment sub-information. Among them, each diagnosis and treatment information corresponds to the above three sub-information, and finally generates the input vector, that is, E1 and the subsequent vectors (for example, E2).

[0080] Furthermore, each group of diagnosis and treatment sub-information and its corresponding input vector are in a one-to-one correspondence. For example, for ultrasound items, N1 is the ultrasound name vector, T1 is the ultrasound type vector, that is, the examination fee vector, and C1 is the ultrasound charge vector, that is, the vector corresponding to the price range of 100-200 yuan. Finally, the ultrasound input vector E1 is obtained by processing the above N1, T1, and C1; for routine blood test items, N2 is the routine blood test name vector, T2 is the routine blood test type vector, that is, the examination fee vector, and C2 is the routine blood test charge vector, that is, the vector corresponding to the price range of 0-100 yuan. Finally, the ultrasound input vector E2 is obtained by processing the above N2, T2, and C2.

[0081] It should be noted that in the above-mentioned first learning model, for medical information, the above-mentioned M vector positions correspond to 4 types of vectors. After being processed by the 12-layer encoder, the final output vectors include O0, O1, and O2, but only O0 is used as the final output vector.

[0082] In this way, the electronic device uses the first learning model in the above manner to process each of the N ongoing diagnosis and treatment information in turn, thereby making full use of the target patient's medical information, and ultimately obtaining accurate diagnosis and treatment result data, and then accurately judging the target patient as an abnormal patient in the subsequent period.

[0083] Optionally, in the embodiment of the present invention, in the above step 204, the information identification method provided by the embodiment of the present invention further includes the following step B:

[0084] Step B1: using the second learning model to identify the concatenated vector and obtain a target recognition value.

[0085] Step B2: When the output value corresponding to the target identification value satisfies a predetermined threshold, the target patient is determined to be an abnormal patient.

[0086] Exemplarily, the target identification value may be preset in advance or may be user-defined.

[0087] Exemplarily, the target identification value is used to characterize the abnormal and normal states of the splicing vector.

[0088] In an example, the target identification value may be 0 and 1, where 0 is used to indicate that the splicing vector is normal, and 1 is used to indicate that the splicing vector is abnormal.

[0089] Exemplarily, the above-mentioned predetermined threshold may be preset in advance or may be user-defined.

[0090] Exemplarily, the above-mentioned predetermined threshold and the target recognition value correspond to each other.

[0091] In one example, when the target identification value can be 0 and 1, the preset threshold is 1, and when the target identification value is 1, the target patient is determined to be an abnormal patient.

[0092] In this way, the abnormal state of the target patient can be accurately and conveniently screened out, thereby helping the user to perform relevant treatment on the target patient in the future.

[0093] Optionally, in the embodiment of the present invention, in the above step 201, the information identification method provided by the embodiment of the present invention further includes the following step C:

[0094] Step C: Input the target patient's N diagnosis and treatment information into the first learning model in sequence, and fill the first vector position in the first learning model with a placeholder mark.

[0095] Exemplarily, as described above, the first position among the M vector positions in the first learning model is a placeholder, that is, a random vector value.

[0096] Furthermore, in an embodiment of the invention, as described above, in the above-mentioned first learning model, M random vector values ​​are pre-filled in M ​​vector positions. When the learning model is started, other vector positions except the first vector position of the first learning model can be used to fill in the medical sub-information corresponding to each of the N medical information, and the first vector position is still filled with random vector values. At this time, the random vector value is a placeholder mark.

[0097] In one example, when the M vector positions are divided into four categories of vectors, the first vector in each category of vectors is a placeholder. For example, in the above example 1, N0, T0, C0, and E0 are all placeholders.

[0098] Optionally, in the embodiment of the present invention, in the above step 201, the information identification method provided by the embodiment of the present invention further includes the following step D:

[0099] Step D: Input the target patient's N diagnosis and treatment information into the first learning model in sequence, starting from the second vector position of the first learning model.

[0100] Optionally, in the embodiment of the present invention, in the above step 202, the information identification method provided by the embodiment of the present invention further includes the following step E:

[0101] Step E1: using the X-layer vector processing encoder of the first learning model to sequentially learn the N times of diagnosis and treatment information, and sequentially obtain the output vector corresponding to each time of the N times of diagnosis and treatment information.

[0102] Exemplarily, as described above, the first learning model includes an X-layer vector processing encoder. In the process of using the first learning model to process the N times of medical information, the first learning model receives each of the N times of medical information in turn, and processes the N times of medical information in turn. For example, after receiving three times of medical information, namely, medical information 1, medical information 2, and medical information 3, the first learning model processes medical information 1 first, and after obtaining output vector 1 of medical information 1, processes medical information 2 again to obtain output vector 2 of medical information 2, and finally processes medical information 3 to obtain output vector 3 of medical information 3, and finally obtains three groups of output vectors.

