Anesthesia equipment regulation and control method and system based on large model

Abstract

The invention discloses an anesthesia equipment regulation and control method and system based on a large model, and belongs to the field of large model application, and the method comprises the steps: based on a medical record number of a patient, extracting preoperative medical record data of the patient from a hospitalization system, and extracting an intraoperative real-time physiological monitoring index of the patient from a surgical anesthesia management system; inputting the preoperative medical record data, the intraoperative real-time physiological monitoring index and the anesthesia operation problem to be decided into an anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information corresponding to the anesthesia operation problem, the anesthetist in the operation regulates and controls the anesthesia equipment in the working state according to the first operation information. By implementing the application, the problem that the accuracy of anesthesia operation decisions is difficult to guarantee in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the field of large model applications, and in particular to a large model-based anesthesia equipment control method and system. Background Art

[0002] In modern medical surgery, anesthesia is a critical step in ensuring smooth surgical progress and patient safety. Currently, anesthesiologists rely primarily on their own clinical experience, preoperative medical records, and intraoperative physiological monitoring data when making decisions about anesthesia procedures. However, this traditional decision-making approach has limitations. On the one hand, the experience levels of different anesthesiologists vary, which can lead to discrepancies in decision-making outcomes. On the other hand, the patient's physiological condition during surgery is complex and changeable, and relying solely on manual judgment makes it difficult to comprehensively and accurately consider all factors, thus affecting the accuracy and safety of anesthesia procedures. Furthermore, existing anesthesia equipment control often relies on manual operation and lacks intelligent, automated decision-making support, making it difficult to adapt to complex and changing clinical needs. Summary of the Invention

[0003] The present invention provides an anesthesia equipment control method and system based on a large model. The method can solve the problem in the prior art that anesthesia operation decisions are overly dependent on the personal experience of anesthesiologists, resulting in large differences in decision results among different anesthesiologists and difficulty in ensuring decision accuracy.

[0004] An embodiment of the present invention provides an anesthesia equipment control method based on a large model, which is applicable to an anesthesia equipment control system, including:

[0005] Obtain the patient's medical record number and the intraoperative anesthesia procedure to be decided;

[0006] Extracting the patient's preoperative medical records from a preset hospitalization system according to the medical record number;

[0007] Extracting the patient's intraoperative real-time physiological monitoring indicators from a preset surgical anesthesia management system according to the medical record number;

[0008] The preoperative medical records, the intraoperative real-time physiological monitoring indicators and the anesthesia operation problems are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesia doctor during the operation can regulate the anesthesia equipment in working state according to the first operation information; wherein, the anesthesia operation consultation model is obtained after training with a pre-labeled anesthesia data set as the input of the open source large model to be fine-tuned, and with theoretical anesthesia operations as the output of the open source large model to be fine-tuned.

[0009] Furthermore, the training process of the anesthesia operation consultation model includes:

[0010] Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records;

[0011] The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model;

[0012] The open source large model after iterative training is used as the anesthesia operation consultation model;

[0013] In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges.

[0014] If yes, stop training and get the open source model after training.

[0015] If not, adjust the parameters of the current open source large model to obtain the open source large model for the next iterative training.

[0016] Furthermore, the type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history;

[0017] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0018] The anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information about the anesthetic dosage based on the bispectral index, the EEG state index, the patient's age, the patient's ASA grade and the past medical history.

[0019] Furthermore, the type of anesthesia operation problem includes the type of muscle relaxation management operation; the intraoperative real-time physiological monitoring index includes the real-time four-stimulation ratio; the preoperative medical history data includes the type of surgery;

[0020] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0021] The anesthetic operation problem based on the muscle relaxant management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model determines the four-time train-of-stimulation ratio range according to the patient's age, the patient's ASA grade, the past medical history and the type of surgery, and outputs the first operation information about the muscle relaxant dosage based on the real-time four-time train-of-stimulation ratio and the four-time train-of-stimulation ratio range.

[0022] Furthermore, the type of anesthesia operation problem includes the type of hemodynamic management operation; the intraoperative real-time physiological monitoring indicators include focused heart rate, blood pressure, blood oxygen saturation, central venous pressure, cardiac output, cardiac index and mixed venous oxygen saturation;

[0023] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0024] The anesthesia operation problem based on the hemodynamic management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information on adjusting the dosage of vasoactive drugs according to the patient's age, the patient's ASA grade, the past medical history, the type of surgery, the focused heart rate, the blood pressure, the blood oxygen saturation, the central venous pressure, the cardiac output, the cardiac index and the mixed venous oxygen saturation.

[0025] Furthermore, after obtaining the intraoperative anesthesia operation problem to be decided, it also includes:

[0026] Identify the information type of the intraoperative anesthesia operation problem to be decided;

[0027] When it is identified that the information type of the intraoperative anesthesia operation question to be decided is a voice type, a preset voice recognition algorithm is called to convert the information type of the intraoperative anesthesia operation question into a text type.

[0028] An embodiment of the present invention further provides an anesthesia equipment control system based on a large model, comprising: a data acquisition module, a preoperative medical record acquisition module, a physiological monitoring index acquisition module, and an anesthesia equipment control module;

[0029] The data acquisition module is used to obtain the patient's medical record number and the intraoperative anesthesia operation problem to be decided;

[0030] The preoperative medical record acquisition module is used to extract the patient's preoperative medical record from a preset hospitalization system according to the medical record number;

[0031] The physiological monitoring index acquisition module is used to extract the patient's intraoperative real-time physiological monitoring index from a preset surgical anesthesia management system according to the medical record number;

[0032] The anesthesia equipment control module is used to input the preoperative medical history data, the intraoperative real-time physiological monitoring indicators and the anesthesia operation problems into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesia doctor during the operation can control the anesthesia equipment in a working state according to the first operation information; wherein, the anesthesia operation consultation model is obtained after training with a pre-labeled anesthesia data set as the input of the open source large model to be fine-tuned, and with theoretical anesthesia operations as the output of the open source large model to be fine-tuned.

[0033] Furthermore, the training process of the anesthesia operation consultation model includes:

[0034] Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records;

[0035] The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model;

[0036] The open source large model after iterative training is used as the anesthesia operation consultation model;

[0037] In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges.

[0038] If yes, stop training and get the open source model after training.

[0039] If not, adjust the parameters of the current open source large model to obtain the open source large model for the next iterative training.

[0040] Furthermore, the type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history;

[0041] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0042] The anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information about the anesthetic dosage based on the bispectral index, the EEG state index, the patient's age, the patient's ASA grade and the past medical history.

[0043] Furthermore, the large model-based anesthesia equipment control system further includes: a voice type conversion module;

[0044] The voice type conversion module is used to identify the information type of the intraoperative anesthesia operation problem to be decided; and when it is identified that the information type of the intraoperative anesthesia operation problem to be decided is a voice type, call a preset voice recognition algorithm to convert the information type of the intraoperative anesthesia operation problem into a text type.

[0045] The following beneficial effects are achieved by implementing the present invention:

[0046] The present invention provides an anesthesia equipment control method and system based on a large model. The method inputs the patient's preoperative medical history data, intraoperative real-time physiological monitoring indicators, and anesthesia operation problems into an anesthesia operation consultation model, so that the anesthesia operation consultation model outputs corresponding first operation information. Since the anesthesia operation consultation model is based on a pre-labeled anesthesia data set as the input of an open source large model to be fine-tuned, and theoretical anesthesia operations as the output of the open source large model to be fine-tuned, and is obtained after training, the open source large model can be fine-tuned to enable the open source large model to mine potential laws and correlations in anesthesia data; therefore, the first operation information output by the anesthesia operation consultation model integrates massive clinical data and professional theoretical knowledge, provides anesthesiologists with an objective and scientific auxiliary decision-making basis, effectively makes up for the shortcomings of relying solely on the experience of anesthesiologists in making decisions, and improves the accuracy and consistency of anesthesia operation decisions;

[0047] This allows the anesthesiologist to adjust the operating parameters of the anesthesia equipment in working state according to the first operation information, thereby realizing intelligent and precise regulation of the anesthesia equipment, better adapting to the patient's changing physiological condition during surgery, further ensuring the patient's life safety, improving the quality and efficiency of surgical anesthesia, and providing strong support for the smooth implementation of medical operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 This is a flowchart of a large-model-based anesthesia equipment control method provided in one embodiment of the present application;

[0050] Figure 2 This is a schematic diagram of the structure of an anesthesia equipment control system based on a large model provided in one embodiment of the present application;

[0051] Figure 3 This is a structural diagram of an anesthesia equipment control system based on a large model provided in another embodiment of the present application. DETAILED DESCRIPTION

[0052] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of this application.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0054] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0055] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0056] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0057] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0058] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0059] The following are the professional terms and known technical contents involved in the embodiments of the present invention:

[0060] (1) ASA classification: ASA classification is the physical condition classification of the American Society of Anesthesiologists (ASA). It is an important tool for anesthesiologists to assess the patient's general health status and surgical risks before surgery. It helps the anesthesia operation consultation model to output the first operation information more comprehensively and accurately. The specific classification standards are as follows:

[0061] Level I: Normal health, no physical, psychological or mental illness, no medication that may affect surgery. This type of patient has good tolerance to anesthesia and surgery and has low risk.

[0062] Level II: Mild systemic diseases without functional limitations, such as mild hypertension, non-severe diabetes, well-controlled asthma, etc. The patient's daily activities are not restricted, and the risks of anesthesia and surgery are relatively low;

[0063] Level III: Patients with severe systemic diseases that limit their daily activities but have not yet lost their ability to work, such as severe hypertension, unstable angina, chronic obstructive pulmonary disease, and chronic renal insufficiency. These patients have poor tolerance to anesthesia and surgery, and the risk of anesthesia is significantly increased;

[0064] IV: Patients with severe systemic diseases who have lost their ability to work and are often life-threatening, such as advanced heart disease, liver failure, kidney failure, and advanced malignant tumors. The risks of anesthesia and surgery are extremely high.

[0065] Level V: Patients who are dying and whose life is unlikely to be sustained for 24 hours regardless of surgery. The risks of anesthesia and surgery are extremely high and are generally performed only in emergency situations.

[0066] V Ι Grade: Declared brain dead.

[0067] See also Figure 1To address the problem in the prior art that anesthesia operation decisions are overly dependent on the anesthesiologist's personal experience, resulting in large differences in decision-making results between different anesthesiologists and difficulty in ensuring decision accuracy, an embodiment of the present invention provides an anesthesia equipment control method based on a large model, which is applicable to an anesthesia equipment control system and includes:

[0068] S1. Obtain the patient's medical record number and the intraoperative anesthesia procedure to be decided;

[0069] In a preferred embodiment, after obtaining the intraoperative anesthesia operation problem to be decided, the process further includes:

[0070] Identify the information type of the intraoperative anesthesia operation problem to be decided;

[0071] When it is recognized that the information type of the intraoperative anesthesia operation question to be decided is a voice type, calling a preset voice recognition algorithm to convert the information type of the intraoperative anesthesia operation question into a text type;

[0072] Specifically, in an actual surgical anesthesia scenario, when the operation is in progress, the anesthesiologist will face various operational issues that need to be decided. In this embodiment, the anesthesiologist will input the current patient's medical record number and the intraoperative anesthesia operation issue to be decided into the anesthesia equipment control system;

[0073] The anesthesia equipment control system will then determine the type of information input by the anesthesiologist, whether it is in voice form or directly entered in text form; if it is determined to be a voice type, the system will immediately start the preset voice recognition algorithm, which is based on natural language processing technology. By collecting, analyzing, and extracting features from voice signals, the voice content is converted into corresponding text content, thereby unifying the information format.

[0074] S2. extracting the patient's preoperative medical records from a preset hospitalization system according to the medical record number;

[0075] Illustratively, in step S2, the preset hospitalization system adopts the hospital's HIS system; it should be noted that the HIS system, through a strict data screening mechanism, only includes patient data with complete hospitalization medical records, clear identity identification and accurate matching data, while excluding cases where medical history information is missing, data is incomplete and cannot be restored, to ensure that the collected data is complete and valid.

[0076] Specifically, in terms of data collection, the patient's preoperative medical records are extracted from the hospitalization system (i.e., the HIS system), wherein the preoperative medical records include the patient's basic information, patient diagnosis information, current medical history, past medical history, patient ASA grade, and test results;

[0077] Specifically, the patient's basic information includes the patient's name, age, gender, weight, height, surgery type, medical record number, etc.

[0078] Secondly, the diagnostic information, the current medical history, and the past medical history are used to present the patient's disease development trajectory and health baseline; for example, a history of drug allergies, chronic diseases, etc. in the past medical history can effectively avoid anesthesia risks; the diagnostic information includes the doctor's initial judgment of the patient's disease after the patient is admitted to the hospital based on the patient's chief complaint, symptoms, physical signs, and preliminary examination results; the current medical history includes the onset of the disease and the main symptom characteristics, that is, the time, place, onset of the patient's disease, prodromal symptoms, etc. (for example, whether the patient's disease onset is sudden or gradual, and whether there are obvious predisposing factors before the onset, such as trauma, infection, mood swings, etc.); the main symptom characteristics include a detailed description of the location, nature, degree, duration, and frequency of the main symptoms;

[0079] In addition, test results quantify the patient's physical condition through objective indicators (such as blood tests, imaging reports, electrocardiograms, etc.);

[0080] Specifically, the patient's preoperative medical record data is retrieved from the hospitalization system through the patient's medical record number to provide basic data support for the operation of the anesthesia operation consultation model.

[0081] S3. Extracting the patient's intraoperative real-time physiological monitoring indicators from a preset surgical anesthesia management system according to the medical record number;

[0082] Specifically, the preset surgical anesthesia management system is a surgical anesthesia information system (AIS), which extracts real-time patient data from multiple dimensions during the surgical process;

[0083] Specifically, on the one hand, the patient's basic information is collected, including name, age, gender, weight, height, medical record number, operation name and anesthesia method. This information is the basis for establishing the patient's intraoperative data file; on the other hand, key intraoperative physiological indicators are collected in real time, that is, real-time collection of focused heart rate, blood pressure (including systolic pressure, diastolic pressure, MAP), blood oxygen saturation, central venous pressure, cardiac output, cardiac index, mixed venous oxygen saturation, respiratory rate, tidal volume, airway pressure, end-tidal carbon dioxide partial pressure, real-time four-beat train-of-stimulation ratio and bispectral index (BIS) and other parameters.

[0084] S4. Inputting the preoperative medical records, the intraoperative real-time physiological monitoring indicators, and the anesthesia operation problem into an anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesiologist during the operation can adjust the anesthesia equipment in a working state according to the first operation information; wherein the anesthesia operation consultation model is obtained after training using a pre-labeled anesthesia dataset as the input of an open source large model to be fine-tuned, and using theoretical anesthesia operations as the output of the open source large model to be fine-tuned;

[0085] In a preferred embodiment, the training process of the anesthesia operation consultation model includes:

[0086] Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records;

[0087] The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model;

[0088] The open source large model after iterative training is used as the anesthesia operation consultation model;

[0089] In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges.

[0090] If yes, stop training and get the open source model after training.

[0091] If not, adjust the parameters of the current open source model to obtain the open source model for the next iterative training;

[0092] In order to make the output results of the anesthesia operation consultation model more accurate, this embodiment fine-tunes the existing open source large model so that the fine-tuned open source large model can better meet the actual needs of anesthesia clinical practice.

[0093] Specifically, we obtained a pre-labeled anesthesia dataset and a Transformer-based open-source large model to be fine-tuned. The pre-labeled anesthesia dataset includes expert knowledge data in the field of anesthesia, which is derived from the experience summary and academic research results of authoritative experts in the field of anesthesia. It contains various anesthesia operation specifications, guidelines, risk assessment standards, etc., representing the professional theoretical system in the field of anesthesia. We also collected information on anesthesia decision-making issues in clinical surgery, such as "How to adjust the anesthesia plan when the patient's blood pressure drops suddenly during surgery" and "Which anesthesia method should be selected for elderly patients with cardiopulmonary diseases";

[0094] Then, the patient diagnosis and treatment data of several patients are integrated, including preoperative medical records and historical intraoperative real-time physiological monitoring indicators of several patients who have completed surgery, reflecting the individual differences of different patients;

[0095] In addition, the actual anesthesia operation records the actual operation plan and final effect feedback taken by the anesthesiologist in response to various anesthesia operation problems in previous surgeries;

[0096] In an illustrative example, after acquiring the data, the open source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open source big model, and the theoretical anesthesia operation as outputs.

[0097] Specifically, by performing a four-tuple mapping of patient diagnosis and treatment data (such as ASA grade, medical history, etc.) with expert knowledge data in the field of anesthesia (such as drug selection and dosage adjustment), anesthesia operation questions (i.e., questions about anesthesia operation asked by anesthesiologists during surgery), and actual anesthesia operation records (i.e., the anesthesia operation finally performed by anesthesiologists on patients during surgery), a corresponding relationship of "clinical characteristics-operation questions-expert knowledge-operation plan" is constructed. Therefore, in this embodiment, through this mapping mechanism, the open source large model can learn the contextualized decision logic of anesthesia, rather than a simple reflexive response.

[0098] For example, when the anesthesia dataset contains the data "elderly patient (age ≥ 65 years) + history of hypertension (systolic blood pressure ≥ 140 mmHg) + heart rate 90 beats / min" and the question "Which anesthetic drug should be selected during the induction period?", the open source big model can learn the following decision logic by annotating the data, combining the corresponding expert knowledge annotations (such as "prioritizing drugs with less circulatory suppression" in the "2024 Guidelines for Anesthesia Management of Elderly Patients") and the actual anesthesia operation records performed by the anesthesiologist in the surgery (such as "select etomidate 0.2 mg / kg instead of propofol"):

[0099] (1) Interpretation of clinical characteristics: Age ≥ 65 years indicates decreased organ reserve function; history of hypertension indicates weakened compensatory capacity of the cardiovascular system; heart rate 90 beats / min indicates close to the upper limit of normal, and further inhibition of myocardial contractility should be avoided;

[0100] (2) Expert knowledge data in the field of anesthesia: drugs with less circulatory suppression are preferred, and etomidate has a weaker inhibitory effect on the cardiovascular system than propofol;

[0101] (3) Operation plan derivation: Based on clinical characteristics and expert knowledge, the optimal plan of "etomidate 0.2 mg / kg intravenous induction" was derived, and the rationality of the theoretical anesthesia operation (i.e., the optimal plan) output by the open source large model was verified through actual anesthesia operation records;

[0102] In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; by calculating the loss function value, it is determined whether the loss function value converges. If so, the training is stopped to obtain the open source large model after training. If not, the parameters of the current open source large model are adjusted to obtain the open source large model for the next iterative training;

[0103] By continuously repeating the above iterative training process, the fine-tuned open source large model gradually learns the professional knowledge in the field of anesthesia, the mapping relationship between different patient conditions and corresponding anesthesia operations, and can output theoretical anesthesia operation recommendations that are more in line with clinical practice, more accurate and reliable based on the input patient information and intraoperative anesthesia operation problems, providing anesthesiologists with scientific and effective auxiliary decision support, significantly improving the accuracy and safety of anesthesia operations.

[0104] In a preferred embodiment, the type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history;

[0105] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0106] Inputting the anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data, and the intraoperative real-time physiological monitoring index into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information on the anesthetic dosage according to the bispectral index, the EEG state index, the patient's age, the patient's ASA grade, and the past medical history;

[0107] Specifically, anesthesia operation questions based on the anesthesia depth management operation type (such as: Is it necessary to adjust the anesthetic concentration of the patient with medical record number AAA?), the preoperative medical record data, and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model. The anesthesia operation consultation model will judge the patient's basic condition. If the patient is over 60 years old or has poor physical function or has a history of cognitive dysfunction, the appropriate range of the bispectral index (BIS value) or the electroencephalogram state index (CSI value) is set to 45-60; if the patient does not have the above special conditions, the normal bispectral index (BIS value) or the electroencephalogram state index (CSI value) is set to 40-60.

[0108] It should be noted that during the application process, the anesthesia operation consultation model will run and output reasonable first operation information based on the acquired preoperative medical records, real-time physiological monitoring indicators during the operation, and anesthesia operation problems, combined with the learned expert knowledge data in the field of anesthesia;

[0109] If the bispectral index (BIS value) or the cerebrospinal state index (CSI value) is below the lower limit of the appropriate range, the first operation information about the anesthetic dose, which is to reduce the concentration of the inhaled anesthetic or reduce the intravenous anesthetic dose, will be output;

[0110] It should be noted that the specific dosage adjustment value will be output according to the actual situation and will not be repeated here;

[0111] If the bispectral index (BIS value) or the cerebrospinal state index (CSI value) is at the boundary of the appropriate range (e.g., the BIS value or CSI value is 60) and the surgical stimulation is strong, the first operation information regarding the increase of the anesthetic dose will be output. For example, the first operation information is "choose to intravenously inject propofol 10-20 mg, propofol 2.5-5.0 mg, or increase the inhaled concentration of sevoflurane or desflurane";

[0112] If the bispectral index (BIS value) or the cerebrospinal state index (CSI value) is within the appropriate range, the first operation information for maintaining the current anesthesia plan will be output and continuous monitoring will be carried out;

[0113] It should be noted that, during the application process, the preoperative medical records and intraoperative real-time physiological monitoring indicators input into the anesthesia operation consultation model correspond to the patient diagnosis and treatment data during the training process of the anesthesia operation consultation model.

[0114] In a preferred embodiment, the type of anesthesia operation problem includes the type of muscle relaxation management operation; the intraoperative real-time physiological monitoring indicator includes the real-time train-of-four stimulation ratio; the preoperative medical history data includes the type of surgery;

[0115] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0116] Inputting an anesthetic operation problem based on the muscle relaxant management operation type, the preoperative medical history data, and the intraoperative real-time physiological monitoring index into the anesthesia operation consultation model, so that the anesthesia operation consultation model determines a train-of-four stimulation ratio range according to the patient's age, the patient's ASA grade, the previous medical history, and the type of surgery, and outputs first operation information about the muscle relaxant dosage based on the real-time train-of-four stimulation ratio and the train-of-four stimulation ratio range;

[0117] Schematically, the anesthesia operation consultation model quantifies the degree of neuromuscular blockade using the TOF ratio: when the TOF ratio = 1.0, it indicates normal neuromuscular function; when the TOF ratio < 1.0, it indicates the presence of neuromuscular blockade; when the TOF ratio < 0.9, it indicates the risk of residual blockade; when the TOF ratio ≥ 0.9, it indicates that the safe extubation criteria have been met;

[0118] The anesthesia operation consultation model then determines the scenario-based TOF ratio based on the type of surgery:

[0119] When rapid sequence induction intubation is performed, the scenario-based TOF ratio is close to 0; when it is a general surgical operation, the scenario-based TOF ratio is maintained at <0.4 during the operation; when it is before extubation after surgery, the scenario-based TOF ratio is ≥0.9; when it is a short operation such as laparoscopy, the scenario-based TOF ratio is maintained at <0.4, and short-acting muscle relaxants are recommended; when it is a long operation such as cardiac surgery, the scenario-based TOF ratio is maintained at <0.2, and medium- and long-acting muscle relaxants are recommended; when it is muscle relaxant antagonism, the scenario-based TOF ratio before antagonism is >0.2, and the scenario-based TOF ratio after antagonism needs to reach ≥0.9;

[0120] After receiving an anesthesia operation question based on the muscle relaxant management operation type (e.g., Is it necessary to adjust the muscle relaxant dosage for the patient with medical record number AAA?), the patient's age, the patient's ASA grade, the past medical history, the surgery type, and the real-time train-of-four stimulation ratio, the anesthesia operation consultation model will dynamically adjust the scenario-based TOF threshold according to the patient's physical condition, obtain the train-of-four stimulation ratio range, and then compare the real-time train-of-four stimulation ratio with the corresponding train-of-four stimulation ratio range:

[0121] Specifically, if the real-time train-of-four stimulation ratio is lower than the lower limit of the train-of-four stimulation ratio range, first operation information indicating that the muscle relaxant dosage is no longer required and monitoring is maintained will be output;

[0122] If the real-time train-of-four stimulation ratio exceeds the upper limit of the train-of-four stimulation ratio range, first operation information regarding the additional muscle relaxant dosage (such as rocuronium bromide 0.1-0.2 mg / kg) will be output.

[0123] In a preferred embodiment, the type of anesthesia operation problem includes a hemodynamic management operation type; the intraoperative real-time physiological monitoring indicators include focused heart rate, blood pressure, blood oxygen saturation, central venous pressure, cardiac output, cardiac index, and mixed venous oxygen saturation;

[0124] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0125] inputting an anesthesia operation question based on the hemodynamic management operation type, the preoperative medical history data, and the intraoperative real-time physiological monitoring index into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information on adjusting the vasoactive drug dosage according to the patient's age, the patient's ASA classification, the previous medical history, the operation type, the focused heart rate, the blood pressure, the blood oxygen saturation, the central venous pressure, the cardiac output, the cardiac index, and the mixed venous oxygen saturation;

[0126] Specifically, the intraoperative real-time physiological monitoring indicators include focused heart rate, blood pressure (including systolic pressure, diastolic pressure, MAP), blood oxygen saturation, central venous pressure, cardiac output, cardiac index, and mixed venous oxygen saturation;

[0127] Specifically, anesthesia operation questions based on the hemodynamic management operation type (e.g., "Do I need to adjust the hemodynamic data of the patient with the medical record number AAA?"), the preoperative medical record data, and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model sets hemodynamic targets based on the patient's condition and the type of surgery: ordinary patients need to maintain a MAP ≥ 55-65 mmHg; elderly hypertensive patients need to appropriately increase the MAP target to 65-70 mmHg; and the ideal range of central venous pressure is set to 5-12 mmHg;

[0128] When MAP < target value: If central venous pressure < 5 mmHg, it is determined to be hypovolemia, and the first operation information for adjusting the dose of vasoactive drugs is output: "Rapidly infuse 200-300 ml of crystalloid solution, to be completed within 10-15 minutes";

[0129] When the central venous pressure is ≥5 mmHg and the MAP is still not up to standard after fluid infusion, and the focused heart rate is <60 beats / min, the first operation information for adjusting the dose of vasoactive drugs is output: "intravenous infusion of a small dose of norepinephrine 0.01-0.05 μg / kg / min, while monitoring the changes in the focused heart rate";

[0130] When the focused heart rate is ≥100 beats / min, the first operation information regarding adjusting the dosage of vasoactive drugs is outputted: "Prioritizely check cardiac function and consider using positive inotropic drugs such as dobutamine";

[0131] When the focused heart rate is >100 beats / min and the MAP decreases, if the central venous pressure is normal, it indicates heart failure, and the first operational information about adjusting the dose of vasoactive drugs is output: "Assess cardiac output and use milrinone to enhance myocardial contractility if necessary";

[0132] When the focused heart rate is less than 50 beats / min and the MAP decreases, the first operation information about adjusting the dosage of the vasoactive drug is outputted: "intravenous injection of atropine 0.5-1 mg, and observation of the heart rate recovery."

[0133] When the mixed venous oxygen saturation is less than 95%, if the central venous pressure is elevated and the focused heart rate is increased, it indicates the possible presence of pulmonary edema, and the first operation information about adjusting the dosage of vasoactive drugs is output: "Adjust respiratory parameters and give furosemide diuresis if necessary."

[0134] Illustratively, an anesthesiologist during surgery will use the first operation information as auxiliary decision-making information, combined with the patient's real-time clinical signs, to determine the second operation information for adjusting the anesthesia equipment;

[0135] Specifically, through the visual operation interface, the anesthesiologist will see the first operation information output by the anesthesia operation consultation model. The anesthesiologist will determine the final second operation information based on the first operation information and the patient's current real-time clinical signs, and input the second operation information, so that the anesthesia equipment control system receives the second operation information and adjusts the operating parameters of the corresponding anesthesia equipment in the working state according to the second operation information;

[0136] For example, the anesthesia equipment includes an anesthetic drug infusion equipment, and the second operation information includes operation information for adjusting the dosage of intravenous anesthetics. The anesthesia equipment control system will automatically send a new infusion rate instruction to the microinjection pump and synchronously update the drug usage data in the surgical anesthesia management system.

[0137] See Figure 2 , is an anesthesia equipment control system based on a large model provided by an embodiment of the present invention, comprising: a data acquisition module, a preoperative medical record acquisition module, a physiological monitoring index acquisition module, and an anesthesia equipment control module;

[0138] The data acquisition module is used to obtain the patient's medical record number and the intraoperative anesthesia operation problem to be decided;

[0139] The preoperative medical record acquisition module is used to extract the patient's preoperative medical record from a preset hospitalization system according to the medical record number;

[0140] The physiological monitoring index acquisition module is used to extract the patient's intraoperative real-time physiological monitoring index from a preset surgical anesthesia management system according to the medical record number;

[0141] The anesthesia equipment control module is used to input the preoperative medical history data, the intraoperative real-time physiological monitoring indicators and the anesthesia operation problems into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesia doctor during the operation can control the anesthesia equipment in a working state according to the first operation information; wherein, the anesthesia operation consultation model is obtained after training with a pre-labeled anesthesia data set as the input of the open source large model to be fine-tuned, and with theoretical anesthesia operations as the output of the open source large model to be fine-tuned.

[0142] In a preferred embodiment, the training process of the anesthesia operation consultation model includes:

[0143] Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records;

[0144] The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model;

[0145] The open source large model after iterative training is used as the anesthesia operation consultation model;

[0146] In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges.

[0147] If yes, stop training and get the open source model after training.

[0148] If not, adjust the parameters of the current open source large model to obtain the open source large model for the next iterative training.

[0149] In a preferred embodiment, the type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history;

[0150] The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes:

[0151] The anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information about the anesthetic dosage based on the bispectral index, the EEG state index, the patient's age, the patient's ASA grade and the past medical history.

[0152] See also Figure 3 ,In a preferred embodiment, the anesthesia equipment control system based on the large model further ,includes: a voice type conversion module;

[0153] The voice type conversion module is used to identify the information type of the intraoperative anesthesia operation problem to be decided; and when it is identified that the information type of the intraoperative anesthesia operation problem to be decided is a voice type, call a preset voice recognition algorithm to convert the information type of the intraoperative anesthesia operation problem into a text type.

[0154] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A large-scale model-based anesthesia equipment control method, applicable to an anesthesia equipment control system, characterized in that: include: Obtain the patient's medical record number and the intraoperative anesthesia procedure to be decided; Extracting the patient's preoperative medical records from a preset hospitalization system according to the medical record number; Extracting the patient's intraoperative real-time physiological monitoring indicators from a preset surgical anesthesia management system according to the medical record number; The preoperative medical records, the intraoperative real-time physiological monitoring indicators and the anesthesia operation problems are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesia doctor during the operation can regulate the anesthesia equipment in working state according to the first operation information; wherein, the anesthesia operation consultation model is obtained after training with a pre-labeled anesthesia data set as the input of the open source large model to be fine-tuned, and with theoretical anesthesia operations as the output of the open source large model to be fine-tuned.

2. The large model-based anesthesia equipment control method according to claim 1, characterized in that: The training process of the anesthesia operation consultation model includes: Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records; The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model; The open source large model after iterative training is used as the anesthesia operation consultation model; In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges. If yes, stop training and get the open source model after training. If not, adjust the parameters of the current open source large model to obtain the open source large model for the next iterative training.

3. The large model-based anesthesia equipment control method according to claim 2, characterized in that: The type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history; The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes: The anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information about the anesthetic dosage based on the bispectral index, the EEG state index, the patient's age, the patient's ASA grade and the past medical history.

4. The large model-based anesthesia equipment control method according to claim 3, characterized in that: The type of anesthesia operation problem includes the type of muscle relaxation management operation; the intraoperative real-time physiological monitoring index includes the real-time four-stimulation ratio; the preoperative medical history data includes the type of surgery; The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes: The anesthetic operation problem based on the muscle relaxant management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model determines the four-time train-of-stimulation ratio range according to the patient's age, the patient's ASA grade, the past medical history and the type of surgery, and outputs the first operation information about the muscle relaxant dosage based on the real-time four-time train-of-stimulation ratio and the four-time train-of-stimulation ratio range.

5. The large model-based anesthesia equipment control method according to claim 4, characterized in that: The types of anesthesia operation problems include hemodynamic management operation types; the intraoperative real-time physiological monitoring indicators include focused heart rate, blood pressure, blood oxygen saturation, central venous pressure, cardiac output, cardiac index, and mixed venous oxygen saturation; The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes: The anesthesia operation problem based on the hemodynamic management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information on adjusting the dosage of vasoactive drugs according to the patient's age, the patient's ASA grade, the past medical history, the type of surgery, the focused heart rate, the blood pressure, the blood oxygen saturation, the central venous pressure, the cardiac output, the cardiac index and the mixed venous oxygen saturation.

6. The large model-based anesthesia equipment control method according to claim 1, characterized in that: After obtaining the intraoperative anesthesia operation issues to be decided, it also includes: Identify the information type of the intraoperative anesthesia operation problem to be decided; When it is identified that the information type of the intraoperative anesthesia operation question to be decided is a voice type, a preset voice recognition algorithm is called to convert the information type of the intraoperative anesthesia operation question into a text type.

7. A large-scale model-based anesthesia equipment control system, characterized in that: include: Data acquisition module, preoperative medical record acquisition module, physiological monitoring index acquisition module and anesthesia equipment control module; The data acquisition module is used to obtain the patient's medical record number and the intraoperative anesthesia operation problem to be decided; The preoperative medical record acquisition module is used to extract the patient's preoperative medical record from a preset hospitalization system according to the medical record number; The physiological monitoring index acquisition module is used to extract the patient's intraoperative real-time physiological monitoring index from a preset surgical anesthesia management system according to the medical record number; The anesthesia equipment control module is used to input the preoperative medical history data, the intraoperative real-time physiological monitoring indicators and the anesthesia operation problems into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information corresponding to the intraoperative anesthesia operation problem based on the input data, so that the anesthesia doctor during the operation can control the anesthesia equipment in a working state according to the first operation information; wherein, the anesthesia operation consultation model is obtained after training with a pre-labeled anesthesia data set as the input of the open source large model to be fine-tuned, and with theoretical anesthesia operations as the output of the open source large model to be fine-tuned.

8. The large model-based anesthesia equipment control system according to claim 7, characterized in that: The training process of the anesthesia operation consultation model includes: Obtain a pre-labeled anesthesia dataset and an open-source large-scale model to be fine-tuned; wherein the open-source large-scale model is a Transformer architecture; the anesthesia dataset includes anesthesia expert knowledge data, anesthesia operation questions, patient diagnosis and treatment data, and actual anesthesia operation records; The open-source big model to be fine-tuned is iteratively trained using the anesthesia expert knowledge data, the anesthesia operation problems, the patient diagnosis and treatment data, and the actual anesthesia operation records as inputs to the open-source big model and theoretical anesthesia operations as outputs of the open-source big model; The open source large model after iterative training is used as the anesthesia operation consultation model; In each iterative training process, the theoretical anesthesia operation predicted by the current open source large model is compared with the corresponding actual anesthesia operation record, and the loss function value is calculated according to the comparison result; it is judged whether the loss function value converges. If yes, stop training and get the open source model after training. If not, adjust the parameters of the current open source large model to obtain the open source large model for the next iterative training.

9. The large model-based anesthesia equipment control system according to claim 8, characterized in that: The type of anesthesia operation problem includes the type of anesthesia depth management operation; the intraoperative real-time physiological monitoring indicators include the bispectral index and the electroencephalogram state index; the preoperative medical history data includes the patient's age, the patient's ASA grade, and past medical history; The step of inputting the preoperative medical history data, the intraoperative real-time physiological monitoring index, and the anesthesia operation question into the anesthesia operation consultation model so that the anesthesia operation consultation model outputs first operation information corresponding to the intraoperative anesthesia operation question based on the input data includes: The anesthesia operation problem based on the anesthesia depth management operation type, the preoperative medical history data and the intraoperative real-time physiological monitoring indicators are input into the anesthesia operation consultation model, so that the anesthesia operation consultation model outputs the first operation information about the anesthetic dosage based on the bispectral index, the EEG state index, the patient's age, the patient's ASA grade and the past medical history.

10. The large model-based anesthesia equipment control system according to claim 9, characterized in that: Also includes: Voice type conversion module; The voice type conversion module is used to identify the information type of the intraoperative anesthesia operation question to be decided; When it is identified that the information type of the intraoperative anesthesia operation question to be decided is a voice type, a preset voice recognition algorithm is called to convert the information type of the intraoperative anesthesia operation question into a text type.

Images