CN117079798A - Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model - Google Patents
Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model Download PDFInfo
- Publication number
- CN117079798A CN117079798A CN202310818964.6A CN202310818964A CN117079798A CN 117079798 A CN117079798 A CN 117079798A CN 202310818964 A CN202310818964 A CN 202310818964A CN 117079798 A CN117079798 A CN 117079798A
- Authority
- CN
- China
- Prior art keywords
- consultation
- result
- department
- history information
- patient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 70
- 238000011282 treatment Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 24
- 201000010099 disease Diseases 0.000 claims abstract description 89
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 89
- 238000004891 communication Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000010365 information processing Effects 0.000 claims abstract description 3
- 238000004422 calculation algorithm Methods 0.000 claims description 40
- 208000024891 symptom Diseases 0.000 claims description 19
- 239000003814 drug Substances 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 10
- 238000010801 machine learning Methods 0.000 claims description 7
- 238000012986 modification Methods 0.000 claims description 3
- 230000004048 modification Effects 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 230000000875 corresponding effect Effects 0.000 description 18
- 229940079593 drug Drugs 0.000 description 8
- 238000012549 training Methods 0.000 description 7
- 238000007689 inspection Methods 0.000 description 6
- 206010020751 Hypersensitivity Diseases 0.000 description 2
- 208000026935 allergic disease Diseases 0.000 description 2
- 230000007815 allergy Effects 0.000 description 2
- 230000036407 pain Effects 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002483 medication Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000008961 swelling Effects 0.000 description 1
- 238000011269 treatment regimen Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Pathology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention provides an intelligent pre-consultation and rapid diagnosis and treatment system and method based on a large language model, which relate to the technical field of medical consultation and comprise the following steps: searching similar historical medical history information in a historical case library based on the disease description result by utilizing a large language model to conduct pre-consultation communication with the current patient, and taking each historical medical history information and the disease description result as pre-consultation results; acquiring recommended registration departments according to the pre-consultation result, the descriptions of a plurality of departments associated with the departments and the information processing of a plurality of doctors of the departments, and feeding the recommended registration departments back to the current patient; performing diagnostic analysis on the pre-consultation result to obtain an analysis result, and then processing according to registration information submitted by the patient and the analysis result to obtain corresponding preliminary diagnosis suggestions comprising a plurality of suggestion examination items and feeding the preliminary diagnosis suggestions back to the current patient; the user is guided to pay after recommending a registration department and checking the item. The method has the advantages that the large language model is used for carrying out pre-consultation on the patient, recommending departments and examination items, and completing diagnosis and treatment processes after the recommendation is completed.
Description
Technical Field
The invention relates to the technical field of medical inquiry, in particular to an intelligent pre-inquiry and rapid diagnosis and treatment system and method based on a large language model.
Background
In the current medical system, patients need to go through complicated treatment procedures, including appointment consultation, registration payment, queuing check-in, doctor inquiry, examination payment, waiting for examination report and other links. Each link needs to be carried out by a single thread in sequence, which easily causes overlong queuing and waiting time, and slows down the patient treatment process and the doctor diagnosis time.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides an intelligent pre-inquiry and rapid diagnosis and treatment system based on a large language model, which is connected with a history case library of a hospital, wherein the history case library comprises a plurality of pieces of history information, and the intelligent pre-inquiry and rapid diagnosis and treatment system comprises:
the pre-consultation module is used for carrying out pre-consultation communication with the current patient by utilizing the large-scale language model to obtain a disease description result, then searching a plurality of similar historical medical history information in the historical case library based on the disease description result, and taking each historical medical history information and the disease description result as pre-consultation results;
the department recommendation module is connected with the pre-consultation module and is used for obtaining a department with highest matching degree according to a pre-consultation result, a plurality of department descriptions related to the department and a plurality of department doctor information processes stored in advance, and feeding the department with highest matching degree back to a current patient as a recommended registration department;
the examination suggestion module is respectively connected with the pre-consultation module and the department recommendation module and is used for analyzing the pre-consultation result to obtain an analysis result, and then processing the analysis result according to registration information submitted by the patient and the analysis result to obtain corresponding project suggestions containing a plurality of suggestion examination projects, and feeding the project suggestions back to the current patient;
the payment module is respectively connected with the department recommendation module and the examination suggestion module and is used for providing a registration payment interface to complete payment after the recommended department is fed back to the current patient and providing a project payment interface to complete payment when the project suggestion is fed back to the current patient.
Preferably, the pre-interrogation module comprises:
the inquiry unit is used for receiving the provided initial illness state description of the current patient, inquiring the current patient based on the initial illness state description by using the large-scale language model to obtain detailed illness state description fed back by the current patient, and obtaining illness state description results according to the initial illness state description and detailed illness state description processing;
the case matching unit is connected with the inquiry unit and is used for matching in the historical case information base according to the disease description result to obtain a plurality of pieces of historical medical history information similar to the disease description result, and integrating the historical medical history information with the disease description result to serve as a pre-inquiry result.
