Full-process diagnosis and treatment path processing method and system, storage medium and electronic equipment

Abstract

The invention provides a whole-process diagnosis and treatment path processing method and system, a storage medium and an electronic device, and relates to the technical field of medical data processing.The processing method comprises the steps that at least two different modes are adopted to construct the corresponding relation between clinical manifestations and treatment items in disease course records; and constructing a corresponding relationship between the treatment items in the disease course record and the medical advice items in the medical advice sheet, so as to obtain a corresponding relationship among the clinical manifestation, the treatment scheme and the medical advice items, and based on the corresponding relationship, obtaining a whole-process diagnosis and treatment path containing'clinical manifestation-treatment scheme-medical advice items', and completely restoring a complete diagnosis and treatment path from symptoms to treatment execution.

Description

Technical Field

[0001] The present invention relates to the field of medical data processing technology, and in particular to a processing method, system, storage medium and electronic device for a full-process diagnosis and treatment pathway. Background Art

[0002] Currently, medical records, medical progress notes, examinations and tests, doctor's orders, and treatment plans are all digitized during the diagnosis and treatment process, generating medical data stored in various information systems. In current medical practice, medical records are generally recorded in free text, resulting in unstructured data. Existing natural language processing (NLP) methods have limited processing capabilities for unstructured data, making it difficult to accurately extract the relationship between each cause of clinical manifestation in the medical record and each treatment item in the treatment plan. Furthermore, there is a lack of direct connection between medical data stored in different information systems.

[0003] To sum up, the data generated during the diagnosis and treatment process is either difficult to extract effective information or is isolated and lacks correlation with each other, making it difficult to restore the complete diagnosis and treatment process from a multimodal perspective. Summary of the Invention

[0004] In view of this, an embodiment of the present invention provides a processing method, system, storage medium and electronic device for a full-process diagnosis and treatment pathway, so as to achieve the purpose of accurately and quickly restoring the diagnosis and treatment process and forming a complete diagnosis and treatment pathway.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A first aspect of an embodiment of the present invention discloses a method for processing a full-process diagnosis and treatment pathway, the method comprising:

[0007] Obtain all medical records of the patient during his / her stay in the hospital and sort them by time;

[0008] Performing a first diagnosis and treatment pathway extraction operation and a second diagnosis and treatment pathway extraction operation based on the sorted medical records;

[0009] Constructing a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway;

[0010] Storing the full-process diagnosis and treatment pathway in a clinical experience database, wherein the clinical experience database stores the full-process diagnosis and treatment pathways of historical patients;

[0011] The first diagnosis and treatment pathway extraction operation includes:

[0012] Extracting the causes of clinical manifestations, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes from the medical records based on the multimodal large language model, constructing a correspondence between the causes and treatment items, and a correspondence between the abnormal examination and test results and treatment items, to obtain a first diagnosis and treatment pathway;

[0013] The second diagnosis and treatment pathway extraction operation includes:

[0014] Extracting the medical order items of the corresponding time in the medical order form according to the chronological order of the medical course record to obtain the specific medical order items within the target time range;

[0015] Based on the knowledge graph, a correspondence between the specific medical order items and the treatment items is constructed to obtain a second diagnosis and treatment path.

[0016] Preferably, it also includes:

[0017] Obtain clinical presentation of new patients;

[0018] The clinical experience database is queried based on the clinical manifestations, and treatment plans and medical advice items corresponding to the clinical manifestations are retrieved and output.

[0019] Preferably, the method of extracting the cause of disease, abnormal examination and test results, and treatment items in the treatment plan corresponding to the cause of disease in the medical record based on the multimodal large language model, and constructing a correspondence between the cause of disease and the treatment items, as well as a correspondence between the abnormal examination and test results and the treatment items, to obtain a first diagnosis and treatment pathway includes:

[0020] Inputting a first free text recording clinical manifestations, a second free text recording treatment plans, and an examination and test sheet recording examination and test results in the medical record into a multimodal large language model, wherein the examination and test sheet includes an examination report, a test report, and / or an imaging image;

[0021] Extracting a first key entity corresponding to a clinical manifestation and a second key entity corresponding to a treatment plan using the multimodal large language model, wherein the first key entity includes at least the cause of the clinical manifestation and abnormal examination and test results, and the second key entity includes at least medications, surgical treatments, examination items, and test items related to the treatment;

[0022] According to a first preset correspondence relationship between the cause of disease and the treatment item, constructing a correspondence relationship between the cause of disease and the treatment item in the first key entity and the second key entity;

[0023] According to a second preset correspondence between the abnormality inspection and test results and the treatment items, constructing a correspondence between the abnormality inspection and test results and the treatment items in the first key entity and the second key entity;

[0024] A first diagnosis and treatment pathway is determined based on the correspondence between the cause of the disease and the treatment item, and the correspondence between the abnormal examination result and the treatment item.

[0025] Preferably, the corresponding relationship between the specific medical order items and the treatment items is constructed based on the knowledge graph to obtain the second diagnosis and treatment path, including:

[0026] Using the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph, the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and treatment items are identified and unified; the treatment items are treatment items in the treatment plan, and the treatment plan is obtained from the medical record;

[0027] For the drugs, examinations and tests, and surgeries with unified names in the specific medical order items and treatment items, the first-level relationships between the medical information pre-constructed in the knowledge graph are used to determine the second-level relationships between the drugs, examinations and tests, and surgeries with unified names;

[0028] A correspondence between the specific medical order item and the treatment item is constructed based on the second hierarchical relationship, and a second diagnosis and treatment pathway is determined using the correspondence between the specific medical order item and the treatment item.

[0029] Preferably, the construction of the correspondence between the specific medical order items and the treatment items in the treatment plan based on the knowledge graph includes:

[0030] Using the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph, the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and treatment items are identified and unified; the treatment items are treatment items in the treatment plan, and the treatment plan is obtained from the medical record;

[0031] By using the correspondence between drugs and drug ingredients in the knowledge graph, we can identify and match treatment items and specific medical order items that represent the same drug, and obtain the second diagnosis and treatment path;

[0032] Alternatively, the drug indications pre-stored in the knowledge graph can be used to identify and match treatment items and specific medical order items with the same drug indications to obtain a second diagnosis and treatment path.