[0103] In order to solve the above technical problems, the embodiment of the present invention also provides an information identification device. Figure 3 , Figure 3 This is a schematic diagram of the basic structure of the information identification device implemented in this embodiment.

[0104] like Figure 3As shown, an information identification device includes: an input module 301, which is used to input N medical information of a target patient into a first learning model in sequence, each of the N medical information includes at least one information, and the first learning model is an unsupervised learning model; an acquisition module 302, which is used to use the first learning model to obtain N output vectors corresponding to the N medical information, and the N output vectors are used to represent the medical result data of the N medical information; a splicing module 303, which is used to use a preset algorithm to splice the N medical information and the N output vectors to obtain a splicing vector; the input module 301 is also used to input the splicing vector into the second learning model for processing, and the second learning model is a supervised learning model; a determination module 304, which is used to determine that the target patient is an abnormal patient when the splicing vector meets a predetermined condition.

[0105] After the information identification device sequentially inputs the target patient's N medical information into the first learning model (each medical information of the N medical information includes at least one piece of information, and the first learning model is an unsupervised learning model), the first learning model is used to obtain the N output vectors corresponding to the above N medical information (the N output vectors represent the medical result data of the N medical information), and then the preset algorithm is used to splice the above N medical information and the N output vectors to obtain the spliced ​​vector. Finally, the spliced ​​vector is input into the second learning model for processing, and the target patient is determined to be an abnormal patient when the predetermined conditions are met. In this way, the limitation of determining abnormal patients through the patient's personal basic information is broken. By obtaining the N medical information of the patient's medical information, that is, obtaining the patient's long-term medical information, the authenticity of the patient's illness can be comprehensively judged using the long-term data, and then the abnormal patients can be more accurately determined, thereby improving the efficiency of determining abnormal patients.

[0106] In some embodiments, the above-mentioned information identification device also includes: the above-mentioned input module, which is specifically used to input the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information of the above-mentioned target patient into the above-mentioned first learning model, and the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information includes at least one information.

[0107] In some embodiments, the diagnosis and treatment sub-information includes: project name information, project type information, and project price information.

[0108] In some embodiments, the above-mentioned information identification device also includes: an acquisition module, used to use the above-mentioned second learning model to identify the above-mentioned splicing vector and obtain a target identification value; the above-mentioned determination module is specifically used to determine that the above-mentioned target patient is an abnormal patient when the output value corresponding to the above-mentioned target identification value meets a predetermined threshold.

[0109] In some embodiments, the input module is specifically used to input N diagnosis and treatment information of the target patient into the first learning model in sequence, and fill the first vector position in the first learning model with a placeholder mark.

[0110] In some embodiments, the first learning model includes M vector positions and X-layer vector processing encoders, and the M vector positions include random vector values, each of the M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each of the N diagnosis and treatment information, and the X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the M vector positions, and M and X are positive integers.

[0111] In some embodiments, the input module is specifically used to input the N diagnosis and treatment information of the target patient into the first learning model in sequence starting from the second vector position of the first learning model.

[0112] In one embodiment, the acquisition module is specifically used to use the above-mentioned X-layer vector processing encoder of the above-mentioned first learning model to sequentially learn the above-mentioned N medical treatment information, and sequentially obtain the output vector corresponding to each medical treatment information in the above-mentioned N medical treatment information.

[0113] To solve the above technical problems, the embodiment of the present invention also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0114] like Figure 4 As shown, a schematic diagram of the internal structure of a computer device. The computer device includes a processor, a non-volatile storage medium, a memory, and a network interface connected via a system bus. Among them, the non-volatile storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The database may store a control information sequence. When the computer-readable instructions are executed by the processor, the processor can implement a data relationship reconstruction method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute a data relationship reconstruction method. The network interface of the computer device is used to connect and communicate with a terminal. Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0115] In this embodiment, the processor is used to execute Figure 3 The memory stores the program codes and various data required to execute the above modules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program codes and data required to execute all sub-modules in the data relationship reconstruction device, and the server can call the program codes and data of the server to execute the functions of all sub-modules.

[0116] The computer equipment constructs the data relationship between the collected transaction nodes into a node feature graph and a corresponding adjacency matrix, and uses a graph neural network to extract features from the node feature graph and the adjacency matrix, and reconstructs the relationship based on the extracted features. When reconstructing the relationship, the graph neural network reconstructs the data based on the learned hidden logical relationship between the data. The process of relationship reconstruction is the process of restoring the data node feature graph and the adjacency matrix. In the process of breaking up and reorganizing the data node feature graph and the adjacency matrix, the graph neural network verifies the correctness of the hidden association relationship between the transaction nodes it has learned, and then reconstructs the relationship between the transaction nodes using the learned hidden relationship, and selects the target transaction nodes that meet the conditions for subsequent use. Through graphics processing technology, the logical relationship behind the data is deeply interpreted, which improves the depth and efficiency of data analysis.

[0117] The present invention also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the data relationship reconstruction method in any of the above embodiments.