Preferably, the case matching unit includes:
the first matching subunit is used for extracting patient symptoms from the disease description result, respectively calculating corresponding symptom similarity with each piece of history information in the history case library through a matching algorithm, and obtaining a plurality of pieces of history information based on the symptom similarity to form a first similarity set;
the second matching subunit is used for extracting the historical disease of the patient from the disease description result, respectively calculating corresponding disease similarity through a matching algorithm and each piece of historical medical history information in the historical case library, and obtaining a plurality of pieces of historical medical history information based on the disease similarity to form a second similarity set;
the third matching subunit is used for extracting a patient medicine record from the disease description result, respectively calculating corresponding medicine similarity with each piece of history information in the history case library through a matching algorithm, and obtaining a plurality of pieces of history information based on the medicine similarity to form a third similarity set;
and the integration subunit is respectively connected with the first matching subunit, the second matching subunit and the third matching subunit and is used for integrating the history information and the illness state description result in the first similar set, the second similar set and the third similar set into a pre-consultation result.
Preferably, the department recommendation module includes:
the extraction unit is used for extracting keywords from the pre-consultation result by using the large-scale language model to obtain a disease keyword set, and extracting keywords from the pre-stored department descriptions and department doctor information associated with a plurality of departments by using the large-scale language model to obtain a corresponding department keyword set;
and the matching unit is connected with the extraction unit and is used for respectively calculating the matching degree between the disease keyword set and each department keyword set by using a matching algorithm, and then feeding back the department corresponding to the department keyword set with the highest matching degree to the current patient as a recommended registration department.
Preferably, the inspection suggestion module includes:
a storage unit for storing a plurality of departments, each department being associated with a plurality of examination items;
the inquiring unit is used for associating the historical medical history information with the pre-inquiry result when the historical medical history information of the current patient is inquired in the historical case library;
the analysis unit is connected with the query unit and is used for extracting keywords from the illness state description result and the related history information in the pre-consultation result by utilizing the large-scale language model to obtain a plurality of illness state keywords, analyzing the illness state keywords to obtain analysis results, screening out a plurality of recommended examination items from the examination items by utilizing the large-scale language model based on pre-stored medical knowledge, the analysis results and the recommended registration department, and feeding back the recommended examination items and the analysis results to the current patient as item suggestions.
Preferably, the inspection advice module further comprises:
the adjusting unit is connected with the analyzing unit and is used for displaying the pre-consultation result to the consultation doctor when the consultation doctor recommending the registering department receives the consultation, and feeding back the project suggestion to the current patient after adjusting each suggested examination project in the project suggestion according to the submitted modification opinion of the consultation doctor before feeding back the project suggestion to the current patient.
Preferably, the matching algorithm comprises one of a text similarity algorithm, a logical matching algorithm, a machine learning algorithm, and a probabilistic model.
Preferably, the case matching unit includes:
the case matching subunit is used for matching in the historical case information base according to the disease description result to obtain a plurality of pieces of historical medical history information similar to the disease description result;
the complete examination subunit is connected with the case matching subunit and is used for respectively extracting information from the illness state description result and the historical medical history information by using the large language model, taking the illness state description result and the historical medical history information as a pre-consultation result when the judgment information extraction result contains preset information, and prompting the current patient to supplement the illness state information when the judgment information extraction result does not contain the preset information.
The invention also provides an intelligent pre-inquiry and rapid diagnosis and treatment method based on a large language model, which is applied to the intelligent pre-inquiry and rapid diagnosis and treatment system, wherein the intelligent pre-inquiry and rapid diagnosis and treatment system is connected with a history case library of a hospital, the history case library comprises a plurality of pieces of history information, and the intelligent pre-inquiry and rapid diagnosis and treatment method comprises the following steps:
step S1, an intelligent pre-consultation and rapid diagnosis and treatment system performs pre-consultation communication with a current patient by using a large language model to obtain a disease description result, and then searches a historical case library based on the disease description result to obtain a plurality of similar historical medical history information, wherein each historical medical history information and the disease description result are used as pre-consultation results;
step S2, the intelligent pre-consultation and rapid diagnosis and treatment system obtains a department with highest matching degree as a recommended registration department according to a pre-consultation result and a plurality of department descriptions and a plurality of department doctor information which are stored in advance and are associated with the departments, and feeds the department with highest matching degree back to a current patient;
step S3, the intelligent pre-consultation and rapid diagnosis and treatment system analyzes the pre-consultation result to obtain an analysis result, and then processes and obtains corresponding project suggestions comprising a plurality of suggestion examination projects according to registration information submitted by a patient and the analysis result to feed back the project suggestions to the current patient;
and S4, the intelligent pre-consultation and rapid diagnosis and treatment system provides a registration payment interface to finish payment after feeding back the recommended department to the current patient, and provides a project payment interface to finish payment when feeding back the project suggestion to the current patient.