[0033] Preferably, the step of constructing a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway includes:

[0034] According to the correspondence between the cause of the disease and the treatment items in the clinical manifestations in the first diagnosis and treatment pathway, as well as the correspondence between the abnormal examination and test results in the clinical manifestations and the treatment items, and the correspondence between the treatment items and the specific medical order items in the second diagnosis and treatment pathway is determined, the correspondence between the cause of the disease, the abnormal examination and test results, the specific medical order items and the treatment items are matched to obtain a full-process diagnosis and treatment pathway that includes the correspondence between clinical manifestations, treatment plans and medical order items.

[0035] A second aspect of an embodiment of the present invention discloses a processing system for a full-process diagnosis and treatment pathway, the processing system comprising:

[0036] The acquisition module is used to obtain all medical records of the patient during his / her stay in the hospital and sort the medical records by time;

[0037] A first extraction module is configured to extract the causes of disease, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes of disease from the sorted medical records based on a multimodal large language model, and to establish a correspondence between the causes of disease and the treatment items, as well as a correspondence between the abnormal examination and test results and the treatment items, to obtain a first diagnosis and treatment pathway;

[0038] The second extraction module is used to extract the medical order items of the corresponding time in the medical order form according to the chronological order of the sorted medical records, and obtain the specific medical order items within the target time range; construct the corresponding relationship between the specific medical order items and the treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway;

[0039] A construction module, configured to construct a full-process diagnosis and treatment pathway including correspondences between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway;

[0040] The storage module is used to store the full-process diagnosis and treatment path in a clinical experience database, in which the full-process diagnosis and treatment path of historical patients is stored.

[0041] Preferably, it also includes:

[0042] The recommendation module is used to obtain the clinical manifestations of a new patient; based on the clinical manifestations, the clinical experience database is searched, and treatment plans and medical orders corresponding to the clinical manifestations are retrieved and output.

[0043] A third aspect of an embodiment of the present invention discloses a computer storage medium having a program stored thereon. When the program is executed by a processor, the processing method of the full-process diagnosis and treatment pathway disclosed in the first aspect of the embodiment of the present invention is implemented.

[0044] The fourth aspect of an embodiment of the present invention discloses an electronic device, including a memory, a processor and a program stored on the memory, wherein the processor executes the program to implement the processing method of the full-process diagnosis and treatment pathway disclosed in the first aspect of the embodiment of the present invention.

[0045] Based on the processing method, system, storage medium and electronic device of the full-process diagnosis and treatment path provided by the above-mentioned embodiments of the present invention, by adopting at least two different methods to respectively construct the correspondence between the clinical manifestations and treatment items in the medical record, and to construct the correspondence between the treatment items in the medical record and the medical order items in the medical order form, the correspondence between the clinical manifestations, treatment plans and medical order items is obtained, and based on this, a full-process diagnosis and treatment path including "clinical manifestations-treatment plans-medical order items" is obtained, which completely restores the complete diagnosis and treatment path from symptoms to treatment execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0047] Figure 1 A schematic diagram of a processing architecture for a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention;

[0048] Figure 2 A flowchart of a method for processing a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention;

[0049] Figure 3 A flowchart of another method for processing a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention;

[0050] Figure 4 A schematic diagram of an inspection report disclosed in an embodiment of the present invention;

[0051] Figure 5 An image diagram disclosed in an embodiment of the present invention;

[0052] Figure 6 A schematic diagram of the structure of a processing system for a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention;

[0053] Figure 7 A schematic diagram of an electronic device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] In the present invention, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. In the present invention, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further limitations, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0056] The following are the technical terms involved in the embodiments of the present invention:

[0057] Clinical manifestations: consist of multiple causes such as the patient's current symptoms, signs, diseases, abnormal examination results, abnormal test results, etc.

[0058] Treatment plan: It consists of multiple treatment items such as medications, examination items, tests, surgery, etc. prescribed by the doctor based on the current clinical manifestations.

[0059] Medical order items: The medical order form consists of multiple specific medical order items such as drug treatment, examination form, test form, surgical plan, etc.

[0060] As we can see from the background, the data generated in actual clinical work is either difficult to extract effective information from or is isolated and lacks correlation, making it difficult to restore the complete diagnosis and treatment process from a multimodal perspective. Without a complete diagnosis and treatment pathway, it is impossible to restore the doctor's complete diagnosis and treatment path from symptoms to treatment execution, further limiting the subsequent doctors' desire to refer to treatments for similar or identical conditions, and hindering the development of intelligent healthcare and auxiliary medical care.

[0061] Therefore, the present invention discloses a processing solution for a full-process diagnosis and treatment path, which is mainly suitable for application scenarios of improving medical electronicization, intelligent treatment and auxiliary medicine. By adopting at least two different methods to respectively construct the correspondence between clinical manifestations and treatment items in the medical record, and to construct the correspondence between treatment items in the medical record and medical order items in the medical order form, the correspondence between clinical manifestations, treatment plans and medical order items is obtained, and based on this, a full-process diagnosis and treatment path including "clinical manifestations-treatment plans-medical order items" is obtained, which completely restores the complete diagnosis and treatment path from symptoms to treatment execution. Furthermore, the full-process diagnosis and treatment path can be used to recommend treatment plans for similar or identical diseases.

[0062] like Figure 1 As shown, it is a schematic diagram of the processing architecture of a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention, which mainly includes: an initialization module 10, a multimodal large language model 11, a multi-granularity matching module 12, a mining module 13 and a clinical experience library 14.

[0063] In real clinical work, starting from the time the patient is admitted to the hospital, the doctor will dynamically adjust the treatment plan based on the patient's current clinical manifestations, current condition, real-time examination and test results and other clinical information during each ward round. The treatment process will be recorded in the medical record in free text (a medical record is generated for each ward round, and each medical record includes the recording time) until the patient recovers and is discharged. Finally, a collection of medical records for the patient is formed and stored in the medical record management system.

[0064] For example, the clinical manifestations and treatment plans recorded in the medical record in free text are shown in Table 1 below:

[0065] Table 1:

[0066]

[0067] The initialization module 10 is mainly used to obtain all medical records of the patient from admission to discharge from the historical electronic medical record data, and sort all medical records according to the recording time of the medical records.