[0118] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0119] The present invention also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the data relationship reconstruction method in any of the above embodiments.

[0120] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0121] Those skilled in the art will appreciate that the various operations, methods, steps, measures, and schemes in the processes discussed in this application may be alternated, altered, combined, or deleted. Further, other steps, measures, and schemes in the various operations, methods, and processes discussed in this application may also be alternated, altered, rearranged, decomposed, combined, or deleted. Further, the steps, measures, and schemes in the prior art that are similar to those disclosed in this application may also be alternated, altered, rearranged, decomposed, combined, or deleted.

[0122] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. An information identification method, characterized in that: include: Inputting N diagnosis and treatment information of the target patient into a first learning model in sequence, wherein each diagnosis and treatment information of the N diagnosis and treatment information includes at least one information, and the first learning model is an unsupervised learning model; Using the first learning model, obtaining N output vectors corresponding to the N times of diagnosis and treatment information, wherein the N output vectors are used to represent diagnosis and treatment result data of the N times of diagnosis and treatment information; Using a preset algorithm, splicing the N diagnosis and treatment information and the N output vectors to obtain a spliced ​​vector; Inputting the splicing vector into a second learning model for processing, wherein the second learning model is a supervised learning model, the second learning model is a learning model obtained after multiple training using diagnosis and treatment information of abnormal patients, and the second learning model is used to detect abnormal conditions of the splicing vector; When the splicing vector satisfies a predetermined condition, determining that the target patient is an abnormal patient; The step of sequentially inputting the target patient's N diagnosis and treatment information into the first learning model includes: Inputting N diagnosis and treatment information of the target patient into the first learning model in sequence, and filling the first vector position in the first learning model with a placeholder mark; The first learning model includes M vector positions and X-layer vector processing encoders, and the M vector positions include random vector values, each vector position in the M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information in the N diagnosis and treatment information, and the X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the M vector positions, and M and X are positive integers; Wherein, the N diagnosis and treatment information of the target patient is sequentially input into the first learning model starting from the second vector position of the first learning model; The using the first learning model to obtain N output vectors corresponding to the N diagnosis and treatment information includes: The X-layer vector processing encoder of the first learning model is used to sequentially learn the N times of medical treatment information, and output vectors corresponding to each time of the N times of medical treatment information are sequentially obtained.

2. The method according to claim 1, characterized in that The step of sequentially inputting the target patient's N diagnosis and treatment information into the first learning model includes: The diagnosis and treatment sub-information corresponding to each diagnosis and treatment information of the target patient's N times of diagnosis and treatment information is input into the first learning model, and the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information includes at least one information.

3. The method according to claim 2, characterized in that The diagnosis and treatment sub-information includes: project name information, project type information and project price information.

4. The method according to claim 1, characterized in that When the splicing vector satisfies a predetermined condition, determining that the target patient is an abnormal patient includes: Using the second learning model to identify the splicing vector, and obtaining a target recognition value; When the output value corresponding to the target identification value meets a predetermined threshold, the target patient is determined to be an abnormal patient.

5. An information recognition device, characterized in that: include: An input module, used to sequentially input N diagnosis and treatment information of a target patient into a first learning model, wherein each diagnosis and treatment information of the N diagnosis and treatment information includes at least one information, and the first learning model is an unsupervised learning model; An acquisition module, used to acquire N output vectors corresponding to the N times of diagnosis and treatment information by using the first learning model, wherein the N output vectors are used to represent diagnosis and treatment result data of the N times of diagnosis and treatment information; A splicing module, used to splice the N diagnosis and treatment information and the N output vectors using a preset algorithm to obtain a splicing vector; The input module is further used to input the splicing vector into a second learning model for processing, wherein the second learning model is a supervised learning model, and the second learning model is a learning model obtained after multiple training using diagnosis and treatment information of abnormal patients, and the second learning model is used to detect abnormal conditions of the splicing vector; A determination module, configured to determine that the target patient is an abnormal patient if the splicing vector satisfies a predetermined condition; The input module is specifically used to input N diagnosis and treatment information of the target patient into the first learning model in sequence, and fill the first vector position in the first learning model with a placeholder mark; The first learning model includes M vector positions and X-layer vector processing encoders, and the M vector positions include random vector values, each vector position in the M vector positions is used to fill in the diagnosis and treatment sub-information corresponding to each diagnosis and treatment information in the N diagnosis and treatment information, and the X-layer vector processing encoder is used to process the diagnosis and treatment sub-information filled in the M vector positions, and M and X are positive integers; The input module is specifically used to input the N diagnosis and treatment information of the target patient into the first learning model in sequence starting from the second vector position of the first learning model; The acquisition module is specifically used to use the X-layer vector processing encoder of the first learning model to sequentially learn the N times of medical treatment information, and sequentially obtain the output vector corresponding to each time of the N times of medical treatment information.

6. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the information identification method according to any one of claims 1 to 4.

7. A storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to perform the steps of the information identification method according to any one of claims 1 to 4.

Images