Preferably, step S1 includes:
step S11, the intelligent pre-consultation and rapid diagnosis and treatment system receives an initial illness state description provided by a current patient, then inquires the current patient based on the initial illness state description by utilizing a large-scale language model to obtain a detailed illness state description fed back by the current patient, and then obtains an illness state description result according to the initial illness state description and detailed illness state description processing;
step S12, the intelligent pre-inquiry and rapid diagnosis and treatment system is matched with the disease description result in the historical case information base to obtain a plurality of pieces of historical medical history information similar to the disease description result, and the pieces of historical medical history information are integrated with the disease description result to serve as pre-inquiry results.
The technical scheme has the following advantages or beneficial effects:
1) The patient is pre-diagnosed through the large language model to obtain the disease description result, so that the doctor's diagnosis time is saved;
2) Recommending registration departments to patients for registration according to the history information which is found in the history case library and is similar to the disease description result, so that queuing links are omitted, and registration time is saved;
3)
drawings
FIG. 1 is a schematic diagram of a large language model-based intelligent pre-consultation and rapid diagnosis and treatment system according to a preferred embodiment of the present invention;
FIG. 2 is a flow chart of an intelligent pre-consultation and rapid diagnosis and treatment method based on a large language model according to a preferred embodiment of the present invention;
fig. 3 is a schematic flow chart of step S1 in the preferred embodiment of the invention.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples. The present invention is not limited to the embodiment, and other embodiments may fall within the scope of the present invention as long as they conform to the gist of the present invention.
In a preferred embodiment of the present invention, based on the above-mentioned problems existing in the prior art, an intelligent pre-consultation and rapid diagnosis and treatment system based on a large language model is now provided, which is connected to a history case library of a hospital, the history case library includes a plurality of history information, and the intelligent pre-consultation and rapid diagnosis and treatment system includes:
the pre-consultation module 1 is used for carrying out pre-consultation communication with the current patient by utilizing a large language model to obtain a disease description result, then searching a plurality of similar historical medical history information in a historical case library based on the disease description result, and taking each historical medical history information and the disease description result as pre-consultation results;
the department recommendation module 2 is connected with the pre-consultation module 1 and is used for obtaining a department with highest matching degree according to a pre-consultation result, a plurality of department descriptions related to the department and a plurality of department doctors information processing, and feeding back the department with highest matching degree to a current patient as a recommendation registration department;
the examination suggestion module 3 is respectively connected with the pre-consultation module 1 and the department recommendation module 2 and is used for analyzing the pre-consultation result to obtain an analysis result, and then processing according to registration information submitted by a patient and the analysis result to obtain corresponding project suggestions containing a plurality of suggestion examination projects and feeding the project suggestions back to the current patient;
the payment module 4 is respectively connected with the department recommendation module 2 and the examination suggestion module, and the examination suggestion module 3 is used for providing a registration payment interface to complete payment after the recommended department is fed back to the current patient and providing an item payment interface to complete payment when the item suggestion is fed back to the current patient.
Specifically, in this embodiment, a large language model (for example, GPT-4) is used to perform a dialogue with a patient to perform a pre-consultation, and a plurality of similar history information is searched in a history case library according to a patient condition description structure to be used as a pre-consultation result, and then the patient is assisted to find a correct department registration according to the pre-consultation result, and the pre-consultation result of the patient is analyzed to obtain an analysis result which automatically matches with an examination item, so that after the patient receives a recommended registration room and a recommended examination item, the patient can also pay for the examination item directly, thereby greatly saving the patient's time and doctor's time. The application of the system can improve the diagnosis efficiency of patients, improve the diagnosis efficiency of doctors, improve the operation efficiency of the whole medical system and provide more convenient and efficient medical services for the patients and the doctors.
In a preferred embodiment of the present invention, as shown in fig. 1, the pre-inquiry module 1 includes:
the inquiry unit 11 is configured to receive an initial condition description provided by a current patient, then query the current patient based on the initial condition description by using a large language model to obtain a detailed condition description fed back by the current patient, and then obtain a condition description result according to the initial condition description and the detailed condition description;
the case matching unit 12 is connected to the inquiry unit 11, and is configured to match the disease description result in the historical case information base to obtain a plurality of pieces of historical medical history information similar to the disease description result, and integrate with the disease description result as a pre-inquiry result.
Specifically, in this embodiment, the patient is prompted to describe his own condition during the pre-consultation process, for example, the patient describes his own discomfort symptoms, pain, swelling and corresponding duration, but generally the patient will not be described in detail, and then the patient is further inquired based on the initial condition description provided by the patient by using a large language model, so as to learn about the degree of pain, accompanying symptoms, history of allergy, and other more detailed problems to more comprehensively understand the condition of the patient; the current large language model (for example, GPT-4) can be networked and has a great deal of medical knowledge, and has a strong logic capability to make a next question based on the patient's answer to get a detailed condition description.