[0068] The multimodal large language model 11 is mainly used to identify the causes of disease, abnormal examination and test results, and treatment items corresponding to the causes in all sorted medical records, and extract all causes of disease, abnormal examination and test results, and treatment items corresponding to each cause of disease, and construct the correspondence between the causes of disease, abnormal examination and test results, and treatment items, and obtain the first diagnosis and treatment path that includes the correspondence between the causes of disease and treatment items, and the correspondence between the abnormal examination and test results and treatment items.

[0069] Typically, causes include, but are not limited to, symptoms, signs, and diseases (in special cases, these diseases also include familial genetic diseases, etc.). In this invention, because the data source and format of abnormal examination and test results differ from those of symptoms, signs, and diseases, the extraction process focuses on extracting the causes that include symptoms, signs, and diseases, while extracting the abnormal examination and test results separately. Symptoms, signs, and diseases are derived from the text information recorded during medical records and are in text format; examination and test results (including abnormal examination and test results) are derived from LIS (Laboratory Information System) and RIS (Radiological Information System), typically in report and image format.

[0070] The treatment items include but are not limited to medications, examinations, tests and surgeries.

[0071] This multimodal large language model 11 is pre-built, and the data used to construct it is derived from multi-source heterogeneous data in the medical field. This ensures that the multimodal large language model 11 has comprehensive medical background knowledge, diverse medical terminology, and expressions in actual clinical scenarios and different contexts. This multi-source heterogeneous data includes, but is not limited to: medical records (such as medical records, diagnostic summaries, treatment plans, etc.), medical images (such as X-rays, CT scans, MRI scans and their diagnostic reports), test reports (such as blood test and biochemical analysis results), medical literature (authoritative journal articles and textbooks), and medical forum discussions (case analysis and question-and-answer records of doctors' exchanges).

[0072] The specific construction process is as follows: first, using existing general multimodal large language models (such as GPT, Transformer and other architectures) and the multi-source heterogeneous data in the above-mentioned medical field for training, an initial multimodal large language model is obtained that adapts to the inherent correlation of professional terminology, logical reasoning and multimodal data in the medical context; then, based on improving the model's understanding depth of medical field knowledge, enhancing its ability to capture the semantics of medical context, and improving the accuracy of the initial multimodal large language model in tasks such as medical text generation, medical image understanding and diagnosis and treatment relationship inference, the output results of the initial multimodal large language model are adjusted to obtain the multimodal large language model required by the present invention, and the obtained multimodal large language model is configured to configure the automated labeling task, which is:

[0073] All causes of disease, abnormal examination and test results, and treatment items corresponding to each cause of disease are extracted, and the corresponding relationship between the cause of disease and the treatment items, as well as the corresponding relationship between the abnormal examination and test results and the treatment items are constructed.

[0074] Among them, the process of adjusting the output results of the initial multimodal large language model is: input the sample course data into the initial multimodal large language model, extract all causes of disease (symptoms, signs, diseases), and abnormal examination and test results for the input sample course data, as well as the treatment items corresponding to each cause of disease (drugs, examination items, test items, surgery), and the treatment items corresponding to the abnormal examination and test structure, and distinguish which are previous treatments and which are current treatment plans. Then, after continuous training and adjustment, the correspondence between the cause of disease and the treatment items, as well as the correspondence between the abnormal examination and test results and the treatment items are output.

[0075] The multi-granularity matching module 12 is mainly used to perform coarse-grained matching and fine-grained matching on the course of disease data and the medical order form, and use the matching results to construct the correspondence between the treatment items and the specific medical order items to obtain the second diagnosis and treatment path.

[0076] Coarse-grained matching involves extracting medical order items from medical order sheets within the same time period based on the timestamps of the medical records. The extracted order items include, but are not limited to, data on medications, surgeries, examinations, and tests.

[0077] It should be noted that the start time of execution of the medical order is less than or equal to the time of medical record, and the end time of execution of the medical order is greater than or equal to the time of medical record, so as to ensure the time consistency between the medical record and the medical order item, and form a preliminary time range association between the medical record and the medical order item.

[0078] Fine-grained matching involves building a correspondence between treatment items and specific medical orders in a treatment plan based on the knowledge graph. The treatment plan is derived from the corresponding medical records and is obtained from the medical records.

[0079] In the present invention, coarse-grained matching in the multi-granularity matching module 12 can quickly screen all potentially relevant medical order items within the target time range. Based on the medical order items screened by the coarse-grained matching, further fine-grained matching is performed to accurately determine the correlation between treatment items and specific medical order items.

[0080] The mining module 13 is mainly used to mine the full-process diagnosis and treatment path including "clinical manifestations-treatment plan-medical order items" based on the corresponding relationships output by the multimodal large language model 11 and the multi-granularity matching module 12.

[0081] The clinical experience database 14 is mainly used to store the full-process diagnosis and treatment pathways obtained by mining the historical electronic medical record data.

[0082] In the present invention, at least two different methods are used to extract key information from different data sources, and the correspondence between clinical manifestations and treatment items in the medical record is constructed respectively, as is the correspondence between treatment items in the medical record and medical order items in the medical order form. Then, the correspondence between clinical manifestations, treatment plans and medical order items is obtained, and based on this, causal relationships are mined to obtain a full-process diagnosis and treatment path including "clinical manifestations-treatment plans-medical order items", thereby achieving the purpose of completely restoring the complete diagnosis and treatment path from symptoms to treatment execution.

[0083] Furthermore, by adding a treatment plan recommendation module to the processing architecture of this full-process diagnosis and treatment pathway, a treatment plan recommendation system architecture can be constructed. This treatment plan recommendation module communicates with the clinical experience library 14. Specifically, when obtaining the clinical manifestations of a new patient, it queries the clinical experience library 14 based on the clinical manifestations, retrieves treatment plans and medical orders that correspond to the clinical manifestations, and recommends them to the doctor. In the present invention, the full-process diagnosis and treatment pathway can be used to recommend treatment plans for similar or identical conditions. At the same time, it provides data support for subsequent intelligent medical care and auxiliary medical care.

[0084] like Figure 2 As shown, it is a flowchart of a method for processing a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention. The processing method can be applied to the processing architecture of the full-process diagnosis and treatment pathway disclosed in the above embodiment of the present invention. The processing method includes:

[0085] S201: Obtain all medical records of the patient during his / her hospital stay and sort the medical records by time.

