CN117954060A - Model training method, medical procedure checking method, system, device and medium - Google Patents
Model training method, medical procedure checking method, system, device and medium Download PDFInfo
- Publication number
- CN117954060A CN117954060A CN202410030273.4A CN202410030273A CN117954060A CN 117954060 A CN117954060 A CN 117954060A CN 202410030273 A CN202410030273 A CN 202410030273A CN 117954060 A CN117954060 A CN 117954060A
- Authority
- CN
- China
- Prior art keywords
- medical
- procedure
- information
- medical procedure
- flow
- 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
- 238000000034 method Methods 0.000 title claims abstract description 358
- 238000012549 training Methods 0.000 title claims abstract description 54
- 238000012795 verification Methods 0.000 claims abstract description 25
- 238000011156 evaluation Methods 0.000 claims description 68
- 238000004590 computer program Methods 0.000 claims description 11
- 239000008280 blood Substances 0.000 description 20
- 210000004369 blood Anatomy 0.000 description 20
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 239000000243 solution Substances 0.000 description 6
- 238000010241 blood sampling Methods 0.000 description 5
- 238000007639 printing Methods 0.000 description 5
- 238000011112 process operation Methods 0.000 description 3
- 230000002354 daily effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002347 injection Methods 0.000 description 2
- 239000007924 injection Substances 0.000 description 2
- 238000000968 medical method and process Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 239000000443 aerosol Substances 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000012377 drug delivery Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011022 operating instruction Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 210000002345 respiratory system Anatomy 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The application discloses a model training method, a medical procedure checking method, a system, a device and a medium. Performing voice recognition and/or voiceprint recognition on voice information to be recognized of the flow related personnel, determining corresponding flow related personnel identity information and medical flow node recognition information, performing medical flow verification according to the medical flow related information, determining a corresponding medical flow verification result, and determining whether to generate flow record information according to the medical flow verification result. The application utilizes the large language model to identify the identity and the flow node information of the flow related personnel, greatly improves the reliability and the relevance of the medical flow checking work, improves the medical flow checking efficiency and meets the medical flow checking requirement. The application is widely applied to the technical field of flow checking.
Description
Technical Field
The present application relates to the field of procedure checking technology, and in particular, to a model training method, a medical procedure checking method, a system, a device, and a medium.
Background
In daily medical procedures, there are very many complicated medical procedures, such as blood drawing, injection, specimen transfer, etc., and the requirements for checking the medical procedures are very high due to the great responsibility, however, the checking of the current medical procedures is very complex, and a series of steps are often required to be carried out: the account number and password are used for logging in an information system, entering a flow interface, selecting operation, confirming operation and the like, so that huge inconvenience and workload are brought to medical daily work, and obvious operation and record hysteresis exists.
Disclosure of Invention
In order to solve at least one technical problem existing in the related art, an embodiment of the present application provides a model training method, a medical procedure checking method, a system, a device and a medium.
In one aspect, an embodiment of the present application provides a model training method, including the following steps:
Collecting multi-modal training data; the multi-modal training data comprises voiceprint recognition data and medical procedure voice recognition data;
constructing a large language model, and performing dialogue matching training on the large language model by utilizing the multi-modal training data;
Acquiring a plurality of preset model evaluation indexes;
And carrying out model evaluation on the large language model according to each model evaluation index to obtain a model evaluation result, and determining whether to continue training the large language model according to the model evaluation result.
In some embodiments, the step of performing model evaluation on the large language model according to each model evaluation index to obtain a model evaluation result, and determining whether to continue training the large language model according to the model evaluation result specifically includes:
Performing model evaluation on the large language model according to each model evaluation index to obtain a model evaluation value corresponding to each model evaluation index;
Obtaining model evaluation thresholds corresponding to the model evaluation indexes;
comparing each model evaluation value with the corresponding model evaluation threshold value to determine the model evaluation result;
And stopping training the large language model to obtain a trained large language model when the model evaluation result is that each model evaluation value is larger than the corresponding model evaluation threshold value, otherwise, continuing training the large language model to enable each model evaluation value to be larger than the corresponding model evaluation threshold value.
In another aspect, an embodiment of the present application provides a medical procedure checking method, including the steps of:
Responding to a checking operation instruction of the medical procedure, and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel; the procedure associated personnel are medical care personnel or patients;
According to the voice information to be recognized of each flow-associated person, performing voice recognition and/or voiceprint recognition on the voice information to be recognized of each flow-associated person through a large language model, and determining corresponding flow-associated person identity information and medical flow node recognition information; the large language model is trained by the model training method;
acquiring medical procedure associated information corresponding to the medical procedure;
According to the medical procedure associated information, medical procedure checking is carried out on the procedure associated personnel identity information and the medical procedure node identification information, and a corresponding medical procedure checking result is determined;
And determining whether to generate flow record information according to the flow related personnel identity information and the medical flow node identification information according to the medical flow checking result.