In the preferred embodiment of the present invention, as shown in fig. 1, the case matching unit 12 includes:
a first matching subunit 121, configured to extract patient symptoms from the disease description result, calculate corresponding symptom similarities through a matching algorithm and each piece of history information in the history case library, and obtain a plurality of pieces of history information based on the symptom similarities to form a first similarity set;
the second matching subunit 122 is configured to extract a patient's historical disease from the disease description result, calculate corresponding disease similarity with each piece of historical medical history information in the historical case library through a matching algorithm, and obtain a plurality of pieces of historical medical history information based on the disease similarity to form a second similarity set;
a third matching subunit 123, configured to extract a patient drug record from the disease description result, calculate corresponding drug similarity with each piece of history information in the history case library through a matching algorithm, and obtain a plurality of pieces of history information based on the drug similarity to form a third similarity set;
and an integrating subunit 124, respectively connected to the first matching subunit 121, the second matching subunit 122, and the third matching subunit 123, for integrating the history information and the disease description result in the first similar set, the second similar set, and the third similar set into a pre-consultation result.
Specifically, in this embodiment, after the detailed disease description is obtained, the large language model compares and matches the collected detailed disease description with a history case library of a hospital to obtain a plurality of similar history information, and integrates and generates a pre-consultation case, and the matching process is divided into three parts:
symptom description alignment: patients will describe their symptoms when they are talking to a large language model. These symptom descriptions will be compared to symptom information in the hospital's historical case base. The comparison process may adopt a text similarity matching algorithm, and the similarity between symptom descriptions is calculated to determine the matching degree so as to obtain similar historical medical history information;
disease diagnosis and comparison: the patient may provide diagnostic information of the historical disease that has been obtained during the session, such as a previous doctor diagnosis or a disease that has been diagnosed. The large language model compares the diagnosis information of the historical diseases with the diagnosis records in the historical case library of the hospital to find similar or related historical medical history information;
drug record comparison: if patients provided medications or medication regimens they had previously used, the large language model would be aligned with each historical medical history information medication record in the hospital's historical case library. Thus, similar historical information can be found for the medication or treatment regimen used to be similar to that of the patient and provide a reference to the current patient;
finally, the three aspects are used for obtaining a plurality of similar historical case information and a disease description result as a pre-consultation result.
After registering the patient, the doctor who receives the doctor can see the similar history information included in the pre-consultation result for the doctor to diagnose and reference, and the pre-consultation result also provides the recommendation of the follow-up registering department and the project suggestion.
The large language model may also answer related questions if the patient is in question for a particular examination item, providing further explanation or understanding.
The doctor can review, confirm or modify project suggestions given by the large language model according to own clinical experience and combining with the illness state description result and analysis result listed by the large language model, the project suggestions are sent to the patient by the system after confirmation, meanwhile, the patient is reminded to enter a fee payment page for payment, and the examination of the project is reminded to be carried out after payment. After the patient is inspected, the system prompts the patient, the inspection result reports the time, and the patient can reasonably arrange the doctor face diagnosis check-in time according to the time condition of the result.
In a preferred embodiment of the present invention, as shown in fig. 1, the department recommendation module 2 includes:
an extracting unit 21, configured to extract keywords from the pre-consultation result by using a large language model to obtain a disease keyword set, and extract keywords from the pre-stored department descriptions and department doctor information associated with multiple departments by using the large language model to obtain a corresponding department keyword set;
department matching unit 22 is connected with extracting unit 21, and is used for calculating the matching degree between the disease keyword set and each department keyword set respectively by using a matching algorithm, and then feeding back the department corresponding to the department keyword set with the highest matching degree to the current patient as the recommended registration department.
Specifically, in this embodiment, after the pre-consultation result is obtained, the large language model will first advance the condition keywords therein to form a condition keyword set, and the large language model is preconfigured with abundant medical knowledge, and also has department descriptions of each department and information of the doctors belonging to the department, where the descriptions of the departments will have symptoms of the disease mainly treated; keyword extraction is also carried out on the department description and the department doctor information of each department to form a department keyword set, the matching degree between the disease keyword set and each department keyword set is calculated by using a matching algorithm, and the department with the highest matching degree is used as a recommended registration department to be fed back to a patient.
The system can automatically push the registration entrance of the department to the patient.
The patient clicks the entrance, selects to input own medical insurance information or doctor card information for registration, and the system can automatically link the medical insurance payment system of the hospital, so that the patient can finish on-line settlement in the system, and the registration time is saved.
In a preferred embodiment of the present invention, as shown in fig. 1, the inspection suggestion module 3 includes:
a storage unit 31 for storing a plurality of departments, each department being associated with a plurality of examination items;
a query unit 32 for associating the history information with the pre-consultation result when the history information of the current patient is queried in the history case library;
the analysis unit 33 is connected with the query unit 32 and the storage unit 31, and is configured to extract keywords from the disease description result and the associated history information in the pre-consultation result by using a large language model to obtain a plurality of disease keywords, analyze the disease keywords to obtain analysis results, and then screen out a plurality of recommended examination items from the examination items by using the large language model based on pre-saved medical knowledge, analysis results and recommended registration departments, and feed back each recommended examination item and analysis result as an item suggestion to the current patient.