[0086] In the specific process of executing S201 , all medical records of the patient during the hospital stay are obtained through historical electronic medical record data, and each medical record is arranged in the order of the recording time of each medical record.

[0087] In S201, the medical record includes but is not limited to clinical manifestations and treatment plans.

[0088] Clinical manifestations include but are not limited to symptoms, signs, causes of diseases, and abnormal test results, such as pain, fever, abnormal heart rate, skin diseases, hypertension, diabetes, and abnormal test results such as elevated white blood cells in routine blood tests and lesions suggested by imaging.

[0089] The treatment plan includes, but is not limited to, prescribed medications, examination items, tests, surgeries, and other treatment items, such as antibiotics, antihypertensive drugs, examination items such as cranial MRI, arterial ultrasound, blood test results, urine test results, coagulation tests, and surgeries such as lung resection.

[0090] After executing S201 , a first diagnosis and treatment pathway extraction operation and a second diagnosis and treatment pathway extraction operation are respectively performed based on the sorted medical records.

[0091] The first diagnosis and treatment pathway extraction operation includes S202.

[0092] S202: Based on the multimodal large language model, the causes of clinical manifestations, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes are extracted from the medical records, and the correspondence between the causes and treatment items, as well as the correspondence between the abnormal examination and test results and the treatment items are constructed to obtain the first diagnosis and treatment path.

[0093] The multimodal large language model used in S202 is consistent with the multimodal large language model 11 in the processing architecture of the above-mentioned full-process diagnosis and treatment path. By taking the sorted medical records as the input of the multimodal large language model, the multimodal large language model is used for extraction and construction processing, and finally outputs the first diagnosis and treatment path that includes the correspondence between the cause of the disease and the treatment items, as well as the correspondence between the abnormal examination test results and the treatment items.

[0094] That is to say, the execution of the above S202 based on the multimodal large model mainly obtains the correspondence between the cause of disease and the treatment item in the medical record.

[0095] In S202, the first diagnosis and treatment pathway displays specific content in text form.

[0096] The second diagnosis and treatment pathway extraction operation includes S203 to S204.

[0097] S203: Extracting medical order items of corresponding time in the medical order form according to the time sequence of the medical course record to obtain specific medical order items within the target time range.

[0098] The specific execution process of S203 can refer to the process of performing coarse-grained matching by the multi-granularity matching module 12 in the processing architecture of the above-mentioned full-process diagnosis and treatment path.

[0099] S204: Construct a correspondence between specific medical order items and treatment items based on the knowledge graph to obtain a second diagnosis and treatment path.

[0100] The specific execution process of S204 can refer to the process of performing fine-grained matching by the multi-granularity matching module 12 in the processing architecture of the above-mentioned full-process diagnosis and treatment path.

[0101] That is to say, executing the above S203 to S204 based on the knowledge graph obtains the correspondence between the treatment items and the specific medical order items.

[0102] S205: Construct a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway.

[0103] In the specific execution of S205, based on the correspondence between the cause of disease and the treatment item in the clinical manifestation in the first diagnosis and treatment pathway, as well as the correspondence between the abnormal examination and test results in the clinical manifestation and the treatment item, and the correspondence between the treatment item and the specific medical order item in the second diagnosis and treatment pathway, the correspondence between the cause of disease, the abnormal examination and test results, the specific medical order item and the treatment item is matched to obtain a full-process diagnosis and treatment pathway including the correspondence between clinical manifestation, treatment plan and medical order item.

[0104] That is to say, by executing S205 and combining the correspondence between the cause of disease and the treatment item, as well as the correspondence between the treatment item and the specific medical order item, a complete treatment path of "cause of disease -> treatment item -> specific medical order item" (clinical manifestation - treatment plan - medical order item) can be obtained.

[0105] It should be noted that the correspondence between clinical manifestations, treatment plans, and medical order items, i.e., the matching relationship, can be displayed in a table. As shown in Table 2 below, it shows the corresponding relationship between clinical manifestations in the medical record, treatment plans in the medical record, and medical order items in the medical order after matching:

[0106] Table 2:

[0107]

[0108] S206: Store the entire diagnosis and treatment pathway in the clinical experience database.

[0109] In S206, the clinical experience database stores the full-process diagnosis and treatment pathways of historical patients.

[0110] In the embodiments disclosed in the present invention, at least two different methods are used to extract key information from different data sources, and the correspondence between clinical manifestations and treatment items in the medical record, as well as the correspondence between treatment items in the medical record and medical order items in the medical order form are constructed respectively, thereby obtaining the correspondence between clinical manifestations, treatment plans and medical order items. By integrating multimodal information such as text, images, and test reports and mining causal relationships, a full-process diagnosis and treatment path including "clinical manifestations-treatment plans-medical order items" is obtained, thereby achieving the purpose of completely restoring the complete diagnosis and treatment path from symptoms to treatment execution.

[0111] like Figure 3 As shown, this is a flowchart of another method for processing a full-process diagnosis and treatment pathway disclosed in an embodiment of the present invention. This processing method is also applicable to the processing architecture of the full-process diagnosis and treatment pathway disclosed in the above embodiment of the present invention. This processing method includes:

[0112] S301: Obtain all medical records of the patient during his / her hospital stay and sort the medical records by time.

[0113] S302: Based on the multimodal large language model, the cause of the disease, the abnormal examination and test results, and the treatment items in the treatment plan corresponding to the cause of the disease are extracted from the medical record, and the correspondence between the cause of the disease and the treatment items, as well as the correspondence between the abnormal examination and test results and the treatment items are constructed to obtain the first diagnosis and treatment path.

[0114] S303: Extracting the medical order items of the corresponding time in the medical order form according to the time sequence of the medical course record to obtain the specific medical order items within the target time range.

[0115] S304: Construct a correspondence between specific medical order items and treatment items based on the knowledge graph to obtain a second diagnosis and treatment path.

[0116] S305: Construct a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway.

[0117] S306: Store the entire diagnosis and treatment pathway in the clinical experience database.

[0118] The specific execution process of the above S301 to S306 is the same as the above Figure 2 The specific execution process of S201 to S206 disclosed in is the same as that of S201 to S206, which will not be repeated here.

[0119] S307: Obtain the clinical manifestations of new patients and analyze them.

[0120] In the specific process of executing S307 , the clinical manifestations of the new patient obtained are analyzed to obtain the causes of the clinical manifestations and abnormal examination results.

