Clinical risk management method and apparatus based on large model and ant colony optimization

Abstract

The present application relates to the technical field of medical treatment, and relates to a clinical risk management method and apparatus based on a large model and ant colony optimization. The method comprises: acquiring medical text data, and calling a medical large model to process the medical text data, so as to extract diagnosis and treatment behavior text from the medical text data; calling AI identification technology to identify diagnosis and treatment behavior text from current medical record text data, and calling ant colony optimization to generate a diagnosis and treatment scheme on the basis of the diagnosis and treatment behavior text, so as to dynamically evaluate a current clinical risk problem; modeling the clinical risk problem into a mathematical model matching the ant colony optimization, and representing the current medical record text data by means of the ant colony optimization; initializing ant positions and ant paths, and selecting an ant path by means of ants and on the basis of a heuristic function and a pheromone concentration, so as to update pheromone on the ant path; and acquiring an optimal solution of the ant colony optimization by means of updating the pheromone on the ant path, and calling the medical large model to process the optimal solution, so as to evaluate the optimal solution.

Description

Method and device for clinical risk management based on large model and ant colony algorithm

[0001] The present application claims priority to the Chinese patent application No. 202410406928.3, filed on April 7, 2024, and entitled "Method and device for clinical risk management based on large model and ant colony algorithm", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the technical field of medical treatment, in particular to a method and device for clinical risk management based on a large model and an ant colony algorithm. BACKGROUND

[0003] Ant Colony Optimization (ACO) is a swarm intelligence optimization algorithm that simulates the process of a colony of ants searching for food. This algorithm was first proposed by the Italian computer scientist Marco Dorigo in 1992, inspired by the cooperative behavior of ants in searching for food. Because this algorithm can be used to solve complex optimization problems, it is currently widely used in various fields. For example, it is applied to emergency power management systems and intelligent patrols, topology optimization methods and systems, and medical technology management based on ant colony algorithm optimization.

[0004] First, in some cases, the intelligent patrol of the emergency power management system based on the ant colony algorithm mainly adopts an emergency power management system based on the ant colony algorithm, which includes a monitoring host and a plurality of lamp units. Each lamp unit includes a power node. Each power node includes a lamp controller and a corresponding power information acquisition module. The lamp controller is used to receive control information sent by the monitoring host and control the power information acquisition module. The power information acquisition module is used to collect power data information and then upload the collected power data information to the monitoring host through the CAN bus. The monitoring host is used to perform intelligent patrol based on the ant colony algorithm on the power node and analyze the state of charge of the power node, and display it on the user interface. The monitoring host determines the patrol path of the power node based on the ant colony algorithm using the active patrol method combined with the periodic timing patrol method, analyzes the state of charge of the power node according to the power data information uploaded by the power node, and displays it on the user interface.

[0005] Secondly, in some cases, the topology optimization method and system based on the ant colony algorithm mainly determines the initial finite element structure according to engineering needs, constructs a topology optimization model with discrete density as the design variable, constructs the unit conversion rule of the topology structure according to the node conversion rule of the ant colony algorithm, determines the density of each unit in the topology optimization model, and obtains the optimized topology structure.

[0006] Again, in some cases, the medical technology management technology based on the ant colony algorithm acquires the medical operation name of the target medical technology and the target monitoring point, determines the monitoring depth of the medical technology core point according to the medical operation name and the preset ant colony algorithm model. The monitoring path of the target monitoring point is determined according to the monitoring depth, the medical operation name and the preset ant colony algorithm model, and the management of the target medical technology is completed according to the monitoring path and the monitoring depth.

[0007] However, the above three cases have strong dependence on the intelligence level of the ant colony algorithm itself, so the above three methods have high requirements for model design, and the selection of objective function, constraint condition and decision variable will directly affect the final result. However, in the medical field, clinical practice frequently occurs, and hospital managers urgently need corresponding technical means to reduce the probability of risk events, and in some cases, it is difficult to meet the needs of the medical field only by relying on the intelligence level of the ant colony algorithm itself. SUMMARY

[0008] Therefore, it is necessary to provide a clinical risk management method and device based on a large model and an ant colony algorithm, which combines the ant colony algorithm with a large model to improve the intelligence level of the ant colony algorithm.

[0009] The present application provides a clinical risk management method based on a large model and an ant colony algorithm, which comprises:

[0010] Acquiring medical text data and calling a medical large model to process the medical text data to extract diagnosis and treatment behavior text in the medical text data, wherein the medical large model is obtained by fine-tuning an open source large model based on historical medical text;

[0011] Calling an AI recognition technology to recognize the diagnosis and treatment behavior text in the current medical record text data, and calling an ant colony algorithm to generate a diagnosis and treatment scheme matched with the current medical record text data based on the diagnosis and treatment behavior text, wherein the diagnosis and treatment scheme is used to dynamically evaluate the clinical risk problem of the current medical record text data;

[0012] Modeling the clinical risk problem into a mathematical model suitable for the ant colony algorithm to determine the objective function, constraint condition and decision variable of the clinical risk problem, and representing the current medical record text data through the ant colony algorithm;

[0013] Based on the representation of the current medical record text data through the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path, wherein the ant path is different diagnosis and treatment schemes;

[0014] An optimal solution or an approximate optimal solution of the ant colony algorithm is obtained by updating pheromone on the ant path, and the medical large model is called to process the optimal solution or the approximate optimal solution, so as to evaluate the optimal solution or the approximate optimal solution, and the optimal solution or the approximate optimal solution includes an optimal diagnosis and treatment scheme.

[0015] In an exemplary embodiment, the medical text data is obtained, and the medical large model is called to process the medical text data to extract diagnosis and treatment behavior text in the medical text data, and the foregoing includes:

[0016] The historical medical text is obtained, and a training data set of an open source large model is constructed based on the historical medical text;

[0017] The open source large model is trained by the training data set to fine-tune the open source large model, and the medical large model is obtained.

[0018] In an exemplary embodiment, the clinical risk problem is modeled into a mathematical model adapted to the ant colony algorithm to determine an objective function, a constraint condition and a decision variable of the clinical risk problem, and the current medical record text data is represented by the ant colony algorithm, including:

[0019] The clinical risk problem is modeled into a mathematical model adapted to the ant colony algorithm, and the mathematical model includes the objective function, the constraint condition and the decision variable, the objective function includes minimized diagnosis and treatment risk of a patient, and the constraint condition includes physical condition of the patient and availability of diagnosis and treatment resources;

[0020] The ant colony algorithm is called to perform graph representation on the current medical record text data, and the graph includes nodes and edges, the nodes are used to represent diagnosis and treatment schemes, and the edges are used to represent relationships between different diagnosis and treatment schemes.

[0021] In an exemplary embodiment, based on the representation of the current medical record text data by the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by an ant according to a heuristic function and pheromone concentration to update pheromone on the ant path, including:

[0022] The ant colony algorithm is parameter configured, and the parameter configuration at least includes an ant number in the ant colony algorithm, a pheromone evaporation rate and a heuristic function;

[0023] The ant path is selected by an ant according to the heuristic function and the pheromone concentration to obtain a current solution of the ant colony algorithm based on the current medical record text data, and the medical large model is called to evaluate the current solution to obtain an evaluation result.

[0024] In an example embodiment, the initialization of the ant position and the ant path based on the representation of the current medical record text data by the ant colony algorithm, and the selection of the ant path by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path further comprises:

[0025] When the evaluation result does not meet the set expectation, the ant path is adjusted, and the ant path is selected again by the ant colony algorithm according to the heuristic function and the pheromone concentration to guide the ant to find the optimal solution or the approximate optimal solution;

[0026] The selection of the ant path by the ant colony algorithm according to the heuristic function and the pheromone concentration is iterated multiple times to obtain the optimal solution or the approximate optimal solution of the ant colony algorithm based on the current medical record text data, and the medical large model based on the optimal solution or the approximate optimal solution meets the evaluation of the set expectation.

[0027] In an example embodiment, the optimal solution or the approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution, comprising:

[0028] The medical large model is called to analyze the current medical text data to obtain the diagnosis and treatment behavior text in the current medical text data;

[0029] The optimal solution or the approximate optimal solution is evaluated by the medical large model based on the diagnosis and treatment behavior text as a baseline to obtain an optimal solution model evaluation result.

[0030] In an embodiment, the method further comprises:

[0031] The current medical text data is identified by AI recognition technology to obtain the clinical risk event and patient satisfaction of the current medical text data;

[0032] The optimal solution or the approximate optimal solution is evaluated based on the clinical risk event and patient satisfaction to obtain an optimal solution AI evaluation result.

[0033] The application also provides a clinical risk management device based on a large model and an ant colony algorithm, the device comprising:

[0034] A model processing module is configured to obtain medical text data and call a medical large model to process the medical text data to extract diagnosis and treatment behavior text in the medical text data, wherein the medical large model is obtained by fine-tuning a historical medical text based on an open source large model;

[0035] an AI recognition module configured to invoke an AI recognition technology to recognize the diagnosis and treatment behavior text in the current medical record text data, and invoke an ant colony algorithm to generate a diagnosis and treatment scheme matching the current medical record text data based on the diagnosis and treatment behavior text, the diagnosis and treatment scheme being used to dynamically evaluate a clinical risk problem of the current medical record text data;

[0036] a mathematical modeling module configured to model the clinical risk problem into a mathematical model adapted to the ant colony algorithm to determine an objective function, a constraint condition, and a decision variable of the clinical risk problem, and represent the current medical record text data through the ant colony algorithm;

[0037] a pheromone updating module configured to initialize an ant position and an ant path based on the representation of the current medical record text data through the ant colony algorithm, and select the ant path according to a heuristic function and a pheromone concentration by an ant to update the pheromone on the ant path, the ant path being different diagnosis and treatment schemes;

[0038] an optimal solution evaluation module configured to obtain an optimal solution or an approximate optimal solution of the ant colony algorithm by updating the pheromone on the ant path, and invoke the medical large model to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution, the optimal solution or the approximate optimal solution including a best diagnosis and treatment scheme.

[0039] The application further provides an electronic device including a memory and a processor, the memory storing a computer program, and the processor implementing the clinical risk management method based on a large model and an ant colony algorithm as described above when executing the computer program.

[0040] The application further provides a computer storage medium storing a computer program, the computer program being executed by a processor to implement the clinical risk management method based on a large model and an ant colony algorithm as described above.

[0041] The application further provides a computer program product including a computer program, the computer program being executed by a processor to implement the clinical risk management method based on a large model and an ant colony algorithm as described above.

[0042] According to the embodiments provided in the application, the application has the following technical effects:

[0043] The application provides a clinical risk management method and device based on a large model and an ant colony algorithm. Medical text data is obtained, and a fine-tuned medical large model is called to process the medical text data to extract diagnosis and treatment behavior text in the medical text data. Then, AI recognition technology is called to identify the diagnosis and treatment behavior text in the current medical record text data, and an ant colony algorithm is called to generate a diagnosis and treatment scheme matching the current medical record text data based on the diagnosis and treatment behavior text to dynamically evaluate the clinical risk problem of the current medical record text data. Then, the clinical risk problem is modeled into a mathematical model suitable for the ant colony algorithm to determine the objective function, constraint condition and decision variable of the clinical risk problem, and the current medical record text data is represented by the ant colony algorithm. The ant position and ant path are initialized, and the ants select the ant path according to the heuristic function and pheromone concentration to update the pheromone on the ant path, which is a different diagnosis and treatment scheme. Finally, the optimal solution or approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution. This method applies the ant colony algorithm to the complex business scenario of clinical risk management, can fully exploit the advantages of the ant colony algorithm, customize the diagnosis and treatment path suitable for each diagnosis and treatment scenario, and constrain the diagnosis and treatment behavior of the clinician through the diagnosis and treatment path, thereby ensuring that the clinical risk control is within a small range. At the same time, the large model technology is integrated into the optimization process of the ant colony algorithm, which can significantly improve the algorithm effect compared with the existing ant colony algorithm, so that the ant colony algorithm can more accurately identify and solve the clinical risk in the diagnosis and treatment scenario, thereby providing a strong technical foundation for managing and controlling clinical risk. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0045] FIG. 1 is a schematic diagram of a clinical risk management method based on a large model and an ant colony algorithm according to one or more embodiments;

[0046] FIG. 2 is a schematic diagram of the overall process of the clinical risk management method based on a large model and an ant colony algorithm in a specific embodiment according to one or more embodiments;

[0047] FIG. 3 is a schematic diagram of a clinical risk management method based on a large model and an ant colony algorithm according to one or more embodiments;

[0048] FIG. 4 is a third schematic diagram of a method for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0049] FIG. 5 is a fourth schematic diagram of a method for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0050] FIG. 6 is a fifth schematic diagram of a method for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0051] FIG. 7 is a sixth schematic diagram of a method for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0052] FIG. 8 is a seventh schematic diagram of a method for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0053] FIG. 9 is a schematic diagram of a device structure for clinical risk management based on a large model and an ant colony algorithm according to one or more embodiments;

[0054] FIG. 10 is a schematic diagram of an internal structure of a computer device according to one or more embodiments. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0056] The method and device for clinical risk management based on a large model and an ant colony algorithm of the present application will be described below with reference to FIGS. 1-10.

[0057] As shown in FIG. 1, in one embodiment, a method for clinical risk management based on a large model and an ant colony algorithm includes the following steps:

[0058] In step S110, medical text data is obtained, and a medical large model is called to process the medical text data to extract diagnosis and treatment behavior text in the medical text data. The medical large model is obtained by fine-tuning an open-source large model based on historical medical text.

[0059] In one embodiment, a server obtains medical text data, and calls a medical large model obtained by fine-tuning an open-source large model based on a large amount of historical medical text to process the medical text data, to complete the extraction of diagnosis and treatment behavior text in the medical text data.

[0060] In combination with FIG. 2, in a specific embodiment, the clinical risk management method based on a large model and an ant colony algorithm provided by the present application first extracts a clinical path based on a large model. Based on an open-source large model, a large model in the medical field with better effect is fine-tuned. Through the domain large model, the diagnosis and treatment path / treatment plan of the target clinical problem in the guideline literature / high-quality medical record is extracted as the baseline for subsequent evaluation of the effect of the ant colony algorithm model.

[0061] In step S120, an AI recognition technology is called to recognize the diagnosis and treatment behavior text in the current medical record text data, and an ant colony algorithm is called to generate a diagnosis and treatment plan matched with the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment plan is used to dynamically evaluate the clinical risk problem of the current medical record text data.

[0062] In one embodiment, the server calls an AI recognition technology to recognize the diagnosis and treatment behavior text in the current medical record text data, and calls an ant colony algorithm to generate a diagnosis and treatment plan matched with the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment plan is used to dynamically evaluate the clinical risk problem of the current medical record text data.

[0063] In combination with FIG. 2, in a specific embodiment, the clinical risk management method based on a large model and an ant colony algorithm provided by the present application, in the process of defining the problem, the clinical risk problem to be solved is defined clearly, such as the risk problem caused by non-standard diagnosis and treatment. The diagnosis and treatment behavior of the current doctor can be recognized based on AI technology, and an adaptive diagnosis and treatment plan can be generated according to the ant colony algorithm to dynamically evaluate the diagnosis and treatment risk. For example, a personalized treatment plan is designed for the patient according to the patient's medical history, symptoms, lifestyle, and other information.

[0064] In step S130, the clinical risk problem is modeled into a mathematical model adapted to the ant colony algorithm to determine the objective function, constraint condition, and decision variable of the clinical risk problem, and the current medical record text data is represented through the ant colony algorithm.

[0065] In one embodiment, the server models the clinical risk problem into a mathematical model adapted to the ant colony algorithm to determine the objective function, constraint condition, and decision variable of the clinical risk problem in the mathematical model, and represents the current medical record text data through the ant colony algorithm.

[0066] In combination with FIG. 2, in a specific embodiment, in the modeling process, the clinical problem is modeled into a mathematical model suitable for the ant colony algorithm, and the objective function, constraint conditions and decision variables of the problem are determined. For example, the objective function can be to minimize the treatment risk of the patient, and the constraint conditions can include the physical condition of the patient, the availability of treatment resources, etc. In the information representation, the information of the patient is represented in a manner suitable for the ant colony algorithm. A graphical representation can be used, in which nodes represent possible treatment decisions and edges represent the relationship between different decisions.

[0067] In this embodiment, ant colony algorithm parameter adaptation is also required, that is, the parameters of the ant colony algorithm are set, including the number of ants, the pheromone evaporation rate, the heuristic function, etc. The configuration of these parameters depends on the specific clinical problem and data. Among them, the configuration logic of the number of ants is: if it is a complex diagnosis and treatment problem, increasing the number of ants can help to more comprehensively explore the solution space, but also requires more computing resources, in a simple scenario, a smaller number of ants is selected because a small number of ants can be sufficient to quickly search the solution space. Volatile coefficient configuration: in a low complexity scenario, a lower volatile coefficient helps to maintain information transmission for a longer period of time, and it is easier to converge to a better solution. In a high complexity scenario, that is, in a more complex problem, moderately increasing the volatile coefficient helps to prevent premature falling into local optimum and encourages ants to explore new paths.

[0068] In step S140, based on the representation of the current medical record text data by the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path, and the ant path is different diagnosis and treatment scheme.

[0069] In one embodiment, the server initializes the ant position and the ant path based on the representation of the current medical record text data by the ant colony algorithm in step S130, and selects the ant path by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path, and the ant path is different diagnosis and treatment scheme.

[0070] In combination with FIG. 2, in a specific embodiment, in the ant colony algorithm initialization process, the position and path of the ant need to be initialized, and in the context, the path can represent different treatment schemes. The ant selects the path according to the heuristic function and the pheromone concentration, and updates the pheromone on the path. Iterative optimization can also be performed by updating the pheromone on the path. Through multiple iterations, the ant colony algorithm continuously updates the pheromone and adjusts the path selection of the ant, gradually optimizing the solution of the problem. In each iteration, the quality of the current solution is evaluated, and the pheromone is updated, which can guide the ant to find a better solution.

[0071] In step S150, the optimal solution or approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution, and the optimal solution or approximate optimal solution includes the best diagnosis and treatment scheme.

[0072] In one embodiment, the server obtains the optimal solution or approximate optimal solution of the ant colony algorithm by updating the pheromone on the ant path, and calls the medical large model to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution, and the optimal solution or approximate optimal solution is the best diagnosis and treatment scheme.

[0073] In combination with FIG. 2, in a specific embodiment, in the process of result evaluation, the optimal solution or approximate optimal solution obtained by the ant colony algorithm is analyzed, and the quality of the treatment scheme is evaluated, including risk reduction, patient satisfaction, etc. The diagnosis and treatment path / scheme obtained by the medical large model is used to evaluate the best treatment scheme of the ant colony algorithm. At the same time, the AI technology can also be used to identify the risk events and patient satisfaction in the medical record, such as the adverse reaction of treatment, etc., to evaluate the effect of the ant colony algorithm treatment scheme. In addition, according to the feedback in actual application, the related parameters and models can be adjusted to improve the performance and adaptability of the ant colony algorithm.

[0074] In this embodiment, the personalized treatment scheme making in clinical risk management may involve multiple factors, including the health status of the patient, medical history, lifestyle, genetic information, etc. The ant colony algorithm simulates the collective intelligence of ants in searching for the optimal path, comprehensively considers these complex factors, and generates a personalized treatment scheme. However, in actual application, close cooperation with clinical professionals is needed to ensure that the treatment scheme generated by the algorithm meets the medical standards and the actual needs of patients, so the method can identify risk events in the diagnosis and treatment scene and remind the doctor in the doctor workstation in the form of quality control, so that the doctor can confirm it, and also provide an interactive interface for the doctor to comment on the risk reminder effect for subsequent effect optimization.

[0075] The above clinical risk management method based on the large model and the ant colony algorithm acquires medical text data, and calls the fine-tuned medical large model to process the medical text data to extract the diagnosis and treatment behavior text in the medical text data. Subsequently, the AI recognition technology is called to identify the diagnosis and treatment behavior text in the current medical record text data, and the ant colony algorithm is called to generate a diagnosis and treatment scheme matched with the current medical record text data based on the diagnosis and treatment behavior text to dynamically evaluate the clinical risk problem of the current medical record text data. Then, the clinical risk problem is modeled into a mathematical model suitable for the ant colony algorithm to determine the objective function, constraint condition and decision variable of the clinical risk problem, and the current medical record text data is represented by the ant colony algorithm. The ant position and ant path are initialized, and the ants select the ant path according to the heuristic function and pheromone concentration to update the pheromone on the ant path, and the ant path is the different diagnosis and treatment scheme. Finally, the optimal solution or approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution. This method applies the ant colony algorithm to the complex business scenarios of clinical risk management, can fully tap the advantages of the ant colony algorithm, customize the diagnosis and treatment path suitable for each diagnosis and treatment scene, and constrain the diagnosis and treatment behavior of the clinician through the diagnosis and treatment path, so as to ensure that the clinical risk control is within a small range. At the same time, the large model technology is integrated into the optimization process of the ant colony algorithm, compared with the existing ant colony algorithm, the algorithm effect can be significantly improved, so that the ant colony algorithm can more accurately identify and solve the clinical risk in the diagnosis and treatment scene, thereby providing a strong technical foundation for controlling the clinical risk.

[0076] As shown in FIG. 3, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application acquires medical text data, and calls the medical large model to process the medical text data to extract the diagnosis and treatment behavior text in the medical text data, which includes the following steps:

[0077] Step S310, acquire historical medical text, and construct a training data set of an open source large model based on the historical medical text.

[0078] In one embodiment, the server acquires a large amount of historical medical text, and constructs a training data set of an open source large model based on the historical medical text.

[0079] Step S320, train the open source large model through the training data set to fine-tune the open source large model, and obtain a medical large model.

[0080] In one embodiment, the server fine-tunes the open-source large model by training the open-source large model through the training data set built in step S310, so that the open-source large model learns to understand the application field and different medical scenarios, and obtains a fine-tuned medical large model.

[0081] As shown in FIG. 4, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application models the clinical risk problem into a mathematical model suitable for the ant colony algorithm to determine the objective function, the constraint condition and the decision variable of the clinical risk problem, and represents the current medical record text data through the ant colony algorithm, specifically including the following steps:

[0082] Step S132, modeling the clinical risk problem into a mathematical model suitable for the ant colony algorithm, the mathematical model including an objective function, a constraint condition and a decision variable, the objective function including the minimum diagnosis and treatment risk of the patient, and the constraint condition including the physical condition of the patient and the availability of diagnosis and treatment resources.

[0083] In one embodiment, the server models the clinical risk problem into a mathematical model suitable for the ant colony algorithm, the mathematical model including an objective function, a constraint condition and a decision variable, the objective function including the minimum diagnosis and treatment risk of the patient, and the constraint condition including the physical condition of the patient and the availability of diagnosis and treatment resources.

[0084] Step S134, calling the ant colony algorithm to graphically represent the current medical record text data, the graph including nodes and edges, the nodes being used to represent diagnosis and treatment schemes, and the edges being used to represent the relationship between different diagnosis and treatment schemes.

[0085] In one embodiment, the server calls the ant colony algorithm to graphically represent the current medical record text data, the graph including nodes and edges, the nodes being used to represent diagnosis and treatment schemes, and the edges being used to represent the relationship between different diagnosis and treatment schemes.

[0086] As shown in FIG. 5, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application, based on the representation of the current medical record text data through the ant colony algorithm, initializes the ant position and the ant path, and selects the ant path according to the heuristic function and the pheromone concentration through the ants to update the pheromone on the ant path, specifically including the following steps:

[0087] Step S142, parameter configuration of the ant colony algorithm, the parameter configuration at least including the number of ants in the ant colony algorithm, the pheromone evaporation rate and the heuristic function.

[0088] In one embodiment, the server parameterizes the ant colony algorithm, and the parameter configuration at least includes the number of ants in the ant colony algorithm, the pheromone evaporation rate and the heuristic function.

[0089] Step S144, the ants select the ant path according to the heuristic function and pheromone concentration, obtain the current solution of the ant colony algorithm based on the current medical record text data, and call the medical large model to evaluate the current solution to obtain an evaluation result.

[0090] In one embodiment, the server obtains the current solution of the ant colony algorithm based on the current medical record text data by selecting the ant path according to the heuristic function and pheromone concentration by the ants in the ant colony algorithm, and calls the medical large model to evaluate the current solution to obtain a corresponding evaluation result.

[0091] As shown in FIG. 6, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application includes the following steps: initializing the ant position and the ant path based on representing the current medical record text data by the ant colony algorithm, and updating the pheromone on the ant path by the ants according to the heuristic function and the pheromone concentration to update the pheromone on the ant path.

[0092] Step S146, when the evaluation result does not meet the set expectation, the ant path is adjusted, and the ant path is selected again by the ant colony algorithm according to the heuristic function and the pheromone concentration to guide the ants to find the optimal solution or the approximate optimal solution.

[0093] In one embodiment, when the evaluation result of the current solution does not meet the set expectation, the server adjusts the ant path and selects the ant path again by the ant colony algorithm according to the heuristic function and the pheromone concentration to guide the ants to find the optimal solution or the approximate optimal solution.

[0094] Step S148, the ant path is selected by the ant colony algorithm according to the heuristic function and the pheromone concentration multiple times, the optimal solution or the approximate optimal solution of the ant colony algorithm based on the current medical record text data is obtained, and the medical large model meets the evaluation set expectation based on the optimal solution or the approximate optimal solution.

[0095] In one embodiment, the server selects the ant path by the ant colony algorithm according to the heuristic function and the pheromone concentration multiple times, guides the ants to find the optimal solution, and then obtains the optimal solution or the approximate optimal solution of the ant colony algorithm based on the current medical record text data, and the medical large model meets the evaluation set expectation based on the optimal solution or the approximate optimal solution.

[0096] As shown in FIG. 7, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application includes the following steps: obtaining the optimal solution or the approximate optimal solution of the ant colony algorithm by updating the pheromone on the ant path, and calling the medical large model to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution.

[0097] At step S152, the medical large model is called to analyze the current medical text data to obtain the diagnosis and treatment behavior text in the current medical text data.

[0098] In one embodiment, the server calls the medical large model to analyze the current medical text data to obtain the diagnosis and treatment behavior text in the current medical text data.

[0099] At step S154, the optimal solution or the approximate optimal solution is evaluated by the medical large model taking the diagnosis and treatment behavior text as a baseline to obtain an optimal solution model evaluation result.

[0100] In one embodiment, the server evaluates the optimal solution or the approximate optimal solution by the medical large model taking the diagnosis and treatment behavior text as a baseline to obtain a corresponding optimal solution model evaluation result.

[0101] As shown in FIG. 8, in one embodiment, the clinical risk management method based on the large model and the ant colony algorithm provided by the present application further includes the following steps:

[0102] At step S810, the current medical text data is identified by AI recognition technology to obtain the clinical risk event and the patient satisfaction of the current medical text data.

[0103] In one embodiment, the server identifies the current medical text data by AI recognition technology to obtain the clinical risk event and the patient satisfaction of the current medical text data, wherein the clinical risk event includes adverse reactions of the patient during the treatment process.

[0104] At step S820, the optimal solution or the approximate optimal solution is evaluated based on the clinical risk event and the patient satisfaction to obtain an optimal solution AI evaluation result.

[0105] In one embodiment, the server evaluates the optimal solution or the approximate optimal solution based on the clinical risk event and the patient satisfaction obtained in step S810 to obtain an optimal solution AI evaluation result.

[0106] The clinical risk management device based on the large model and the ant colony algorithm provided by the present application is described below, and the clinical risk management device based on the large model and the ant colony algorithm described below can be mutually corresponding to the clinical risk management method based on the large model and the ant colony algorithm described above.

[0107] As shown in FIG. 9, in one embodiment, a clinical risk management device based on a large model and an ant colony algorithm includes a model processing module 910, an AI recognition module 920, a mathematical modeling module 930, a pheromone updating module 940, and an optimal solution evaluation module 950.

[0108] The model processing module 910 is configured to acquire medical text data, and call a medical large model to process the medical text data to extract diagnosis and treatment behavior text in the medical text data. The medical large model is a medical large model that is obtained by fine-tuning an open source large model based on historical medical text.

[0109] The AI identification module 920 is configured to call an AI identification technology to identify diagnosis and treatment behavior text in current medical record text data, and call an ant colony algorithm to generate diagnosis and treatment schemes matched with the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment schemes are used to dynamically evaluate clinical risk problems of the current medical record text data.

[0110] The mathematical modeling module 930 is configured to model the clinical risk problems into a mathematical model suitable for the ant colony algorithm, to determine an objective function, a constraint condition and a decision variable of the clinical risk problems, and to represent the current medical record text data by the ant colony algorithm.

[0111] The pheromone updating module 940 is configured to initialize an ant position and an ant path based on the representation of the current medical record text data by the ant colony algorithm, and to update pheromones on the ant path by the ants according to a heuristic function and a pheromone concentration, the ant path being different diagnosis and treatment schemes.

[0112] The optimal solution evaluation module 950 is configured to obtain an optimal solution or an approximate optimal solution of the ant colony algorithm by updating the pheromones on the ant path, and to call the medical large model to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution, the optimal solution or the approximate optimal solution including a best diagnosis and treatment scheme.

[0113] In this embodiment, the clinical risk management device based on the large model and the ant colony algorithm provided by the present application further includes a model fine-tuning module, configured to:

[0114] acquire historical medical text, and construct a training data set of the open source large model based on the historical medical text.

[0115] train the open source large model by the training data set to fine-tune the open source large model, and obtain the medical large model.

[0116] In this embodiment, the clinical risk management device based on the large model and the ant colony algorithm provided by the present application, the mathematical modeling module is specifically configured to:

[0117] model the clinical risk problems into a mathematical model suitable for the ant colony algorithm, the mathematical model including an objective function, a constraint condition and a decision variable, the objective function including a minimized diagnosis and treatment risk of a patient, and the constraint condition including a physical condition of the patient and availability of diagnosis and treatment resources.

[0118] ​The ant colony algorithm is called to perform graph representation on the current medical record text data, the graph includes nodes and edges, the nodes are used to represent diagnosis and treatment schemes, and the edges are used to represent the relationship between different diagnosis and treatment schemes.

[0119] In the embodiment, the information updating module is specifically used for:

[0120] The ant colony algorithm is configured with parameters, and the parameters at least include the number of ants in the ant colony algorithm, the evaporation rate of the pheromone, and the heuristic function.

[0121] The ant path is selected according to the heuristic function and the pheromone concentration by the ant, the current solution of the ant colony algorithm based on the current medical record text data is obtained, and the current solution is evaluated by calling the medical large model to obtain an evaluation result.

[0122] In the embodiment, the information updating module is specifically used for:

[0123] When the evaluation result does not meet the set expectation, the ant path is adjusted, and the ant path is selected again according to the heuristic function and the pheromone concentration by the ant colony algorithm, so as to guide the ant to find the optimal solution or the approximate optimal solution.

[0124] The ant path is selected according to the heuristic function and the pheromone concentration by the ant colony algorithm through multiple iterations, the optimal solution or the approximate optimal solution of the ant colony algorithm based on the current medical record text data is obtained, and the evaluation set expectation is met based on the optimal solution or the approximate optimal solution by the medical large model.

[0125] In the embodiment, the optimal solution evaluation module is specifically used for:

[0126] The current medical text data is analyzed by calling the medical large model to obtain diagnosis and treatment behavior text in the current medical text data.

[0127] The optimal solution or the approximate optimal solution is evaluated by the medical large model taking the diagnosis and treatment behavior text as a baseline to obtain an optimal solution model evaluation result.

[0128] In the embodiment, the clinical risk management device based on the large model and the ant colony algorithm is further provided with an optimal solution evaluation submodule, which is used for:

[0129] The current medical text data is identified by AI recognition technology to obtain a clinical risk event and patient satisfaction of the current medical text data.

[0130] An optimal solution or an approximate optimal solution is evaluated based on clinical risk events and patient satisfaction to obtain an optimal solution AI evaluation result.

[0131] FIG. 10 illustrates a schematic diagram of a physical structure of an electronic device, which can be a smart terminal, and an internal structure diagram of the electronic device can be as shown in FIG. 10. The electronic device includes a processor, a memory, and a network interface connected through a system bus. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the electronic device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a clinical risk management method based on a large model and an ant colony algorithm, and the method includes:

[0132] Medical text data is obtained, and a medical large model is called to process the medical text data to extract diagnosis and treatment behavior text in the medical text data. The medical large model is a large open source model fine-tuned based on historical medical text.

[0133] An AI recognition technology is called to recognize diagnosis and treatment behavior text in the current medical record text data, and an ant colony algorithm is called to generate a diagnosis and treatment scheme matching the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment scheme is used to dynamically evaluate the clinical risk problem of the current medical record text data.

[0134] The clinical risk problem is modeled into a mathematical model suitable for the ant colony algorithm to determine the objective function, constraint condition, and decision variable of the clinical risk problem, and the current medical record text data is represented by the ant colony algorithm.

[0135] Based on the representation of the current medical record text data by the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path. The ant path is different diagnosis and treatment schemes.

[0136] The optimal solution or the approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution. The optimal solution or the approximate optimal solution includes the best diagnosis and treatment scheme.

[0137] Those skilled in the art can understand that the structure shown in FIG. 10 is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0138] In an embodiment, the present application also provides a computer storage medium storing a computer program, the computer program being executed by a processor to implement a clinical risk management method based on a large model and an ant colony algorithm, the method comprising:

[0139] Obtaining medical text data, and calling a medical large model to process the medical text data to extract diagnosis and treatment behavior text in the medical text data, the medical large model being obtained by fine-tuning an open-source large model based on historical medical text.

[0140] Calling an AI recognition technology to recognize diagnosis and treatment behavior text in the current medical record text data, and calling an ant colony algorithm to generate diagnosis and treatment schemes matched with the current medical record text data based on the diagnosis and treatment behavior text, the diagnosis and treatment schemes being used to dynamically evaluate clinical risk problems of the current medical record text data.

[0141] Modeling the clinical risk problems into a mathematical model suitable for the ant colony algorithm to determine an objective function, a constraint condition and a decision variable of the clinical risk problems, and representing the current medical record text data by the ant colony algorithm.

[0142] Initializing an ant position and an ant path based on representing the current medical record text data by the ant colony algorithm, and selecting the ant path by the ant according to a heuristic function and pheromone concentration to update pheromone on the ant path, the ant path being different diagnosis and treatment schemes.

[0143] Obtaining an optimal solution or an approximate optimal solution of the ant colony algorithm by updating the pheromone on the ant path, and calling the medical large model to process the optimal solution or the approximate optimal solution to evaluate the optimal solution or the approximate optimal solution, the optimal solution or the approximate optimal solution including a best diagnosis and treatment scheme.

[0144] In an embodiment, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions stored in a computer readable storage medium. A processor of an electronic device reads the computer instructions from the computer readable storage medium, and the processor implements a clinical risk management method based on a large model and an ant colony algorithm when executing the computer instructions, the method comprising:

[0145] Obtaining medical text data, and calling a medical large model to process the medical text data to extract diagnosis and treatment behavior text in the medical text data, the medical large model being obtained by fine-tuning an open-source large model based on historical medical text.

[0146] The AI recognition technology is called to recognize the diagnosis and treatment behavior text in the current medical record text data, and the ant colony algorithm is called to generate a diagnosis and treatment scheme matched with the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment scheme is used to dynamically evaluate the clinical risk problem of the current medical record text data.

[0147] The clinical risk problem is modeled into a mathematical model suitable for the ant colony algorithm to determine the objective function, constraint condition and decision variable of the clinical risk problem, and the current medical record text data is represented by the ant colony algorithm.

[0148] Based on the representation of the current medical record text data by the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by the ant according to the heuristic function and the pheromone concentration to update the pheromone on the ant path. The ant path is different diagnosis and treatment schemes.

[0149] The optimal solution or approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the medical large model is called to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution. The optimal solution or approximate optimal solution includes the best diagnosis and treatment scheme.

[0150] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory.

[0151] As an illustration but not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0152] Any combination of the technical features in the above embodiments can be made, and for the sake of brevity, not all possible combinations are described above, however, as long as the combination of the technical features does not exist in contradiction, it shall be considered within the scope of the present disclosure.

[0153] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it shall not be understood as a limitation on the patent scope of the present application. It shall be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these shall be within the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A clinical risk management method based on a large model and ant colony algorithm, characterized in that: The method comprises: Obtain medical text data and call a medical big model to process the medical text data to extract diagnosis and treatment behavior text from the medical text data. The medical big model is an open source big model fine-tuned based on historical medical text; Calling AI recognition technology to identify the diagnosis and treatment behavior text in the current medical record text data, and calling the ant colony algorithm to generate a diagnosis and treatment plan that matches the current medical record text data based on the diagnosis and treatment behavior text, and the diagnosis and treatment plan is used to dynamically evaluate the clinical risk issues of the current medical record text data; Modeling the clinical risk problem into a mathematical model adapted to the ant colony algorithm to determine the objective function, constraints, and decision variables of the clinical risk problem, and representing the current medical record text data through the ant colony algorithm; Based on the representation of the current medical record text data by the ant colony algorithm, the ant position and the ant path are initialized, and the ant path is selected by the ants according to the heuristic function and the pheromone concentration to update the pheromone on the ant path, and the ant path is different diagnosis and treatment plans; The optimal solution or approximate optimal solution of the ant colony algorithm is obtained by updating the pheromone on the ant path, and the optimal solution or approximate optimal solution is processed by calling the medical big model to evaluate the optimal solution or approximate optimal solution, and the optimal solution or approximate optimal solution includes the best diagnosis and treatment plan.

2. The clinical risk management method based on large model and ant colony algorithm according to claim 1, characterized in that: The obtaining of medical text data and calling a medical big model to process the medical text data to extract the diagnosis and treatment behavior text in the medical text data previously includes: Obtaining the historical medical texts, and constructing an open-source large-scale model training dataset based on the historical medical texts; The open source big model is trained using the training data set to fine-tune the open source big model to obtain the medical big model.

3. The clinical risk management method based on large model and ant colony algorithm according to claim 1, characterized in that: Modeling the clinical risk problem into a mathematical model adapted to the ant colony algorithm to determine the objective function, constraints, and decision variables of the clinical risk problem, and representing the current medical record text data by the ant colony algorithm, includes: Modeling the clinical risk problem into a mathematical model adapted to the ant colony algorithm, the mathematical model including the objective function, constraints, and decision variables, the objective function including minimizing the patient's diagnosis and treatment risk, and the constraints including the patient's physical condition and the availability of diagnosis and treatment resources; The ant colony algorithm is called to perform a graphical representation on the current medical record text data, where the graph includes nodes and edges, the nodes are used to represent diagnosis and treatment plans, and the edges are used to represent the relationship between different diagnosis and treatment plans.

4. The clinical risk management method based on large model and ant colony algorithm according to claim 1, characterized in that: The method of representing the current medical record text data by the ant colony algorithm, initializing ant positions and ant paths, and selecting the ant path by ants according to a heuristic function and pheromone concentration to update the pheromone on the ant path includes: Performing parameter configuration on the ant colony algorithm, wherein the parameter configuration includes at least the number of ants in the ant colony algorithm, the pheromone volatility rate, and the heuristic function; The ant colony algorithm selects the ant path according to the heuristic function and the pheromone concentration to obtain the current solution of the ant colony algorithm based on the current medical record text data, and the medical big model is called to evaluate the current solution to obtain an evaluation result.

5. The clinical risk management method based on large model and ant colony algorithm according to claim 4, characterized in that: The method of representing the current medical record text data by the ant colony algorithm, initializing ant positions and ant paths, and selecting the ant path by ants according to a heuristic function and pheromone concentration to update the pheromone on the ant path, further includes: When the evaluation result does not meet the set expectation, the ant path is adjusted, and the ant path is selected again by the ant colony algorithm according to the heuristic function and the pheromone concentration, so as to guide the ants to find the optimal solution or the approximate optimal solution; The ant colony algorithm selects the ant path through multiple iterations according to the heuristic function and the pheromone concentration to obtain the optimal solution or approximate optimal solution of the ant colony algorithm based on the current medical record text data, and the medical macro model evaluates the set expectations based on whether the optimal solution or approximate optimal solution satisfies the set expectations.

6. The clinical risk management method based on large model and ant colony algorithm according to claim 1, characterized in that: The obtaining of the optimal solution or approximate optimal solution of the ant colony algorithm by updating the pheromone on the ant path, and calling the medical macro model to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution includes: Calling the medical big model to analyze the current medical record text data to obtain the diagnosis and treatment behavior text in the current medical record text data; The optimal solution or approximate optimal solution is evaluated by the medical macro model with the diagnosis and treatment behavior text as a baseline to obtain an optimal solution model evaluation result.

7. The clinical risk management method based on a large model and ant colony algorithm according to any one of claims 1 to 6, characterized in that: The method further comprises: Identify the current medical record text data using AI recognition technology to obtain clinical risk events and patient satisfaction from the current medical record text data; The optimal solution or approximate optimal solution is evaluated based on the clinical risk events and patient satisfaction to obtain an optimal solution AI evaluation result.

8. A clinical risk management device based on a large model and ant colony algorithm, characterized in that: The device comprises: A model processing module is used to obtain medical text data and process the medical text data using a medical big model to extract the diagnosis and treatment behavior text in the medical text data. The medical big model is an open source big model fine-tuned based on historical medical text; An AI recognition module is configured to use AI recognition technology to identify the diagnosis and treatment behavior text in the current medical record text data, and to use an ant colony algorithm to generate a diagnosis and treatment plan that matches the current medical record text data based on the diagnosis and treatment behavior text. The diagnosis and treatment plan is used to dynamically assess the clinical risk issues of the current medical record text data; a mathematical modeling module, configured to model the clinical risk problem into a mathematical model adapted to the ant colony algorithm, so as to determine the objective function, constraints, and decision variables of the clinical risk problem, and to represent the current medical record text data through the ant colony algorithm; a pheromone updating module, configured to initialize ant positions and ant paths based on the representation of the current medical record text data by the ant colony algorithm, and to update the pheromones on the ant paths by selecting the ant paths according to the heuristic function and the pheromone concentration by the ants, wherein the ant paths represent different diagnosis and treatment plans; The optimal solution evaluation module is used to obtain the optimal solution or approximate optimal solution of the ant colony algorithm by updating the pheromones on the ant path, and call the medical big model to process the optimal solution or approximate optimal solution to evaluate the optimal solution or approximate optimal solution, wherein the optimal solution or approximate optimal solution includes the best diagnosis and treatment plan.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images