Specifically, in this embodiment, in addition to this, the large language model uses the medical knowledge in the large language model and the information in the hospital database to match the patient's own medical history information.
If the patient has history case information recorded by history visit, the large language model can select keywords in the patient's illness state description according to the history case information of the patient in a hospital system and the illness state description of the patient, and the large language model automatically gives preliminary diagnosis suggestions (similar to preliminary judgment of doctors) by analyzing the extracted keywords: including predictions of possible diseases or conditions, suggested exam items, historical allergy information, etc., while notes related to the disease are noted in the preliminary diagnostic advice for convenient review by the physician and patient.
If the patient has no history visit record, the large language model can select keywords in the complaint condition description of the patient according to the condition description of the patient, and the large language model analyzes the extracted keywords according to internal medical knowledge and corresponding clinical departments of hospitals to give preliminary disease or condition diagnosis suggestions: including predictions of possible diseases or conditions, and suggested exam items.
The large language model (taking GPT-4 as an example) used in the embodiment is trained by aiming at training, and a training set is pre-constructed before training, wherein the training set comprises disease descriptions of a plurality of patients and diagnosis results given by corresponding doctors, and mass data training is carried out by taking the disease descriptions of the patients as input and the diagnosis results given by the corresponding doctors as output during training, so that accurate prediction results can be obtained.
And in order to maintain the accuracy and reliability of the system, the large language model requires regular knowledge updates and model iterations. This may include introducing new medical knowledge, learning up-to-date guidelines and disease patterns, etc.
In the whole system, it is important to protect the privacy and security of patient data. Necessary security measures, such as data encryption, authentication, access control, etc., are taken to ensure confidentiality and integrity of patient information.
In the preferred embodiment of the present invention, as shown in fig. 1, the inspection suggestion module 3 further includes:
the adjusting unit 34 is connected to the analyzing unit 32, and is configured to display the pre-consultation result to the consultation physician when the consultation physician in the registration department is recommended to take the consultation, and to feed back the item suggestion to the current patient after adjusting each suggested examination item in the item suggestion according to the submitted modification opinion of the consultation physician before feeding back the item suggestion to the current patient.
In a preferred embodiment of the present invention, the matching algorithm comprises one of a text similarity algorithm, a logical matching algorithm, a machine learning algorithm, and a probabilistic model.
Specifically, in the present embodiment, the first embodiment,
text similarity based algorithm: these algorithms determine the degree of matching between texts by comparing their similarities. Common text similarity measures include cosine similarity, jaccard similarity, edit distance, and the like. During the matching process, the textual descriptions provided by the patient are compared with the text in the case library, a similarity score is calculated, and the degree of matching is determined based on the score.
Logical matching algorithm: the logical matching algorithm performs matching based on predefined rules and logical conditions. These rules and conditions may be based on medical knowledge and experience, such as associations between symptoms and diseases, matching of medical history, etc. By logically matching the information provided by the patient with the rules, cases can be determined that are consistent with their condition.
Machine learning algorithm: machine learning algorithms can be used for training and optimization of the matching model. By taking as input the information provided by the patient and the data in the historic case base of the hospital, the machine learning model can learn patterns and associations between the data and predict the degree of matching. Common machine learning algorithms include decision trees, random forests, support Vector Machines (SVMs), neural networks, and the like.
Probability model: the probability model may be used to calculate the probability or similarity of the matches. For example, a Bayesian inference based algorithm can calculate the probability of a match based on patient provided information and statistics in a case library. These algorithms may take into account the relationships between multiple features and variables and provide probability-based matching results.
The algorithms have advantages and disadvantages, and proper algorithms or multiple algorithms are selected or combined according to the matching requirements and the data characteristics. For example, in text similarity matching, a cosine similarity algorithm or a Jaccard similarity algorithm may be used. For logical matching, matching rules and logical conditions may be designed and defined.
In the preferred embodiment of the present invention, as shown in fig. 1, the case matching unit 12 includes:
the complete checking subunit 125 is connected to the integrating subunit 124, and is configured to extract information from the disease description result and each of the historical medical history information by using the large language model, and when the information extraction result includes the preset information, use the disease description result and each of the historical medical history information as a pre-diagnosis result, and when the information extraction result does not include the preset information, prompt the current patient to supplement the disease information.