[0121] S308: Based on the clinical manifestations, the clinical experience database is queried, and treatment plans and medical orders corresponding to the clinical manifestations are retrieved and output.

[0122] During the specific execution of S308, by querying the correspondence between the clinical manifestations, treatment plans and medical orders stored in the clinical experience database, the cause of the clinical manifestations of the new patient, as well as the treatment plan and medical order corresponding to the abnormal examination results can be queried, and the three can be matched, and the matched results can be output, that is, recommended to the doctor.

[0123] If the new patient's clinical manifestation is hypertension, the treatment plan matched with Table 2 is: regular blood pressure checks, and the doctor's order item is: Check: [{'2015-05-25 10:37:26': '24-hour ambulatory blood pressure monitoring'}].

[0124] In the processing method of the full-process diagnosis and treatment path disclosed in the embodiment of the present invention, based on the mining and complete restoration of the full-process diagnosis and treatment path from symptoms to treatment execution including "clinical manifestations-treatment plan-medical order items", it can provide data support for subsequent intelligent medical care and auxiliary medical care, and at the same time, it can realize the recommendation of treatment plans for similar or identical diseases.

[0125] In one embodiment of the present invention, based on the above Figure 2 and Figure 3 The disclosed method for processing a full-process diagnosis and treatment pathway includes the following steps: S202 and S203 extract the cause of disease, abnormal examination and test results, and treatment items in the treatment plan corresponding to the cause of disease from the medical record based on a multimodal large language model, and establish a correspondence between the cause of disease and the treatment item, as well as a correspondence between the abnormal examination and test results and the treatment item. The specific process of obtaining the first diagnosis and treatment pathway includes the following steps:

[0126] S11: Input the first free text recording clinical manifestations, the second free text recording treatment plans, and the examination and test sheet recording examination and test results in the medical record into the multimodal large language model.

[0127] In S11 , the inspection and test form includes an inspection report form, a test report form and / or an image.

[0128] For example, the clinical manifestations and treatment plans in the free text medical records are shown in Table 1. The test report is shown in Figure 4 As shown, the input image is as follows Figure 5 shown.

[0129] S12: Use a multimodal large language model to extract the first key entity corresponding to the clinical manifestation and the second key entity corresponding to the treatment plan.

[0130] In S12, the first key entity includes at least the cause of the clinical manifestation and the abnormal examination and test results; the second key entity includes at least the drugs, surgical treatments, examination items and test items related to the treatment.

[0131] In the specific process of executing S12, the multimodal large language model is used to identify key entities, namely the first key entity and the second key entity, from the input first free text, the second free text, and the third free text.

[0132] S13: According to the first preset correspondence between the cause of disease and the treatment item, construct the correspondence between the cause of disease and the treatment item in the first key entity and the second key entity.

[0133] In S13, the first preset correspondence between the cause of disease and the treatment item is such as fever corresponding to anti-infection treatment; chest pain corresponding to analgesics and coronary angiography.

[0134] S14: According to the second preset correspondence between the abnormality inspection result and the treatment item, a correspondence between the abnormality inspection result and the treatment item in the first key entity and the second key entity is constructed.

[0135] In S14, a second preset correspondence between abnormal examination results and treatment items is established, such as a positive blood culture corresponding to the use of antibiotics; an image indicating a mass corresponding to resection surgery.

[0136] S15: Determine the first diagnosis and treatment path based on the correspondence between the cause of the disease and the treatment items, and the correspondence between the abnormal examination and test results and the treatment items.

[0137] During the specific execution of S15, the multimodal large language model can output a first diagnosis and treatment path based on the correspondence between the cause of the disease and the treatment item, and the correspondence between the abnormal examination test results and the treatment item.

[0138] In S15, the output first diagnosis and treatment pathway can be displayed in text form. Figure 4 and Figure 5 The first diagnosis and treatment path with as input and specific output is as follows.

[0139] Output:

[0140]

[0141] In the embodiment of the present invention, the multimodal large language model technology that has been trained and adjusted can simultaneously process information from different data sources, such as text descriptions in medical records, medical images, examination and test reports, etc. Based on this, a first diagnosis and treatment path can be obtained, which includes the correspondence between the cause of the disease and the treatment items, as well as the correspondence between abnormal examination and test results and the treatment items.

[0142] In one embodiment of the present invention, based on the above Figure 2 and Figure 3 The disclosed method for processing a full-process diagnosis and treatment pathway includes the following steps: S204 and S304 construct a correspondence between specific medical order items and treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway:

[0143] S21: Utilize the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph to identify and unify the actually identical drug names, examination and test names, and surgery names in specific medical order items and treatment items.

[0144] In S21 , the treatment item is a treatment item in a treatment plan, and the treatment plan is derived from a corresponding medical record and is obtained from the medical record.

[0145] The pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph represent entities that may have different names or abbreviations in different contexts. By identifying and processing these alias relationships, we ensure that the same drug, examination, surgery, or test in medical records and orders is accurately mapped.

[0146] Drug aliases, such as "Plavix Tablets" and "Plavix Tablets," actually refer to the same drug, but have different names or spellings. By identifying these aliases, unified processing can be achieved to avoid information loss.

[0147] Examination and test aliases: For example, "blood routine" and "complete blood count" may refer to the same examination. Through the examination and test alias relationship in the knowledge graph, they can be correctly matched to the corresponding medical order items.

[0148] Surgical alias: Some surgical names may have multiple expressions. For example, "lung cancer resection" and "lung cancer surgical resection" should be considered as the same surgery, and accurate matching can be achieved through mapping of surgical alias relationships.

[0149] It should be noted that the alias relationships in the knowledge graph are not limited to drug alias relationships, examination and inspection alias relationships, and surgery alias relationships. Relevant alias relationships can be added based on medical needs.

[0150] S22: For the drugs, examinations and tests, and surgeries with unified names in specific medical orders and treatment items, the first-level relationships between the medical information pre-constructed in the knowledge graph are used to determine the second-level relationships between the drugs, examinations and tests, and surgeries with unified names.

[0151] In the specific execution of S22, the first-level relationship pre-constructed in the knowledge graph can be used to achieve accurate correspondence between the treatment items (drugs, examinations, tests, surgery) in the treatment plan and the specific medical order items (drugs, examinations, tests, surgery).