In some embodiments, the step of determining the corresponding procedure-associated person identity information and medical procedure node identification information by performing voice recognition and/or voiceprint recognition on the to-be-recognized voice information of each procedure-associated person through a large language model according to the to-be-recognized voice information of each procedure-associated person specifically includes:
When the flow related personnel is the medical personnel, voiceprint recognition and voice recognition are carried out on voice information to be recognized corresponding to the medical personnel through the large language model, and corresponding medical personnel identity information and first medical flow node recognition information are determined; the first medical procedure node identification information comprises procedure node identification time and procedure operation instructions;
And when the flow related personnel is the patient, carrying out voice recognition on the voice information to be recognized corresponding to the patient through the large language model, and determining corresponding patient identity information and second medical flow node recognition information.
In some embodiments, when the process-related person is the medical care person, performing voiceprint recognition and voice recognition on voice information to be recognized corresponding to the medical care person through the large language model, and determining corresponding medical care person identity information and first medical process node identification information specifically includes:
Acquiring voiceprint input information of medical staff;
voiceprint recognition is carried out on the voice information of the medical staff through the large language model, and corresponding identity information of the medical staff is determined from the voiceprint input information of the medical staff;
Acquiring voice recognition input information of a medical procedure;
According to the medical procedure voice recognition input information, voice recognition is carried out on the medical staff voice information through the large language model, the procedure node recognition time is determined, and the procedure operation instruction is determined from the medical procedure voice recognition input information.
In some embodiments, the medical procedure related information includes procedure related personnel identity input information and procedure node input information, and the step of performing medical procedure verification on the procedure related personnel identity information and the medical procedure node identification information according to the medical procedure related information to determine a corresponding medical procedure verification result specifically includes:
When the flow related personnel is the medical personnel, checking whether the medical personnel identity information is contained in the flow related personnel identity input information or not through a data exchange platform according to the medical flow related information, and determining an identity checking result;
when the identity check result is that the medical personnel identity information is contained in the flow associated personnel identity input information, determining a target medical personnel operation authority corresponding to the medical personnel identity information from the flow associated personnel identity input information;
Checking whether the medical staff has the authority to execute the flow operation instruction according to the operation authority of the target medical staff, and obtaining a corresponding first medical flow checking result;
When the procedure association personnel is the patient, checking whether the patient identity information is consistent with the patient identity input information contained in the procedure association personnel identity input information, and obtaining a corresponding second medical procedure checking result.
In some embodiments, the step of determining whether to generate the procedure record information according to the procedure associated person identity information and the medical procedure node identification information according to the medical procedure checking result specifically includes:
When the first medical procedure checking result is that the medical staff has the right to execute the procedure operation instruction, generating corresponding first procedure record information according to the medical staff identity information and the first medical procedure node identification information, otherwise, sending out a procedure checking alarm and stopping executing the medical procedure node associated with the medical staff;
When the second medical procedure checking result is that the patient identity information is consistent with the patient identity input information contained in the procedure association personnel identity input information, generating corresponding second procedure record information according to the patient identity information and the second medical procedure node identification information, otherwise, sending out a procedure checking alarm and stopping executing the medical procedure node associated with the patient.
In another aspect, an embodiment of the present application provides a medical procedure verification system, including:
the first module is used for responding to the checking operation instruction of the medical procedure and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel; the procedure associated personnel are medical care personnel or patients;
The second module is used for carrying out voice recognition and/or voiceprint recognition on the voice information to be recognized of each flow related person through a large language model according to the voice information to be recognized of each flow related person, and determining corresponding flow related person identity information and medical flow node recognition information; the large language model is trained by the model training method;
A third module, configured to obtain medical procedure association information corresponding to the medical procedure;
A fourth module, configured to perform medical procedure verification on the procedure-related personnel identity information and the medical procedure node identification information according to the medical procedure-related information, and determine a corresponding medical procedure verification result;
And a fifth module, configured to determine, according to the medical procedure checking result, whether to generate procedure record information according to the procedure-related personnel identity information and the medical procedure node identification information.
In another aspect, an embodiment of the present application provides a medical procedure checking device, where the device includes a memory and a processor, where the memory stores a computer program, and the processor implements the medical procedure checking method described above when executing the computer program.
In yet another aspect, embodiments of the present application provide a computer readable storage medium storing a computer program which, when executed by a processor, implements the medical procedure verification method described above.
The application provides a model training method, a medical procedure checking method, a system, a device and a medium, which train a large language model by collecting multi-mode data, carry out voice recognition and/or voiceprint recognition on voice information to be recognized of procedure related personnel by utilizing the large language model, determine corresponding procedure related personnel identity information and medical procedure node recognition information, carry out medical procedure checking according to the medical procedure related information, determine corresponding medical procedure checking results, and determine whether to generate procedure record information according to the medical procedure checking results. The application utilizes the large language model to identify the identity and the flow node information of the flow related personnel, thereby greatly improving the reliability and the relevance of the medical flow checking work and improving the medical flow checking efficiency.
Drawings
FIG. 1 is a flow chart of a model training method provided by an embodiment of the present application;
FIG. 2 is a schematic diagram of training a large language model in an embodiment of the application;
FIG. 3 is another schematic diagram of training a large language model in an embodiment of the application;
FIG. 4 is a flow chart of a medical procedure verification method provided by an embodiment of the present application;
FIG. 5 is a schematic diagram of a medical procedure checking system according to an embodiment of the present application;
Fig. 6 is a schematic hardware structure of a medical procedure checking device according to an embodiment of the present application.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
It should be noted that although functional block division is performed in a device diagram and a logic sequence is shown in a flowchart, in some cases, the steps shown or described may be performed in a different order than the block division in the device, or in the flowchart. The terms first, second and the like in the description and in the claims and in the above-described figures, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the application only and is not intended to be limiting of the application.
Referring to fig. 1, fig. 1 is an optional flowchart of a model training method according to an embodiment of the present application, where the method may include, but is not limited to, steps S1 to S4:
Step S1, collecting multi-mode training data; the multimodal training data includes voiceprint recognition data and medical procedure speech recognition data;
s2, constructing a large language model, and performing dialogue matching training on the large language model by utilizing multi-mode training data;
s3, acquiring a plurality of preset model evaluation indexes;
And S4, carrying out model evaluation on the large language model according to the model evaluation indexes to obtain a model evaluation result, and determining whether to continue training the large language model according to the model evaluation result.
In step S1 of some embodiments, the voiceprint identification data is voiceprint data of each healthcare worker, corresponding to identity information of each healthcare worker, optionally, in a model training stage, the healthcare worker may continuously record voiceprint data periodically, including voiceprint data in various states, for example, voiceprint data in a cold state, voiceprint data before working every day, and the like; the medical procedure voice recognition data includes voice recognition information related to each medical procedure, for example, voice recognition information related to the blood drawing procedure includes "your good, the blood drawing procedure is now being performed" and "your good, i are XXX, the blood drawing is now performed by i, and the like.
In step S2 of some embodiments, dialogue matching training is performed on the large language model by using the multimodal training data, specifically, voice print data of medical staff is input into the large language model for dialogue matching, the large language model outputs identity information of the corresponding medical staff, the training is repeated continuously, in addition, voice recognition data of a medical procedure is input into the large language model for dialogue matching, and the large language model outputs the corresponding medical procedure and an operating instruction of the medical procedure being executed, and the training is repeated continuously.
Referring to fig. 2, fig. 2 is an optional schematic diagram of training a large language model according to an embodiment of the present application, in which voiceprint recognition data a is input into the large language model for dialogue matching, the large language model outputs corresponding healthcare worker a identity information, voiceprint recognition data B is input into the large language model for dialogue matching, and the large language model outputs corresponding healthcare worker B identity information.
Referring to fig. 3, fig. 3 is another alternative schematic diagram of training a large language model in an embodiment of the present application, in which medical speech recognition data A1 is input into the large language model for dialogue matching, the large language model outputs operation instructions 1 of the corresponding medical procedure a and the corresponding medical procedure a, voiceprint recognition data A2 is input into the large language model for dialogue matching, the large language model outputs operation instructions 2 of the corresponding medical procedure a and the corresponding medical procedure a, voiceprint recognition data B2 is input into the large language model for dialogue matching, and the large language model outputs operation instructions 2 of the corresponding medical procedure B and the corresponding medical procedure B.
In some embodiments, step S4 may include, but is not limited to including, step S41 to step S44:
Step S41, performing model evaluation on the large language model according to each model evaluation index to obtain a model evaluation value corresponding to each model evaluation index;
step S42, obtaining model evaluation thresholds corresponding to the model evaluation indexes;
Step S43, comparing each model evaluation value with a corresponding model evaluation threshold value to determine a model evaluation result;
And S44, stopping training the large language model to obtain a trained large language model when the model evaluation result is that each model evaluation value is larger than the corresponding model evaluation threshold, otherwise, continuing training the large language model to enable each model evaluation value to be larger than the corresponding model evaluation threshold.
In some embodiments, the model pre-estimation index includes a dialogue matching hit rate of the large language model, optionally, the model evaluation threshold is 99%, when the dialogue matching hit rate of the large language model is greater than 99%, training of the large language model is stopped, and when the dialogue matching hit rate of the large language model is less than 99%, training of the large language model is continued, and it is noted that the model evaluation index and the corresponding model evaluation threshold of the large language model may be set accordingly according to actual requirements.
Referring to fig. 4, fig. 4 is an optional flowchart of a medical procedure verification method according to an embodiment of the present application, which may include, but is not limited to, steps S101 to S105:
step S101, responding to a checking operation instruction of the medical procedure, and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel;
step S102, according to the voice information to be recognized of each flow-associated person, voice recognition and/or voiceprint recognition are carried out on the voice information to be recognized of each flow-associated person through a large language model, and corresponding flow-associated person identity information and medical flow node recognition information are determined;
step S103, acquiring medical procedure associated information corresponding to the medical procedure;
Step S104, according to the medical procedure associated information, checking the medical procedure of the identity information of the procedure associated personnel and the medical procedure node identification information, and determining a corresponding medical procedure checking result;
Step S105, determining whether to generate flow record information according to the flow related personnel identity information and the medical flow node identification information according to the medical flow checking result.
Wherein, the procedure associated personnel are medical care personnel or patients; the large language model is trained by the model training method.
In some embodiments, step S102 may include, but is not limited to including, step S201 to step S202:
Step S201, when the flow related personnel is medical personnel, voiceprint recognition and voice recognition are carried out on voice information to be recognized corresponding to the medical personnel through a large language model, and corresponding medical personnel identity information and first medical flow node recognition information are determined; the first medical procedure node identification information comprises procedure node identification time and a procedure operation instruction;
Step S202, when the procedure association personnel is a patient, performing voice recognition on voice information to be recognized corresponding to the patient through a large language model, and determining corresponding patient identity information and second medical procedure node recognition information.
In step S201 of some embodiments, the process node recognition time is the time of completing voice recognition and voiceprint recognition through the large language model, and the process operation instruction is an instruction corresponding to an operation performed in advance by a medical staff obtained after voice recognition through the large language model, and exemplary process operation instructions of the blood drawing process include "ready to draw blood", "blood drawing barcode printing", and "start drawing blood", etc.
In step S202 of some embodiments, the second medical procedure node identification information includes procedure node identification time and medical procedure information currently engaged by the patient, for example, the patient is engaged in an inhalation procedure of respiratory tract, the voice information to be identified of the patient is identified by using a large language model to confirm the identity of the patient, the identified time is yyyyy year-MM month-DD day-HH, FF is the time of the identification, and the medical procedure information currently engaged by the patient is the inhalation procedure information engaged by the patient, including an inhalation procedure code, an inhalation procedure name, and a procedure-associated healthcare personnel identity, etc.
In some embodiments, step S201 may include, but is not limited to including, step S301 to step S304:
Step S301, acquiring voiceprint input information of medical staff;
Step S302, voice print recognition is carried out on voice information of medical staff through a large language model, and corresponding medical staff identity information is determined from voice print input information of the medical staff;
step S303, acquiring voice recognition input information of a medical procedure;
Step S304, according to the medical procedure voice recognition input information, voice recognition is carried out on the medical staff voice information through the large language model, the procedure node recognition time is determined, and the procedure operation instruction is determined from the medical procedure voice recognition input information.
In step S302 of some embodiments, the voice print input information of the medical staff includes voice print data corresponding to each of the medical staff identity information, specifically, voice print recognition is performed on voice print information of the medical staff through a large language model, voice print data corresponding to the voice print information of the medical staff is determined from the voice print input information of the medical staff, and the medical staff identity information corresponding to the voice print data is determined as the medical staff identity information corresponding to the voice print information of the medical staff.
In step S304 of some embodiments, the medical procedure voice recognition input information includes voice recognition information and procedure operation instruction information corresponding to each medical procedure, optionally, voice recognition is performed on the medical staff voice information through a large language model, a medical procedure and a procedure operation instruction corresponding to the medical staff voice information are determined from the medical procedure voice recognition input information, and a procedure node recognition time is determined after the recognition is completed.
In some embodiments, the medical procedure associated information includes procedure associated personnel identity entry information and procedure node entry information, and step S104 may include, but is not limited to including, steps S401 through S404:
Step S401, when the flow related personnel is a medical personnel, checking whether the identity information of the medical personnel is contained in the identity input information of the flow related personnel or not through the data exchange platform according to the medical flow related information, and determining an identity checking result;
Step S402, when the identity check result is that the medical personnel identity information is contained in the flow associated personnel identity input information, determining the target medical personnel operation authority corresponding to the medical personnel identity information from the flow associated personnel identity input information;
Step S403, checking whether the medical staff has the authority for executing the flow operation instruction according to the operation authority of the target medical staff, and obtaining a corresponding first medical flow checking result;
Step S404, when the procedure association person is a patient, checking whether the patient identity information is consistent with the patient identity input information contained in the procedure association person identity input information, and obtaining a corresponding second medical procedure checking result.
In some embodiments, the data exchange platform checks whether the medical personnel identity information is contained in the flow associated personnel identity entry information, and if the medical personnel identity information is contained in the flow associated personnel identity entry information, the data exchange platform issues a corresponding digital authentication certificate, optionally, the data exchange platform may be a digital authentication platform.
In step S403 of some embodiments, according to the operation authority of the target medical staff, it is checked whether the medical staff has the authority to execute the procedure operation instruction, for example, the medical staff a has the operation authority from the medical procedure a to the medical procedure C, if the procedure operation instruction a is the related operation instruction of the medical procedure D, since the medical staff a does not have the authority to execute the related operation of the medical procedure D, the corresponding first medical procedure checking result is that the medical staff a does not have the authority to execute the procedure operation instruction a, and if the procedure operation instruction B is the related operation instruction of the medical procedure C, since the medical staff a has the authority to execute the related operation of the medical procedure C, the corresponding first medical procedure checking result is that the medical staff a has the authority to execute the procedure operation instruction B.
In step S105 of some embodiments, specifically, when the first medical procedure checking result is that the medical staff has the right of executing the procedure operation instruction, corresponding first procedure record information is generated according to the identity information of the medical staff and the first medical procedure node identification information, otherwise, a procedure checking alarm is sent out and the medical procedure node associated with the medical staff is stopped executing;
When the second medical procedure checking result is that the patient identity information is consistent with the patient identity input information contained in the procedure association personnel identity input information, generating corresponding second procedure record information according to the patient identity information and the second medical procedure node identification information, otherwise, sending out a procedure checking alarm and stopping executing the patient associated medical procedure node.
In some embodiments, the output result of the large language model, including patient identity information, medical personnel identity information, process node identification time, process operation instructions or process node identification information, etc., may be transmitted to an intelligent program or an intelligent application platform, and corresponding process record information may be generated by the intelligent program or the intelligent application platform.
In some embodiments, a number of procedure-related personnel in a medical procedure, such as a medical staff or a patient, may include, but is not limited to, medical procedures involving both medical staff and patient, such as blood drawing procedures, aerosol inhalation procedures, medical fluid injection procedures, and drug delivery procedures, etc., and medical procedures involving both medical staff, such as test sample delivery and delivery procedures, test report delivery procedures, etc.
In some embodiments, taking a blood drawing process as an example, the blood drawing process is checked by applying the medical process checking method provided by the embodiment of the application, which is specifically as follows:
The method comprises the steps that firstly, a patient enters an operated position, the identity of the patient is recognized, and first identity information of the patient is determined;
Secondly, medical staff obtains operation requests, such as a patient gives a blood drawing sheet to a nurse, and the medical staff looks at the blood drawing sheet to initiate recognition requests through voice, such as the medical staff says: "do you good, i now help you draw blood, ask you what is you' name? During the process, voice recognition and voice recognition are carried out on voice information to be recognized of medical staff by utilizing a large language model, the identity information of the medical staff and a flow operation instruction are recognized, the operation instruction for preparing blood drawing is recognized, data exchange is carried out through a data exchange platform, whether the identity information of the medical staff has flow operation authority is checked, a corresponding digital authentication certificate is obtained, the process node recognition time is related, and corresponding flow record information is generated through an intelligent program: time: time1 healthcare personnel: XXX ready to perform a blood drawing operation ";
Thirdly, the patient says "yes, i say that me is Zhang San", the large language model is utilized to carry out voice recognition on the voice information to be recognized of the patient, the second identity information of the patient is determined, at the moment, whether the second identity information of the patient is consistent with the first identity information of the patient obtained through face recognition in the first step is checked, double check of face recognition and voice recognition is achieved, if the identities are consistent, the process node recognition information is associated, and corresponding process record information is generated through an intelligent program: time: time2 patient: YYY is ready to draw blood ";
Fourth, medical staff say "good, I print the bar code of blood sampling for you now", similarly, utilize the large language model to discern medical staff's identity information and flow operation instruction, the flow operation instruction is printing the bar code of blood sampling, the association flow node discerns the time, through intelligent procedure generation flow record information: time: time3 healthcare personnel: XXX is patient: printing a blood sampling bar code' by YYY, then identifying an instruction of printing the bar code by using the large language model again, obtaining the bar code of the patient needing blood sampling this time through scanning information such as the bar code on a blood sampling application form of the patient and a preset interface, and starting printing the bar codes;
Fifthly, the medical staff say "I give you the start to draw blood now", and similarly, the medical staff identity information and the related flow operation instruction of "start to draw blood" are identified by using the large language model, the flow node identification time is related, and the flow record information is generated by the intelligent program: time: time4 healthcare: XXX is patient: YYY draw blood ";
Sixthly, if checking abnormal conditions such as checking errors occur when identity checking and/or checking flow operation authority checking are performed in the medical flow nodes in each step by using the large language model and the data exchange platform, sending out a flow checking alarm and stopping executing the corresponding medical flow nodes;
and seventh, packaging all the flow record information into a flow report form, automatically transferring and waiting for checking, and pushing the flow report form to both operation parties for confirmation.
Referring to fig. 5, fig. 5 is a schematic diagram of an alternative configuration of a medical procedure verification system provided in an embodiment of the present application, which may include, but is not limited to, a system comprising:
The first module is used for responding to the checking operation instruction of the medical procedure and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel; the procedure associated personnel are medical care personnel or patients;
The second module is used for carrying out voice recognition and/or voiceprint recognition on the voice information to be recognized of each flow associated person through the large language model according to the voice information to be recognized of each flow associated person, and determining corresponding flow associated person identity information and medical flow node recognition information; training the large language model by the model training method;
a third module, configured to obtain medical procedure association information corresponding to a medical procedure;
a fourth module, configured to perform medical procedure verification according to the medical procedure related information, the procedure related personnel identity information, and the medical procedure node identification information, and determine a corresponding medical procedure verification result;
And a fifth module for determining whether to generate flow record information according to the flow related personnel identity information and the medical flow node identification information according to the medical flow checking result.
The specific implementation of the medical procedure checking system is basically the same as the specific embodiment of the medical procedure checking method, and will not be described herein.
The embodiment of the application also provides a medical procedure checking device, which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the medical procedure checking method when executing the computer program. The medical procedure checking device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer and the like.
Referring to fig. 6, fig. 6 illustrates a hardware configuration of a medical procedure checking apparatus according to another embodiment, the medical procedure checking apparatus including:
The processor 901 may be implemented by a general purpose CPU (central processing unit), a microprocessor, an application specific integrated circuit (ApplicationSpecificIntegratedCircuit, ASIC), or one or more integrated circuits, etc. for executing related programs, so as to implement the technical solution provided by the embodiments of the present application;
The memory 902 may be implemented in the form of read-only memory (ReadOnlyMemory, ROM), static storage, dynamic storage, or random access memory (RandomAccessMemory, RAM), among others. The memory 902 may store an operating system and other application programs, and when the technical solutions provided in the embodiments of the present disclosure are implemented by software or firmware, relevant program codes are stored in the memory 902, and the processor 901 invokes a medical procedure checking method for executing the embodiments of the present disclosure;
an input/output interface 903 for inputting and outputting information;
The communication interface 904 is configured to implement communication interaction between the device and other devices, and may implement communication in a wired manner (e.g. USB, network cable, etc.), or may implement communication in a wireless manner (e.g. mobile network, WIFI, bluetooth, etc.);
A bus 905 that transfers information between the various components of the device (e.g., the processor 901, the memory 902, the input/output interface 903, and the communication interface 904);
Wherein the processor 901, the memory 902, the input/output interface 903 and the communication interface 904 are communicatively coupled to each other within the device via a bus 905.
The embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the medical procedure checking method when being executed by a processor.
The memory, as a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory remotely located relative to the processor, the remote memory being connectable to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The embodiment of the application provides a model training method, a medical procedure checking method, a system, a device and a medium, which are used for training a large language model by collecting multi-mode data, carrying out voice recognition and/or voiceprint recognition on voice information to be recognized of procedure-related personnel by utilizing the large language model, determining corresponding procedure-related personnel identity information and medical procedure node recognition information, carrying out medical procedure checking according to the medical procedure-related information, determining corresponding medical procedure checking results, and determining whether to generate procedure record information according to the medical procedure checking results. The application utilizes the large language model to identify the identity and the flow node information of the flow related personnel, thereby greatly improving the reliability and the relevance of the medical flow checking work and improving the medical flow checking efficiency.
The embodiments described in the embodiments of the present application are for more clearly describing the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application, and those skilled in the art can know that, with the evolution of technology and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.
It will be appreciated by persons skilled in the art that the embodiments of the application are not limited by the illustrations, and that more or fewer steps than those shown may be included, or certain steps may be combined, or different steps may be included.
The preferred embodiments of the present application have been described above with reference to the accompanying drawings, and are not thereby limiting the scope of the claims of the embodiments of the present application. Any modifications, equivalent substitutions and improvements made by those skilled in the art without departing from the scope and spirit of the embodiments of the present application shall fall within the scope of the claims of the embodiments of the present application.
Claims (10)
1. A method of model training, the method comprising the steps of:
Collecting multi-modal training data; the multi-modal training data comprises voiceprint recognition data and medical procedure voice recognition data;
constructing a large language model, and performing dialogue matching training on the large language model by utilizing the multi-modal training data;
Acquiring a plurality of preset model evaluation indexes;
And carrying out model evaluation on the large language model according to each model evaluation index to obtain a model evaluation result, and determining whether to continue training the large language model according to the model evaluation result.
2. The model training method according to claim 1, wherein the step of performing model evaluation on the large language model according to each model evaluation index to obtain a model evaluation result, and determining whether to continue training the large language model according to the model evaluation result comprises:
Performing model evaluation on the large language model according to each model evaluation index to obtain a model evaluation value corresponding to each model evaluation index;
Obtaining model evaluation thresholds corresponding to the model evaluation indexes;
comparing each model evaluation value with the corresponding model evaluation threshold value to determine the model evaluation result;
And stopping training the large language model to obtain a trained large language model when the model evaluation result is that each model evaluation value is larger than the corresponding model evaluation threshold value, otherwise, continuing training the large language model to enable each model evaluation value to be larger than the corresponding model evaluation threshold value.
3. A medical procedure verification method, the method comprising the steps of:
Responding to a checking operation instruction of the medical procedure, and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel; the procedure associated personnel are medical care personnel or patients;
According to the voice information to be recognized of each flow-associated person, performing voice recognition and/or voiceprint recognition on the voice information to be recognized of each flow-associated person through a large language model, and determining corresponding flow-associated person identity information and medical flow node recognition information; the large language model is trained by the model training method of any one of claims 1 to 2;
acquiring medical procedure associated information corresponding to the medical procedure;
According to the medical procedure associated information, medical procedure checking is carried out on the procedure associated personnel identity information and the medical procedure node identification information, and a corresponding medical procedure checking result is determined;
And determining whether to generate flow record information according to the flow related personnel identity information and the medical flow node identification information according to the medical flow checking result.
4. The medical procedure verification method according to claim 3, wherein the step of determining the corresponding procedure-related person identity information and medical procedure node identification information by performing voice recognition and/or voiceprint recognition on the to-be-recognized voice information of each procedure-related person through a large language model based on the to-be-recognized voice information of each procedure-related person specifically comprises:
When the flow related personnel is the medical personnel, voiceprint recognition and voice recognition are carried out on voice information to be recognized corresponding to the medical personnel through the large language model, and corresponding medical personnel identity information and first medical flow node recognition information are determined; the first medical procedure node identification information comprises procedure node identification time and procedure operation instructions;
And when the flow related personnel is the patient, carrying out voice recognition on the voice information to be recognized corresponding to the patient through the large language model, and determining corresponding patient identity information and second medical flow node recognition information.
5. The medical procedure verification method according to claim 4, wherein when the procedure-related person is the medical staff, the step of performing voiceprint recognition and voice recognition on voice information to be recognized corresponding to the medical staff through the large language model, and determining corresponding medical staff identity information and first medical procedure node identification information, specifically includes:
Acquiring voiceprint input information of medical staff;
voiceprint recognition is carried out on the voice information of the medical staff through the large language model, and corresponding identity information of the medical staff is determined from the voiceprint input information of the medical staff;
Acquiring voice recognition input information of a medical procedure;
According to the medical procedure voice recognition input information, voice recognition is carried out on the medical staff voice information through the large language model, the procedure node recognition time is determined, and the procedure operation instruction is determined from the medical procedure voice recognition input information.
6. The medical procedure verification method according to claim 4, wherein the medical procedure related information includes procedure related personnel identity entry information and procedure node entry information, and the step of performing medical procedure verification on the procedure related personnel identity information and the medical procedure node identification information according to the medical procedure related information to determine a corresponding medical procedure verification result specifically includes:
When the flow related personnel is the medical personnel, checking whether the medical personnel identity information is contained in the flow related personnel identity input information or not through a data exchange platform according to the medical flow related information, and determining an identity checking result;
when the identity check result is that the medical personnel identity information is contained in the flow associated personnel identity input information, determining a target medical personnel operation authority corresponding to the medical personnel identity information from the flow associated personnel identity input information;
Checking whether the medical staff has the authority to execute the flow operation instruction according to the operation authority of the target medical staff, and obtaining a corresponding first medical flow checking result;
When the procedure association personnel is the patient, checking whether the patient identity information is consistent with the patient identity input information contained in the procedure association personnel identity input information, and obtaining a corresponding second medical procedure checking result.
7. The medical procedure verification method according to claim 6, wherein the step of determining whether to generate procedure record information based on the procedure-related person identification information and the medical procedure node identification information based on the medical procedure verification result, specifically comprises:
When the first medical procedure checking result is that the medical staff has the right to execute the procedure operation instruction, generating corresponding first procedure record information according to the medical staff identity information and the first medical procedure node identification information, otherwise, sending out a procedure checking alarm and stopping executing the medical procedure node associated with the medical staff;
When the second medical procedure checking result is that the patient identity information is consistent with the patient identity input information contained in the procedure association personnel identity input information, generating corresponding second procedure record information according to the patient identity information and the second medical procedure node identification information, otherwise, sending out a procedure checking alarm and stopping executing the medical procedure node associated with the patient.
8. A medical procedure verification system, the system comprising:
the first module is used for responding to the checking operation instruction of the medical procedure and determining procedure related personnel corresponding to each medical procedure node and voice information to be recognized of each procedure related personnel; the procedure associated personnel are medical care personnel or patients;
The second module is used for carrying out voice recognition and/or voiceprint recognition on the voice information to be recognized of each flow related person through a large language model according to the voice information to be recognized of each flow related person, and determining corresponding flow related person identity information and medical flow node recognition information; the large language model is trained by the model training method of any one of claims 1 to 2;
A third module, configured to obtain medical procedure association information corresponding to the medical procedure;
A fourth module, configured to perform medical procedure verification on the procedure-related personnel identity information and the medical procedure node identification information according to the medical procedure-related information, and determine a corresponding medical procedure verification result;
And a fifth module, configured to determine, according to the medical procedure checking result, whether to generate procedure record information according to the procedure-related personnel identity information and the medical procedure node identification information.
9. A medical procedure collation apparatus, characterised in that the apparatus comprises a memory storing a computer program and a processor which when executing the computer program implements the medical procedure collation method as claimed in any one of claims 3 to 7.
10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the medical procedure collation method according to any one of claims 3 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410030273.4A CN117954060A (en) | 2024-01-08 | 2024-01-08 | Model training method, medical procedure checking method, system, device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410030273.4A CN117954060A (en) | 2024-01-08 | 2024-01-08 | Model training method, medical procedure checking method, system, device and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117954060A true CN117954060A (en) | 2024-04-30 |
Family
ID=90795710
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410030273.4A Pending CN117954060A (en) | 2024-01-08 | 2024-01-08 | Model training method, medical procedure checking method, system, device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117954060A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105380764A (en) * | 2015-10-26 | 2016-03-09 | 中国人民解放军第306医院 | Novel movable nursing system |
| CN106251865A (en) * | 2016-08-04 | 2016-12-21 | 华东师范大学 | A kind of medical treatment & health record Auto-writing method based on speech recognition |
| CN107657675A (en) * | 2017-10-27 | 2018-02-02 | 上海京颐科技股份有限公司 | A kind of medical care is registered method and device, storage medium and medical terminal |
| CN107978353A (en) * | 2017-11-29 | 2018-05-01 | 上海京颐科技股份有限公司 | The monitoring method and device of height warning medicine double-core pair, storage medium, medical mobile terminal |
| CN110570868A (en) * | 2019-09-12 | 2019-12-13 | 深圳市华创技术有限公司 | Medical interaction method and system based on voice recognition |
| CN110931099A (en) * | 2018-09-19 | 2020-03-27 | 北京航天长峰股份有限公司 | Method for automatically generating doctor order record through voice recognition |
| CN113421624A (en) * | 2021-06-07 | 2021-09-21 | 中山大学中山眼科中心 | Eye identification method, device, equipment and medium |
| CN116564315A (en) * | 2023-05-06 | 2023-08-08 | 平安科技（深圳）有限公司 | Voiceprint recognition method, voiceprint recognition device, voiceprint recognition equipment and storage medium |
| CN116612762A (en) * | 2023-05-05 | 2023-08-18 | 中山大学附属第六医院 | Voiceprint recognition-based doctor-patient identity verification method, system and device and storage medium |
-
2024
- 2024-01-08 CN CN202410030273.4A patent/CN117954060A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105380764A (en) * | 2015-10-26 | 2016-03-09 | 中国人民解放军第306医院 | Novel movable nursing system |
| CN106251865A (en) * | 2016-08-04 | 2016-12-21 | 华东师范大学 | A kind of medical treatment & health record Auto-writing method based on speech recognition |
| CN107657675A (en) * | 2017-10-27 | 2018-02-02 | 上海京颐科技股份有限公司 | A kind of medical care is registered method and device, storage medium and medical terminal |
| CN107978353A (en) * | 2017-11-29 | 2018-05-01 | 上海京颐科技股份有限公司 | The monitoring method and device of height warning medicine double-core pair, storage medium, medical mobile terminal |
| CN110931099A (en) * | 2018-09-19 | 2020-03-27 | 北京航天长峰股份有限公司 | Method for automatically generating doctor order record through voice recognition |
| CN110570868A (en) * | 2019-09-12 | 2019-12-13 | 深圳市华创技术有限公司 | Medical interaction method and system based on voice recognition |
| CN113421624A (en) * | 2021-06-07 | 2021-09-21 | 中山大学中山眼科中心 | Eye identification method, device, equipment and medium |
| CN116612762A (en) * | 2023-05-05 | 2023-08-18 | 中山大学附属第六医院 | Voiceprint recognition-based doctor-patient identity verification method, system and device and storage medium |
| CN116564315A (en) * | 2023-05-06 | 2023-08-08 | 平安科技（深圳）有限公司 | Voiceprint recognition method, voiceprint recognition device, voiceprint recognition equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105913846A (en) | Method, device and system for realizing voice registration | |
| US8899478B2 (en) | System and method for medical data transfer | |
| CN105095645A (en) | Method and system for querying health report under O2O mode | |
| CN109660533B (en) | Method and device for identifying abnormal flow in real time, computer equipment and storage medium | |
| CN104537290A (en) | Password system and achievement method based on multi-fingerprint identification | |
| CA2952079A1 (en) | Method and system for intelligent completion of medical record based on big data analytics | |
| CN108109689A (en) | Diagnosis and treatment session method and device, storage medium, electronic equipment | |
| CN110288995A (en) | Exchange method, device, storage medium and electronic equipment based on speech recognition | |
| CN109088853A (en) | A kind of diagnostic device Activiation method, terminal and server | |
| CN105147299A (en) | Method and device for confirming identity of user of wireless probe in fetal monitoring system | |
| CN104137110A (en) | Medication reminder method, device and system | |
| CN112908473A (en) | Model-based data processing method and device, computer equipment and storage medium | |
| CN113299294B (en) | Task type dialogue robot interaction method, device, equipment and storage medium | |
| CN107833600A (en) | Medical data typing check method and device, storage medium, electronic equipment | |
| WO2020004556A1 (en) | Patient assessment support device, patient assessment support method, and recording medium | |
| CN117954060A (en) | Model training method, medical procedure checking method, system, device and medium | |
| CN110739052A (en) | registration and treatment guiding method and system | |
| CN103559673A (en) | Organizing system and method for treatment closed-loop information | |
| CN113707157A (en) | Identity verification method and device based on voiceprint recognition, electronic equipment and medium | |
| CN113205716A (en) | Anesthesia nursing staff service capability assessment method and system | |
| CN112802598A (en) | Real-time auxiliary diagnosis and treatment method and system based on voice diagnosis and treatment data | |
| CN103473722A (en) | Doctor-patient dispute early-warning system | |
| CN108052811A (en) | Terminal control method and system based on face texture identification | |
| CN114446301A (en) | Intelligent follow-up method, device, equipment and storage medium | |
| CN111402985A (en) | Hospitalization report writing system and use method thereof |
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 |