Specifically, in this embodiment, the large language model may perform integrity check before the pre-consultation case is generated, so as to ensure that the extracted information has sufficient integrity. This includes checking for missing symptoms, medical history, inspection results, etc., and ensuring that the resulting case provides the basic information required by the physician. The large language model may take into account patient individuation factors when generating the pre-consultation results. For example, the patient's descriptions are appropriately adjusted and modified according to the age, sex, underlying health, etc., of the patient to provide a more accurate and personalized case record. In order to facilitate understanding and use of the pre-consultation results, the large language model may employ a medical professional language, follow the expression specifications of medical literature, and ensure that the language style is accurate and concise. This helps the physician to understand and use the pre-consultation case more quickly during subsequent diagnosis and treatment. When the pre-consultation result is generated, the large language model can organize information through logic association, and internal logic and causal relation among the information are ensured. Meanwhile, the case information can be presented in a structured mode, so that doctors can more conveniently search and understand the contents of different parts. If there is uncertainty or potential risk factors in generating the pre-consultation results, the large language model may alert the physician by way of identification or alerting. This helps the physician to have a comprehensive understanding of the patient's condition and take appropriate measures for further diagnosis and treatment.
In general, the generation of pre-consultation results is the presentation of the patient's condition and medical history in a structured and accurate manner by integrating the information provided by the patient with the matching historical cases. This helps the physician to learn the condition of the patient more quickly during diagnosis and treatment and provides initial references and advice. However, it should be noted that the pre-consultation results are merely a reference for the physician, and that professional judgment and further examination by the physician is still required in the final diagnosis and treatment decision.
The invention also provides an intelligent pre-inquiry and rapid diagnosis and treatment method based on a large language model, which is applied to the intelligent pre-inquiry and rapid diagnosis and treatment system, wherein the intelligent pre-inquiry and rapid diagnosis and treatment system is connected with a history case library of a hospital, the history case library comprises a plurality of pieces of history information, and as shown in fig. 2, the intelligent pre-inquiry and rapid diagnosis and treatment method comprises the following steps:
step S1, an intelligent pre-consultation and rapid diagnosis and treatment system performs pre-consultation communication with a current patient by using a large language model to obtain a disease description result, and then searches a historical case library based on the disease description result to obtain a plurality of similar historical medical history information, wherein each historical medical history information and the disease description result are used as pre-consultation results;
step S2, the intelligent pre-consultation and rapid diagnosis and treatment system obtains a department with highest matching degree as a recommended registration department according to a pre-consultation result and a plurality of department descriptions and a plurality of department doctor information which are stored in advance and are associated with the departments, and feeds the department with highest matching degree back to a current patient;
step S3, the intelligent pre-consultation and rapid diagnosis and treatment system analyzes the pre-consultation result to obtain an analysis result, and then processes and obtains corresponding project suggestions comprising a plurality of suggestion examination projects according to registration information submitted by a patient and the analysis result to feed back the project suggestions to the current patient;
and S4, the intelligent pre-consultation and rapid diagnosis and treatment system provides a registration payment interface to finish payment after feeding back the recommended department to the current patient, and provides a project payment interface to finish payment when feeding back the project suggestion to the current patient.
In a preferred embodiment of the present invention, as shown in fig. 3, step S1 includes:
step S11, the intelligent pre-consultation and rapid diagnosis and treatment system receives an initial illness state description provided by a current patient, then inquires the current patient based on the initial illness state description by utilizing a large-scale language model to obtain a detailed illness state description fed back by the current patient, and then obtains an illness state description result according to the initial illness state description and detailed illness state description processing;
step S12, the intelligent pre-inquiry and rapid diagnosis and treatment system is matched with the disease description result in the historical case information base to obtain a plurality of pieces of historical medical history information similar to the disease description result, and the pieces of historical medical history information are integrated with the disease description result to serve as pre-inquiry results.
The foregoing is merely illustrative of the preferred embodiments of the present invention and is not intended to limit the embodiments and scope of the present invention, and it should be appreciated by those skilled in the art that equivalent substitutions and obvious variations may be made using the description and illustrations herein, which should be included in the scope of the present invention.
Claims (10)
1. An intelligent pre-consultation and rapid diagnosis and treatment system based on a large language model is characterized by being connected with a history case library of a hospital, wherein the history case library comprises a plurality of pieces of history information, and the intelligent pre-consultation and rapid diagnosis and treatment system comprises:
the pre-consultation module is used for carrying out pre-consultation communication with the current patient by utilizing a large language model to obtain a disease description result, then searching in the historical case library based on the disease description result to obtain a plurality of similar historical medical history information, and taking each of the historical medical history information and the disease description result as a pre-consultation result;
the department recommendation module is connected with the pre-consultation module and used for obtaining the department with the highest matching degree according to the pre-consultation result, a plurality of department descriptions related to the department and a plurality of department doctors information processing and feeding back the department with the highest matching degree to the current patient as a recommended registration department;
the examination suggestion module is respectively connected with the pre-consultation module and the department recommendation module and is used for analyzing the pre-consultation result to obtain an analysis result, and then processing according to registration information submitted by the patient and the analysis result to obtain corresponding project suggestions containing a plurality of suggestion examination projects and feeding the project suggestions back to the current patient;
and the payment module is respectively connected with the department recommendation module and the examination suggestion module and is used for providing a registration payment interface to finish payment after the recommended department is fed back to the current patient and providing a project payment interface to finish payment when the project suggestion is fed back to the current patient.
2. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 1, wherein the pre-consultation module includes:
the inquiry unit is used for receiving the provided initial condition description of the current patient, inquiring the current patient based on the initial condition description by using a large language model to obtain detailed condition description fed back by the current patient, and obtaining the condition description result according to the initial condition description and the detailed condition description;
and the case matching unit is connected with the inquiry unit and is used for matching in the historical case information base according to the disease description result to obtain a plurality of pieces of historical medical history information similar to the disease description result, and integrating the historical medical history information with the disease description result to serve as a pre-inquiry result.
3. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 2, wherein the case matching unit includes:
the first matching subunit is used for extracting patient symptoms from the disease description result, respectively calculating corresponding symptom similarity with each piece of history information in the history case library through a matching algorithm, and obtaining a plurality of pieces of history information based on the symptom similarity to form a first similarity set;
the second matching subunit is used for extracting the patient history diseases from the disease description result, respectively calculating corresponding disease similarity with each piece of history information in the history case library through a matching algorithm, and obtaining a plurality of pieces of history information based on the disease similarity to form a second similarity set;
the third matching subunit is used for extracting a patient medicine record from the disease description result, respectively calculating corresponding medicine similarity with each piece of history information in the history case library through a matching algorithm, and obtaining a plurality of pieces of history information based on the medicine similarity to form a third similarity set;
and the integration subunit is respectively connected with the first matching subunit, the second matching subunit and the third matching subunit and is used for integrating the historical medical history information and the illness state description result in the first similar set, the second similar set and the third similar set into the pre-consultation result.
4. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 1, wherein the department recommendation module includes:
the extraction unit is used for extracting keywords from the pre-consultation result by using a large-scale language model to obtain a disease keyword set, and extracting keywords from a plurality of pre-stored department descriptions and department doctor information associated with the departments by using the large-scale language model to obtain a corresponding department keyword set;
and the matching unit is connected with the extracting unit and is used for respectively calculating the matching degree between the illness state keyword set and each department keyword set by using a matching algorithm, and then feeding back the department corresponding to the department keyword set with the highest matching degree to the current patient as the recommended registration department.
5. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 1, wherein the examination suggestion module includes:
a storage unit for storing a plurality of departments, each of which is associated with a plurality of examination items;
a query unit, configured to associate the historical medical history information with the pre-inquiry result when the historical medical history information of the current patient is queried in the historical case library;
the analysis unit is respectively connected with the query unit and the storage unit, and is used for extracting keywords from the illness state description result and the associated history medical history information in the pre-consultation result by utilizing a large language model to obtain a plurality of illness state keywords, analyzing according to the illness state keywords to obtain the analysis result, and then screening a plurality of recommended examination items from the examination items by the large language model based on pre-stored medical knowledge, the analysis result and the recommended registration department, and feeding back the recommended examination items and the analysis result to the current patient as item suggestions.
6. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 5, wherein the examination suggestion module further comprises:
the adjusting unit is connected with the analyzing unit and is used for displaying the pre-consultation result to the consultation doctor when the consultation doctor recommending the registering department makes a consultation, and feeding back the project suggestion to the current patient after adjusting each suggestion examination project in the project suggestion according to the submitted modification opinion of the consultation doctor before feeding back the project suggestion to the current patient.
7. The intelligent pre-consultation and rapid diagnostic system of claim 3 or 4 wherein the matching algorithm includes one of a text similarity algorithm, a logical matching algorithm, a machine learning algorithm and a probabilistic model.
8. The intelligent pre-consultation and rapid diagnosis and treatment system of claim 2, wherein the case matching unit further includes:
and the complete checking subunit is connected with the integrating subunit and is used for respectively extracting information from the illness state description result and the historical medical history information by utilizing a large language model, taking the illness state description result and the historical medical history information as the pre-consultation result when judging that the information extraction result contains preset information, and prompting the current patient to supplement illness state information when judging that the information extraction result does not contain the preset information.