[0152] For example, a treatment plan includes antibiotics, and a specific medical order includes aspirin. In the first-level relationship in the knowledge graph, antibiotics are a broad category of drugs, and amoxicillin is a specific drug within that category. This first-level relationship automatically identifies the corresponding relationship between the antibiotics in the treatment plan and the amoxicillin in the specific medical order.

[0153] S23: Constructing a correspondence between specific medical order items and treatment items based on the second-level relationship, and determining a second diagnosis and treatment path using the correspondence between the specific medical order items and treatment items.

[0154] In an embodiment of the present invention, the correspondence between each specific medical order item and the treatment item confirmed through the hierarchical relationship in the knowledge graph can accurately identify which treatments correspond to which medical order items, and ultimately obtain a second diagnosis and treatment path containing the correspondence between the specific medical order items and the treatment items.

[0155] In one embodiment of the present invention, based on the above Figure 2 and Figure 3 The disclosed method for processing a full-process diagnosis and treatment pathway includes the following steps: S204 and S304 construct a correspondence between specific medical order items and treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway:

[0156] S31: Utilize the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph to identify and unify the actually identical drug names, examination and test names, and surgery names in specific medical order items and treatment items.

[0157] In S31 , the treatment item is a treatment item in a treatment plan, and the treatment plan is derived from a corresponding medical record and is obtained from the medical record.

[0158] The specific execution process of S31 is the same as that of the above S21, and will not be repeated here.

[0159] S32: Using the correspondence between drugs and drug ingredients in the knowledge graph, identify and match treatment items and specific medical order items that represent the same drug to obtain a second diagnosis and treatment path.

[0160] In S32, the knowledge graph is provided with a correspondence between drugs and drug ingredients. During the specific execution of S32, the correspondence between drugs and drug ingredients can be used to identify and match treatment items and specific medical order items that represent the same drug, thereby obtaining a second diagnosis and treatment path.

[0161] For example, the knowledge graph contains a corresponding relationship between the drug "Amoxicillin" and its ingredient "Amoxicillin trihydrate." During recognition, the knowledge graph can identify the amoxicillin trihydrate mentioned in the treatment plan and match it with the correct medical order, such as the specific dosage and usage of amoxicillin in the specific medical order.

[0162] In an embodiment of the present invention, through the correspondence between drugs and drug ingredients in the knowledge graph, the drug indicated by the finished drug can be accurately identified and matched with the medical order containing the drug in the specific medical order item, and finally a second diagnosis and treatment path containing the correspondence between the specific medical order item and the treatment item is obtained.

[0163] In one embodiment of the present invention, based on the above Figure 2 and Figure 3The disclosed method for processing a full-process diagnosis and treatment pathway includes the following steps: S204 and S304 construct a correspondence between specific medical order items and treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway:

[0164] S41: Utilize the pre-constructed drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph to identify and unify the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and the treatment items.

[0165] In S41 , the treatment item is a treatment item in a treatment plan, and the treatment plan is derived from a corresponding medical record and is obtained from the medical record.

[0166] The specific execution process of S41 is the same as that of the above S21 and will not be repeated here.

[0167] S42: Using the drug indications pre-stored in the knowledge graph, identify and match treatment items and specific medical order items with the same drug indications to obtain a second diagnosis and treatment path.

[0168] Because different drugs usually have specific indications, and these indications are directly related to the doctor's treatment plan decision, drug indications are pre-stored in the knowledge graph; drug indications belong to the knowledge of drug entities in the knowledge graph, and are mainly used to associate the relationship between the drugs in the medical order items and the causes of the disease in the treatment plan.

[0169] In the specific process of executing S42, the adaptability of the drugs can be used to identify which medical order items are related to which causes of disease, and which treatment methods are related to those causes of disease. Based on the same cause of disease, the medical order items and treatment methods are matched.

[0170] In an embodiment of the present invention, through the drug adaptability in the knowledge graph, it is possible to accurately identify which specific medical order items or treatment items are related to which causes of disease, and based on this, match the specific medical order items and treatment items, and finally obtain a second diagnosis and treatment path containing the correspondence between the specific medical order items and the treatment items.

[0171] Because some drugs in the medical records and the drugs in the medical orders can be matched through the drug hierarchical relationship, and some drugs cannot be matched through the hierarchical relationship, but can be matched through the drug ingredients, therefore, the above multiple embodiments of constructing the correspondence between the specific medical order items and the treatment items in the treatment plan based on the knowledge graph can be executed independently or in parallel.

[0172] Based on the processing method of the full-process diagnosis and treatment pathway disclosed in the above embodiment of the present invention, Figure 6As shown, an embodiment of the present invention further discloses a processing system for a full-process diagnosis and treatment pathway, which includes: an acquisition module 601, a first extraction module 602, a second extraction module 603, a construction module 604 and a storage module 605.

[0173] The acquisition module 601 is used to acquire all medical records of the patient during his / her stay in the hospital and sort the medical records by time;

[0174] A first extraction module 602 is configured to extract, based on a multimodal large language model, the causes of clinical manifestations, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes of the disease from the sorted medical records, and to establish a correspondence between the causes of the disease and the treatment items, as well as between the abnormal examination and test results and the treatment items, to obtain a first diagnosis and treatment pathway;

[0175] The second extraction module 603 is used to extract the medical order items of the corresponding time in the medical order form according to the chronological order of the sorted medical records to obtain the specific medical order items within the target time range; and to construct the corresponding relationship between the specific medical order items and the treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway;

[0176] A construction module 604 is configured to construct a full-process diagnosis and treatment pathway including correspondences between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway;

[0177] The storage module 605 is used to store the full-process diagnosis and treatment pathway in a clinical experience database, where the full-process diagnosis and treatment pathway of historical patients is stored.

[0178] In one embodiment of the present invention, the processing system further includes:

[0179] The recommendation module is used to obtain the clinical manifestations of new patients; based on the clinical manifestations, it queries the clinical experience database, retrieves the treatment plans and medical orders that correspond to the clinical manifestations, and outputs them.

[0180] In one embodiment of the present invention, the first extraction module 602 is specifically configured to:

[0181] The first free text recording clinical manifestations, the second free text recording treatment plans, and the inspection and test sheets recording inspection and test results in the medical records are input into the multimodal large language model, where the inspection and test sheets include inspection reports, test reports and / or imaging images; the first key entity corresponding to the clinical manifestations and the second key entity corresponding to the treatment plan are extracted using the multimodal large language model, where the first key entity at least includes the cause, symptoms and abnormal inspection and test results used to characterize the clinical manifestations, and the second key entity at least includes drugs, surgical treatments, inspection items and test items related to the treatment; based on the first preset correspondence between the cause and the treatment item, the correspondence between the cause and the treatment item in the first key entity and the second key entity is constructed; based on the second preset correspondence between the abnormal inspection and test results and the treatment item, the correspondence between the abnormal inspection and test results and the treatment item in the first key entity and the second key entity is constructed; based on the correspondence between the cause and the treatment item, and the correspondence between the abnormal inspection and test results and the treatment item, the first diagnosis and treatment path is determined.

[0182] In one embodiment of the present invention, a correspondence between the specific medical order items and the treatment items is constructed based on the knowledge graph to obtain a second extraction module 603 of the second diagnosis and treatment pathway, which is specifically used to:

[0183] Utilize the pre-constructed drug alias relationship, examination and test alias relationship, and surgery alias relationship in the knowledge graph to identify and unify the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and treatment items; the treatment items are the treatment items in the treatment plan, and the treatment plan is obtained from the corresponding medical record; for the drugs, examinations and tests, and surgeries with unified names in the specific medical order items and treatment items, utilize the first-level relationship between the medical information pre-constructed in the knowledge graph to determine the second-level relationship between the drugs, examinations and tests, and surgeries with unified names; construct the corresponding relationship between the specific medical order items and the treatment items based on the second-level relationship, and use the corresponding relationship between the specific medical order items and the treatment items to determine the second diagnosis and treatment path.

[0184] In one embodiment of the present invention, a correspondence between specific medical order items and treatment items is constructed based on the knowledge graph to obtain a second extraction module 603 of the second diagnosis and treatment pathway, which is specifically used to:

[0185] Using the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph, we identify and unify the same drug names, examination and test names, and surgery names in specific medical order items and treatment items; treatment items are the treatment items in the treatment plan in the medical record; using the correspondence between drugs and drug ingredients in the knowledge graph, we identify and match treatment items and specific medical order items that represent the same drug to obtain the second diagnosis and treatment path;

[0186] In one embodiment of the present invention, a correspondence between specific medical order items and treatment items is constructed based on the knowledge graph to obtain a second extraction module 603 of the second diagnosis and treatment pathway, which is specifically used to:

[0187] Utilize the pre-built drug alias relationships, examination and test alias relationships, and surgical alias relationships in the knowledge graph to identify and unify the actually identical drug names, examination and test names, and surgical names in specific medical order items and treatment items; the treatment items are the treatment items in the treatment plan, which are obtained from the corresponding medical records; utilize the pre-stored drug indications in the knowledge graph to identify and match treatment items and specific medical order items with the same drug indications to obtain the second diagnosis and treatment path.

[0188] In one embodiment of the present invention, the construction module 604 is specifically configured to:

[0189] According to the correspondence between symptoms and treatment items in the clinical manifestations in the first diagnosis and treatment pathway, as well as the correspondence between abnormal examination and test results in the clinical manifestations and treatment items, and the correspondence between treatment items and specific medical order items in the second diagnosis and treatment pathway, the correspondence between the causes of disease, abnormal examination and test results, specific medical order items and treatment items is matched to obtain a full-process diagnosis and treatment pathway that includes the correspondence between clinical manifestations, treatment plans and medical order items.

[0190] In the processing system of the full-process diagnosis and treatment pathway disclosed in the embodiment of the present invention, at least two different methods are used to respectively construct the correspondence between clinical manifestations and treatment items in the medical record, and to construct the correspondence between treatment items in the medical record and medical order items in the medical order form, thereby obtaining the correspondence between clinical manifestations, treatment plans, and medical order items. Based on this, a full-process diagnosis and treatment pathway including "clinical manifestations-treatment plans-medical order items" is obtained, completely restoring the complete diagnosis and treatment pathway from symptoms to treatment execution. Furthermore, the full-process diagnosis and treatment pathway can be used to recommend treatment plans for similar or identical conditions.

[0191] Based on the processing system for the full-process diagnosis and treatment pathway disclosed in the above-mentioned embodiment of the present disclosure, each of the above modules can be implemented by a hardware device composed of a processor and a memory. Specifically, each of the above modules is stored in the memory as a program unit, and the processor executes the program unit stored in the memory to implement thread control.

[0192] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and database expansion can be achieved by adjusting kernel parameters.

[0193] An embodiment of the present invention discloses a computer storage medium having a program stored thereon. When the program is executed by a processor, the processing method of the full-process diagnosis and treatment pathway disclosed in the above-mentioned embodiment of the present invention is implemented.

[0194] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology to store information. 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 RAM (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 technologies, 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 such as modulated data signals and carrier waves.

[0195] An embodiment of the present invention discloses an electronic device, including a memory, a processor, and a program stored in the memory. The processor executes the above program to implement the processing method of the full-process diagnosis and treatment path disclosed in the above embodiment of the present invention.

[0196] Specifically, such as Figure 7 FIG. 1 is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device 700 includes at least one processor 701 , at least one memory 702 connected to the processor, and a bus 703 .

[0197] The processor 701 and the memory 702 communicate with each other via the bus 703 .

[0198] The processor 701 is configured to execute the program stored in the memory.

[0199] The memory 702 is used to store a program, which is at least used to implement the processing method of the full-process diagnosis and treatment pathway disclosed in the above-mentioned embodiment of the present invention.

[0200] In a typical configuration, the device includes one or more processors (CPUs), memory, and a bus. The device may also include input / output interfaces, network interfaces, and the like.

[0201] Memory may include non-permanent memory in a computer-readable medium, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip. Memory is an example of a computer-readable medium.

[0202] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple. For relevant parts, refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0203] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0204] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for processing a full-process diagnosis and treatment pathway, characterized in that: The processing method comprises: Obtain all medical records of the patient during his / her stay in the hospital and sort them by time; Performing a first diagnosis and treatment pathway extraction operation and a second diagnosis and treatment pathway extraction operation based on the sorted medical records; Constructing a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway; Storing the full-process diagnosis and treatment pathway in a clinical experience database, wherein the clinical experience database stores the full-process diagnosis and treatment pathways of historical patients; The first diagnosis and treatment pathway extraction operation includes: Extracting the causes of clinical manifestations, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes from the medical records based on the multimodal large language model, constructing a correspondence between the causes and treatment items, and a correspondence between the abnormal examination and test results and treatment items, to obtain a first diagnosis and treatment pathway; The second diagnosis and treatment pathway extraction operation includes: Extracting the medical order items of the corresponding time in the medical order form according to the chronological order of the medical course record to obtain the specific medical order items within the target time range; Based on the knowledge graph, a correspondence between the specific medical order items and the treatment items is constructed to obtain a second diagnosis and treatment path.

2. The processing method according to claim 1, characterized in that Also includes: Obtain clinical presentation of new patients; The clinical experience database is queried based on the clinical manifestations, and treatment plans and medical advice items corresponding to the clinical manifestations are retrieved and output.

3. The method according to claim 1, characterized in that The multimodal large language model is used to extract the cause of disease, abnormal examination and test results, and treatment items in the treatment plan corresponding to the cause of disease from the medical record, and to establish a correspondence between the cause of disease and the treatment items, as well as a correspondence between the abnormal examination and test results and the treatment items, to obtain a first diagnosis and treatment pathway, including: Inputting a first free text recording clinical manifestations, a second free text recording treatment plans, and an examination and test sheet recording examination and test results in the medical record into a multimodal large language model, wherein the examination and test sheet includes an examination report, a test report, and / or an imaging image; Extracting a first key entity corresponding to a clinical manifestation and a second key entity corresponding to a treatment plan using the multimodal large language model, wherein the first key entity includes at least the cause of the clinical manifestation and abnormal examination and test results, and the second key entity includes at least medications, surgical treatments, examination items, and test items related to the treatment; According to a first preset correspondence relationship between the cause of disease and the treatment item, constructing a correspondence relationship between the cause of disease and the treatment item in the first key entity and the second key entity; According to a second preset correspondence between the abnormality inspection and test results and the treatment items, constructing a correspondence between the abnormality inspection and test results and the treatment items in the first key entity and the second key entity; A first diagnosis and treatment pathway is determined based on the correspondence between the cause of the disease and the treatment item, and the correspondence between the abnormal examination result and the treatment item.

4. The method according to claim 1, wherein The corresponding relationship between the specific medical order items and the treatment items is constructed based on the knowledge graph to obtain the second diagnosis and treatment path, including: Using the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph, the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and treatment items are identified and unified; the treatment items are treatment items in the treatment plan, and the treatment plan is obtained from the medical record; For the drugs, examinations and tests, and surgeries with unified names in the specific medical order items and treatment items, the first-level relationships between the medical information pre-constructed in the knowledge graph are used to determine the second-level relationships between the drugs, examinations and tests, and surgeries with unified names; A correspondence between the specific medical order item and the treatment item is constructed based on the second hierarchical relationship, and a second diagnosis and treatment pathway is determined using the correspondence between the specific medical order item and the treatment item.

5. The method according to claim 1, wherein The corresponding relationship between the specific medical order item and the treatment item in the treatment plan is constructed based on the knowledge graph, including: Using the pre-built drug alias relationships, examination and test alias relationships, and surgery alias relationships in the knowledge graph, the drug names, examination and test names, and surgery names that are actually the same in the specific medical order items and treatment items are identified and unified; the treatment items are treatment items in the treatment plan, and the treatment plan is obtained from the medical record; By using the correspondence between drugs and drug ingredients in the knowledge graph, we can identify and match treatment items and specific medical order items that represent the same drug, and obtain the second diagnosis and treatment path; Alternatively, the drug indications pre-stored in the knowledge graph can be used to identify and match treatment items and specific medical order items with the same drug indications to obtain a second diagnosis and treatment path.

6. The method according to any one of claims 1 to 5, characterized in that The step of constructing a full-process diagnosis and treatment pathway including the correspondence between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway includes: According to the correspondence between the cause of the disease and the treatment items in the clinical manifestations in the first diagnosis and treatment pathway, as well as the correspondence between the abnormal examination and test results in the clinical manifestations and the treatment items, and the correspondence between the treatment items and the specific medical order items in the second diagnosis and treatment pathway is determined, the correspondence between the cause of the disease, the abnormal examination and test results, the specific medical order items and the treatment items are matched to obtain a full-process diagnosis and treatment pathway that includes the correspondence between clinical manifestations, treatment plans and medical order items.

7. A full-process diagnosis and treatment pathway processing system, characterized by: The processing system comprises: The acquisition module is used to obtain all medical records of the patient during his / her stay in the hospital and sort the medical records by time; A first extraction module is configured to extract the causes of disease, abnormal examination and test results, and treatment items in the treatment plan corresponding to the causes of disease from the sorted medical records based on a multimodal large language model, and to establish a correspondence between the causes of disease and the treatment items, as well as a correspondence between the abnormal examination and test results and the treatment items, to obtain a first diagnosis and treatment pathway; The second extraction module is used to extract the medical order items of the corresponding time in the medical order form according to the chronological order of the sorted medical records, and obtain the specific medical order items within the target time range; construct the corresponding relationship between the specific medical order items and the treatment items based on the knowledge graph to obtain the second diagnosis and treatment pathway; A construction module, configured to construct a full-process diagnosis and treatment pathway including correspondences between clinical manifestations, treatment plans, and medical order items based on the first diagnosis and treatment pathway and the second diagnosis and treatment pathway; The storage module is used to store the full-process diagnosis and treatment path in a clinical experience database, in which the full-process diagnosis and treatment path of historical patients is stored.

8. The processing system according to claim 7, characterized in that Also includes: Recommendation module, used to obtain clinical presentation of new patients; The clinical experience database is queried based on the clinical manifestations, and treatment plans and medical advice items corresponding to the clinical manifestations are retrieved and output.

9. A computer storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the processing method of the full-process diagnosis and treatment pathway as described in any one of claims 1 to 6 is implemented.

10. An electronic device comprising a memory, a processor, and a program stored in the memory, characterized in that: The processor executes the program to implement the processing method of the full-process diagnosis and treatment pathway as described in any one of claims 1 to 6.

Images