9. An intelligent pre-consultation and rapid diagnosis and treatment method based on a large language model, which is applied to the intelligent pre-consultation and rapid diagnosis and treatment system according to any one of claims 1 to 8, wherein the intelligent pre-consultation and rapid diagnosis and treatment system is connected with a history case library of a hospital, the history case library comprises a plurality of pieces of history information, and the intelligent pre-consultation and rapid diagnosis and treatment method comprises:
step S1, the intelligent pre-consultation and rapid diagnosis and treatment system performs pre-consultation communication with a current patient by using a large language model to obtain a disease description result, then searches for a plurality of similar historical medical history information in the historical case library based on the disease description result, and takes each of the historical medical history information and the disease description result as a pre-consultation result;
step S2, the intelligent pre-consultation and rapid diagnosis and treatment system obtains the department with highest matching degree as a recommended registration department according to the pre-consultation result and a plurality of department descriptions and a plurality of department doctor information which are stored in advance and associated with the department, and feeds the department with highest matching degree back to the current patient;
step S3, the intelligent pre-consultation and rapid diagnosis and treatment system analyzes the pre-consultation result to obtain an analysis result, and then processes and obtains corresponding project suggestions comprising a plurality of suggested examination projects according to registration information submitted by the patient and the analysis result and feeds the project suggestions back to the current patient;
and S4, the intelligent pre-consultation and rapid diagnosis and treatment system provides a registration payment interface to finish payment after feeding back the recommended department to the current patient, and provides a project payment interface to finish payment when feeding back the project suggestion to the current patient.
10. The intelligent pre-consultation and rapid diagnosis and treatment method according to claim 9, characterized in that the step S1 includes:
step S11, the intelligent pre-consultation and rapid diagnosis and treatment system receives an initial condition description provided by the current patient, then inquires the current patient based on the initial condition description by using a large language model to obtain a detailed condition description fed back by the current patient, and then obtains a condition description result according to the initial condition description and the detailed condition description;
step S12, the intelligent pre-consultation and rapid diagnosis and treatment system matches the disease description result in the historical case information base to obtain a plurality of pieces of historical medical history information similar to the disease description result, and integrates the pieces of historical medical history information with the disease description result to serve as a pre-consultation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310818964.6A CN117079798A (en) | 2023-07-05 | 2023-07-05 | Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310818964.6A CN117079798A (en) | 2023-07-05 | 2023-07-05 | Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117079798A true CN117079798A (en) | 2023-11-17 |
Family
ID=88703161
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310818964.6A Pending CN117079798A (en) | 2023-07-05 | 2023-07-05 | Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117079798A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117373658A (en) * | 2023-12-08 | 2024-01-09 | 北京回龙观医院(北京心理危机研究与干预中心) | Data processing-based auxiliary diagnosis and treatment system and method for depression |
-
2023
- 2023-07-05 CN CN202310818964.6A patent/CN117079798A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117373658A (en) * | 2023-12-08 | 2024-01-09 | 北京回龙观医院(北京心理危机研究与干预中心) | Data processing-based auxiliary diagnosis and treatment system and method for depression |
| CN117373658B (en) * | 2023-12-08 | 2024-03-08 | 北京回龙观医院(北京心理危机研究与干预中心) | Data processing-based auxiliary diagnosis and treatment system and method for depression |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10885150B2 (en) | System and a method for assessing patient treatment risk using open data and clinician input | |
| WO2022041729A1 (en) | Medication recommendation method, apparatus and device, and storage medium | |
| US10831863B2 (en) | System and a method for assessing patient risk using open data and clinician input | |
| US20180293354A1 (en) | Clinical content analytics engine | |
| TWI444921B (en) | Online system, method, and computer readable medium to facilitate providing health and wellness assistance | |
| CN109036545B (en) | Medical information processing method, apparatus, computer device and storage medium | |
| US20100076786A1 (en) | Computer System and Computer-Implemented Method for Providing Personalized Health Information for Multiple Patients and Caregivers | |
| US20050240439A1 (en) | System and method for automatic assignment of medical codes to unformatted data | |
| US20090062623A1 (en) | Identifying possible medical conditions of a patient | |
| US11551813B2 (en) | Augmented intelligence for next-best-action in patient care | |
| WO2016097886A1 (en) | Differential medical diagnosis apparatus adapted in order to determine an optimal sequence of diagnostic tests for identifying a pathology by adopting diagnostic appropriateness criteria | |
| US20140129246A1 (en) | Extension of clinical guidelines based on clinical expert recommendations | |
| CN106663136A (en) | System and method for scheduling healthcare follow-up appointments based on written recommendations | |
| CN111383754B (en) | Medical decision method, medical decision device, electronic device, and storage medium | |
| US20170364640A1 (en) | Machine learning algorithm to automate healthcare communications using nlg | |
| CN117079798A (en) | Intelligent pre-consultation and rapid diagnosis and treatment system and method based on large language model | |
| CN111429985B (en) | Electronic medical record data processing method and system | |
| JP4955197B2 (en) | Receipt file generation system | |
| CN116860935A (en) | Content management method, device, equipment and medium based on prompt word question-answer interaction | |
| WO2023217737A1 (en) | Health data enrichment for improved medical diagnostics | |
| US20220293253A1 (en) | Systems and methods using natural language processing to improve computer-assisted coding | |
| CN113096795B (en) | Multi-source data-aided clinical decision support system and method | |
| GB2548627A (en) | A system and a method for assessing patient treatment risk using open data and clinician input | |
| US11869674B2 (en) | Translation of medical evidence into computational evidence and applications thereof | |
| RU2818874C1 (en) | Medical decision support system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |