Intelligent auxiliary medication system for patient suffering from pyosis

Abstract

The invention discloses an intelligent pyosis patient auxiliary medication system. The system comprises a data acquisition module, a data expansion model building module, a nursing auxiliary model building module and a patient nursing module. The invention relates to the technical field of medication management, in particular to an intelligent auxiliary medication system for pyogenic patients, which is characterized in that a dynamic weight adjustment mechanism and clustering constraints based on a topological structure are introduced, and an improved distance metric function is designed, so that the clustering accuracy and adaptability are improved; more scientific and more reliable auxiliary support is provided for medication decision of sepsis patients; a parameter generation strategy and a parameter movement mechanism are designed, information intensity and driving parameters are introduced, an information adjustment coefficient and attenuation coefficient mechanism is combined, a structure parameter iteration mechanism is generated, a clustering structure can be dynamically optimized according to data features, local optimum is effectively avoided, the global search capability is enhanced, and the search efficiency is improved. And a more efficient and more reliable solution is provided for clustering analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of medication management, and specifically refers to an intelligent auxiliary medication system for septic patients. Background Technique

[0002] An intelligent auxiliary medication system for septic patients is a medical auxiliary tool innovatively generated based on artificial intelligence and big data technologies. It integrates the clinical data, physiological indicators, and laboratory test results of patients through intelligent algorithms, and uses advanced algorithms to achieve accurate medication recommendations and treatment response predictions, improving the treatment effect of septic patients.

[0003] Traditional clustering methods have problems of being insensitive to feature differences and difficult to adapt to dynamic changes when dealing with data of septic patients; traditional clustering methods have problems of low optimization efficiency, difficulty in dynamically adjusting parameters, and being prone to falling into local optima when dealing with complex data structures. Summary of the Invention

[0004] In view of the above situation, to overcome the defects of the prior art, the present invention provides an intelligent auxiliary medication system for septic patients. Aiming at the problems of being insensitive to feature differences and difficult to adapt to dynamic changes of traditional clustering methods when dealing with data of septic patients, this solution improves the flexibility and accuracy of clustering by introducing a dynamic weight adjustment mechanism; designs an improved distance metric function, enhances the sensitivity to large differences through a penalty function, and further improves the robustness of clustering; introduces clustering constraints based on topological structure, enabling the clustering process to fully consider the spatial topological relationship of data, significantly improving the accuracy and adaptability of clustering, and providing more scientific and reliable auxiliary support for the medication decision-making of septic patients; aiming at the problems of low optimization efficiency, difficulty in dynamically adjusting parameters, and being prone to falling into local optima of traditional clustering methods when dealing with complex data structures, this solution realizes the dynamic generation and optimization of parameters by introducing a clustering structure evolution module; designs a parameter generation strategy and a parameter movement mechanism, which can adaptively generate initial parameter points and dynamically adjust parameter values, thereby significantly improving the optimization efficiency and adaptability; by introducing information intensity and driving parameters, combining the mechanisms of information adjustment coefficient and attenuation coefficient, dynamically adjusts the search direction and speed of parameter points, effectively avoiding local optima and enhancing the global search ability; through a structure parameter iteration mechanism, it can dynamically optimize the clustering structure according to data characteristics, further improving the accuracy and robustness of clustering, and providing a more efficient and reliable solution for clustering analysis.

[0005] The technical solution adopted by the present invention is as follows: An intelligent auxiliary medication system for septic patients provided by the present invention includes a data acquisition module, an auxiliary medication clustering generation module, a clustering structure evolution module, and an auxiliary medication module;

[0006] The data acquisition module collects the attribute data, clinical physiological indicators, laboratory test data, and medication data of historical sepsis patients;

[0007] The adjuvant medication clustering generation module generates adjuvant medication clusters by neighborhood division, distance measurement, generation of dynamic weights, update of cluster centers, introduction of clustering constraints, and design of objective functions;

[0008] The clustering structure evolution module realizes the evolution of the clustering structure by determining structure parameters, designing parameter generation strategies, calculating information intensity, generating driving parameters, defining parameter movement speeds, parameter movement, and structure parameter iteration;

[0009] The adjuvant medication module collects the attribute data, clinical physiological indicators, and laboratory test data of patients, and uses adjuvant medication clustering to output predicted drugs and treatment responses to assist patients in taking medications.

[0010] Furthermore, the data acquisition module collects the attribute data, clinical physiological indicator data, laboratory test data, medication data, and patient status data of historical sepsis patients; the attribute data includes the age, gender, and medical history of the patient; the clinical physiological indicator data includes body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation; the laboratory test data includes the test data of white blood cell count, platelet count, lactate level, inflammatory markers, liver function, and kidney function; the medication data includes the medications already used and treatment responses.

[0011] Furthermore, the adjuvant medication clustering generation module specifically includes the following:

[0012] Neighborhood division: Calculate the average position of all data points, select 30% of the data points closest to the average position of the data points, calculate the Euclidean distance between the average position of the data points and the data point farthest from these 30% of the data points, set this distance as the neighborhood width, and divide the area within the neighborhood width into the neighborhood range;

[0013] Distance measurement: Considers the distance in space, increases the penalty for large differences through a logarithmic term, and sets dynamic feature weighting, which is expressed as follows:

[0014] ;

[0015] where i and j represent the indices of the data points, and respectively represent the positions of the i-th and j-th data points, represents the distance measurement function, represents the distance measurement between the i-th data point and the j-th data point, represents the Euclidean distance between the $i$-th data point and the $j$-th data point, 、 and represent the metric weights, $k$ represents the dimension index of the data points, $d$ represents the total number of dimensions of the data points, represents the logarithmic weight of the $k$-th data dimension, represents the penalty function, represents the value of the $k$-th dimension of the $i$-th data point, represents the value of the $k$-th dimension of the $j$-th data point, represents the number of data points within the neighborhood range of the $i$-th data point, represents the number of data points within the neighborhood range of the $j$-th data point, represents taking the absolute value, represents the logarithmic function, represents taking the maximum dimension value;

[0016] Generate dynamic weights to adjust the contribution of each data point to the cluster center according to its local density and global distribution, as shown below:

[0017] ;

[0018] where, represents the dynamic weight of the $i$-th data point, represents the exponential function with the natural constant as the base, represents the Euclidean distance from the $i$-th data point to the nearest cluster center, represents the weight distribution parameter, $N$ represents the total number of data points;

[0019] Cluster center update, introducing an adaptive update mechanism to make the update of the cluster center consider the position, weight, and local density of the data points, as shown below:

[0020] ;

[0021] where, $v$ represents the cluster index, represents the position after the update of the $v$-th cluster center, represents the indicator factor, which takes the value of 1 when belongs to the $v$-th cluster, otherwise 0;

[0022] Introduce clustering constraints, introducing clustering constraints based on the topological structure to make the clustering process consider the topological relationship of the data points in space, as shown below:

[0023] ;

[0024] where, represents the topological constraint term, and V represents the total number of cluster centers. represents the position of the v-th cluster center. represents taking the modulus length. represents the topological scaling factor.

[0025] The design objective function is expressed as follows:

[0026] ;

[0027] Among them, represents the objective function value. represents the objective function weight.

[0028] For clustering, set the objective function threshold, randomly select V initial cluster centers, calculate the distance between each data point and each cluster center using the distance metric function, assign the data points to the cluster with the closest distance, and after all data points are assigned, update the cluster centers. Repeat this process until the objective function value is less than the objective function threshold, and the clustering is completed.

[0029] Further, the cluster structure evolution module specifically includes the following contents:

[0030] Determine the structure parameters, set the total number of cluster centers, the metric weight, and the weight distribution parameters in the clustering process as the cluster structure parameters, and set the reciprocal of the objective function value of the clustering as the fitness of the parameters.

[0031] Design a parameter generation strategy, create a parameter space, introduce random numbers and non-linear transformations, and randomly generate initial parameter points in the parameter space, which is expressed as follows:

[0032] ;

[0033] Among them, represents the generated parameter point. represents the lower limit of the parameter space. represents a random number with a value range between 0 and 1. represents the upper limit of the parameter space.

[0034] Calculate the information intensity to quantify the importance of the parameter points. The initial information intensity is initialized by random perturbation and information constants, and the information intensity is adjusted by the normalization of the fitness to reflect the importance of the parameter points in the evolution process. At the same time, the reduction of information is controlled by the information adjustment coefficient, which is expressed as follows:

[0035] ;

[0036] Among them, l represents the index of the parameter point. represents the information intensity of the l-th parameter point at the first iteration. represents an information constant, r2 represents a number with a value range between 0 and 0.1, and t represents the number of parameter iteration times. represents the information intensity of the l-th parameter point at the t-th iteration. represents an information adjustment coefficient. represents the position of the l-th parameter point at the t-th iteration. represents the parameter fitness of the l-th parameter point at the

[0037] Generate driving parameters, introduce the normalization of the attenuation coefficient and information intensity, generate driving parameters that can gradually weaken with the increase of the iteration times, and at the same time adjust the driving intensity through the reciprocal of the position difference, which is expressed as follows:

[0038] ;

[0039] where represents the attenuation coefficient. represents the driving parameter of the l-th parameter point at the t-th iteration. represents the average value of the information intensities of all parameter points at the t-th iteration. represents the average value of the positions of all parameter points at the

[0040] Define the parameter movement speed, which combines the inertia weight and random perturbation to define the parameter movement speed. The inertia weight gradually decreases with the increase of the iteration times, so that the parameter points can move quickly in the initial stage and then gradually stabilize.

[0041] Parameter movement, update the position of the parameter point according to the calculated parameter movement speed.

[0042] Structural parameter iteration, first set the maximum number of parameter iteration times, generate the initial parameter points by using the parameter generation strategy, calculate the parameter fitness of the initial parameter points and calculate the average value of the fitness, set 1.5 times of the average value of the fitness as the fitness target value, perform parameter movement, and calculate the parameter fitness after each parameter movement; when the parameter fitness is greater than the fitness target value, the evolution ends; when the iteration times are greater than the maximum iteration times, regenerate the initial parameter points and evolve again; otherwise, add 1 to the iteration times and continue the iteration.

[0043] Furthermore, the auxiliary medication module collects the patient's attribute data, clinical physiological indicators, and laboratory test data, inputs the data into the auxiliary medication clustering, finds the cluster to which the input patient data belongs, provides the medication data in the cluster for the user, and outputs the predicted drugs and treatment responses to assist the patient in taking medications.

[0044] The beneficial effects achieved by the present invention using the above solution are as follows:

[0045] (1) Aiming at the problems of the traditional clustering method in processing sepsis patient data, such as being insensitive to feature differences and difficult to adapt to dynamic changes, this solution improves the flexibility and accuracy of clustering by introducing a dynamic weight adjustment mechanism; designs an improved distance metric function to enhance the sensitivity to large differences through a penalty function, further improving the robustness of clustering; introduces clustering constraints based on topological structure, enabling the clustering process to fully consider the spatial topological relationship of the data, significantly improving the accuracy and adaptability of clustering, and providing more scientific and reliable auxiliary support for the medication decision-making of sepsis patients.

[0046] (2) Aiming at the problems of the traditional clustering method in processing complex data structures, such as low optimization efficiency, difficulty in dynamically adjusting parameters, and being prone to falling into local optima, this solution realizes the dynamic generation and optimization of parameters by introducing a clustering structure evolution module; designs a parameter generation strategy and a parameter movement mechanism, which can adaptively generate initial parameter points and dynamically adjust parameter values, thus significantly improving the optimization efficiency and adaptability; by introducing information intensity and driving parameters, combined with the mechanism of information adjustment coefficient and attenuation coefficient, dynamically adjusts the search direction and speed of parameter points, effectively avoiding local optima and enhancing the global search ability; through the structure parameter iteration mechanism, it can dynamically optimize the clustering structure according to data characteristics, further improving the accuracy and robustness of clustering, and providing a more efficient and reliable solution for clustering analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of an intelligent auxiliary medication system for sepsis patients provided by the present invention;

[0048] Figure 2 is a schematic diagram of the auxiliary medication clustering generation module;

[0049] Figure 3 is a schematic diagram of the clustering structure evolution module.

[0050] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0051] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0052] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0053] Embodiment 1, referring to Figure 1 , an intelligent auxiliary medication system for sepsis patients provided by the present invention includes a data acquisition module, an auxiliary medication clustering generation module, a clustering structure evolution module, and an auxiliary medication module;

[0054] The data acquisition module collects the attribute data, clinical physiological indicators, laboratory test data, and medication data of historical sepsis patients, and sends the data to the auxiliary medication clustering generation module;

[0055] The auxiliary medication clustering generation module receives the data sent by the data acquisition module, generates auxiliary medication clusters by neighborhood division, distance measurement, generating dynamic weights, updating cluster centers, introducing clustering constraints, and designing objective functions, and sends the data to the clustering structure evolution module;

[0056] The clustering structure evolution module receives the data sent by the auxiliary medication clustering generation module, realizes the evolution of the clustering structure by determining structure parameters, designing parameter generation strategies, calculating information intensity, generating driving parameters, defining parameter movement speeds, parameter movement, and structure parameter iteration, and sends the data to the auxiliary medication module;

[0057] The auxiliary medication module receives the data sent by the clustering structure evolution module, collects the attribute data, clinical physiological indicators, and laboratory test data of the patient, and uses the auxiliary medication clusters to output predicted drugs and treatment responses to assist the patient in taking medications.

[0058] Embodiment 2, referring to Figure 1, This embodiment is based on the above embodiment. The data acquisition module acquires the attribute data, clinical physiological index data, laboratory examination data, medication data, and patient status data of historical sepsis patients. The attribute data includes the patient's age, gender, and medical history. The clinical physiological index data includes body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation. The laboratory examination data includes the examination data of white blood cell count, platelet count, lactate level, inflammatory markers, liver function, and kidney function. The medication data includes the administered drugs and treatment responses.

[0059] Embodiment Three. Refer to Figure 1 and Figure 2 , This embodiment is based on the above embodiment. The auxiliary medication clustering generation module specifically includes the following:

[0060] Neighborhood division: Calculate the average position of all data points, select the 30% of the data points closest to the average position of the data points, calculate the Euclidean distance between the average position of the data points and the data point farthest from these 30% of the data points, and set this distance as the neighborhood width. Divide the area within the neighborhood width into the neighborhood range.

[0061] Distance metric: Consider the distance in space, add a penalty for large differences through a logarithmic term, and set dynamic feature weighting, which is expressed as follows:

[0062] ;

[0063] where i and j represent the indices of the data points, and represent the positions of the i-th and j-th data points respectively, represents the distance metric function, represents the distance metric between the i-th data point and the j-th data point, represents the Euclidean distance between the i-th data point and the j-th data point, , and represent the metric weights, k represents the dimension index of the data point, d represents the total number of dimensions of the data point, represents the logarithmic weight of the k-th data dimension, represents the penalty function, represents the value of the k-th dimension of the i-th data point, represents the value of the k-th dimension of the j-th data point, represents the number of data points within the neighborhood range of the i-th data point, represents the number of data points within the neighborhood range of the j-th data point, represents taking the absolute value, represents the logarithmic function. Indicates taking the maximum dimension value;

[0064] Generate dynamic weights so that the contribution of each data point to the cluster center is adjusted according to its local density and global distribution, as shown below:

[0065] ;

[0066] Where, Indicates the dynamic weight of the i-th data point, Indicates the exponential function with the natural constant as the base, Indicates the Euclidean distance from the i-th data point to the nearest cluster center, Indicates the weight distribution parameter, and N represents the total number of data points;

[0067] Cluster center update, introducing an adaptive update mechanism so that the update of the cluster center takes into account the position, weight, and local density of the data points, as shown below:

[0068] ;

[0069] Where, v represents the index of the cluster, Indicates the position after the update of the v-th cluster center, Indicates the indicator factor, when Belongs to the v-th cluster, the value is 1, otherwise it is 0;

[0070] Introduce cluster constraints, introduce cluster constraints based on the topological structure so that the clustering process takes into account the topological relationship of data points in space, as shown below:

[0071] ;

[0072] Where, Indicates the topological constraint term, V represents the total number of cluster centers, Indicates the position of the v-th cluster center, Indicates taking the modulus length, Indicates the topological stretching factor;

[0073] Design the objective function, as shown below:

[0074] ;

[0075] Where, Indicates the value of the objective function, Indicates the objective function weight;

[0076] Clustering, set the threshold of the objective function, randomly select V initial clustering centers, calculate the distance between each data point and each clustering center using the distance metric function, assign the data points to the cluster with the closest distance, and after all data points are assigned, update the clustering centers. Repeat this process until the objective function value is less than the objective function threshold, and the clustering is completed.

[0077] By performing the above operations, for the problems of the traditional clustering method being insensitive to feature differences and difficult to adapt to dynamic changes when dealing with sepsis patient data, this solution improves the flexibility and accuracy of clustering by introducing a dynamic weight adjustment mechanism; designs an improved distance metric function to enhance the sensitivity to large differences through a penalty function, further improving the robustness of clustering; introduces clustering constraints based on topological structure, enabling the clustering process to fully consider the spatial topological relationship of the data, significantly improving the accuracy and adaptability of clustering, and providing more scientific and reliable auxiliary support for the drug use decision-making of sepsis patients.

[0078] Example 4, refer to Figure 1 and Figure 3 , based on the above example, the clustering structure evolution module specifically includes the following content:

[0079] Determine the structure parameters, set the total number of clustering centers, metric weights, and weight distribution parameters of the clustering process as clustering structure parameters, and set the reciprocal of the objective function value of the clustering as the fitness of the parameters;

[0080] Design a parameter generation strategy, create a parameter space, introduce random numbers and non-linear transformations, and randomly generate initial parameter points within the parameter space, expressed as follows:

[0081] ;

[0082] Among them, represents the generated parameter point, represents the lower limit of the parameter space, represents a random number with a value range between 0 and 1, represents the upper limit of the parameter space;

[0083] Calculate the information intensity, quantify the importance of the parameter points, initialize the initial information intensity through random perturbation and information constants, adjust the information intensity through the normalization of the fitness, reflect the importance of the parameter points in the evolution process, and at the same time control the reduction of information through the information adjustment coefficient, expressed as follows:

[0084] ;

[0085] Among them, l represents the index of the parameter point, denotes the information intensity of the $l$-th parameter point at the first iteration, denotes the information constant, $r2$ denotes a number with a value range between 0 and 0.1, and $t$ denotes the parameter iteration number, denotes the information intensity of the $l$-th parameter point at the $t$-th iteration, denotes the information adjustment coefficient, denotes the position of the $l$-th parameter point at the $t$-th iteration, denotes the parameter fitness of the $l$-th parameter point at the

[0086] Generate driving parameters, introduce the normalization of the attenuation coefficient and information intensity, generate driving parameters that can gradually weaken with the increase of the iteration number, and at the same time adjust the driving intensity through the reciprocal of the position difference, which is expressed as follows:

[0087] ;

[0088] where, denotes the attenuation coefficient, denotes the driving parameter of the $l$-th parameter point at the $t$-th iteration, denotes the average value of the information intensity of all parameter points at the $t$-th iteration, denotes the average value of the positions of all parameter points at the

[0089] Define the parameter movement speed, which combines the inertia weight and random perturbation to define the parameter movement speed. The inertia weight gradually decreases with the increase of the iteration number, so that the parameter points can move quickly in the initial stage and then gradually stabilize, which is expressed as follows:

[0090] ;

[0091] where, denotes the parameter movement speed of the $l$-th parameter point at the $(t + 1)$-th iteration, and respectively denote the maximum parameter movement weight and the minimum parameter movement weight, $T$ denotes the maximum parameter iteration number, denotes the parameter movement speed of the $l$-th parameter point at the $t$-th iteration, denotes a random number with a value range between 0 and 0.85, denotes the driving increment, $m$ denotes the index of the parameter point, denotes the position of the $m$-th parameter point at the $t$-th iteration;

[0092] Parameter movement: Update the position of the parameter point according to the calculated parameter movement speed, as shown below:

[0093] ;

[0094] Among them, represents the position of the l-th parameter point at the (t + 1)-th iteration;

[0095] Structural parameter iteration: First, set the maximum number of parameter iterations. Generate the initial parameter points using the parameter generation strategy, calculate the parameter fitness of the initial parameter points and the average fitness value, set 1.5 times the average fitness value as the fitness target value, and perform parameter movement. Calculate the fitness of the parameters after each parameter movement; when the fitness of the parameters is greater than the fitness target value, the evolution ends; when the number of iterations is greater than the maximum number of iterations, regenerate the initial parameter points and evolve again; otherwise, increment the number of iterations by one and continue the iteration.

[0096] By performing the above operations, for the problems of low optimization efficiency, difficulty in dynamically adjusting parameters, and easy to fall into local optimum when the traditional clustering method is dealing with complex data structures, this solution realizes the dynamic generation and optimization of parameters by introducing a clustering structure evolution module; designs a parameter generation strategy and a parameter movement mechanism, which can adaptively generate initial parameter points and dynamically adjust the parameter values, thus significantly improving the optimization efficiency and adaptability; by introducing information intensity and driving parameters, combined with the mechanism of information regulation coefficient and attenuation coefficient, dynamically adjust the search direction and speed of the parameter points, effectively avoid local optimum, and enhance the global search ability; through the structural parameter iteration mechanism, it can dynamically optimize the clustering structure according to data characteristics, further improving the accuracy and robustness of clustering, and providing a more efficient and reliable solution for clustering analysis.

[0097] Example 5, refer to Figure 1 Based on the above example, the auxiliary medication module collects the patient's attribute data, clinical physiological indicators, and laboratory test data, inputs the data into the auxiliary medication clustering, finds the cluster to which the input patient data belongs, provides the medication data in the cluster for the user, and outputs the predicted drugs and treatment responses to assist the patient in taking medication.

[0098] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device.

[0099] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

[0100] The above description of the present invention and its embodiments is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In general, if those of ordinary skill in the art are inspired by it and design similar structural modes and embodiments without creative efforts without departing from the purpose of the present invention, they shall fall within the protection scope of the present invention.

Claims

1. An intelligent medication-assisted system for sepsis patients, characterized by: It includes data collection module, auxiliary medication cluster generation module, cluster structure evolution module and auxiliary medication module; The data collection module collects attribute data, clinical physiological indicators, laboratory test data and medication data of historical sepsis patients; The auxiliary medication cluster generation module generates auxiliary medication clusters by neighborhood division, distance measurement, generation of dynamic weights, cluster center update, introduction of cluster constraints and design of objective functions; The distance metric takes into account the spatial distance and increases the penalty for large differences through logarithmic terms. At the same time, dynamic feature weighting is set, which is expressed as follows: ； Among them, i and j represent the index of the data point, and Respectively represent the positions of the i-th and j-th data points, represents the distance metric function, represents the distance measure between the i-th data point and the j-th data point, represents the Euclidean distance between the i-th data point and the j-th data point, , and represents the measurement weight, k represents the dimension index of the data point, and d represents the total number of dimensions of the data point. represents the logarithmic weight of the kth data dimension, represents the penalty function, represents the value of the kth dimension of the ith data point, represents the value of the kth dimension of the jth data point, represents the number of data points in the neighborhood of the i-th data point, represents the number of data points in the neighborhood of the jth data point, Indicates taking the absolute value, represents the logarithmic function, Indicates taking the maximum dimension value; The cluster structure evolution module realizes cluster structure evolution by determining structure parameters, designing parameter generation strategy, calculating information intensity, generating driving parameters, defining parameter movement speed, parameter movement and structure parameter iteration; The auxiliary medication module collects the patient's attribute data, clinical physiological indicators and laboratory test data, and uses auxiliary medication clustering to output predicted drugs and treatment responses to assist patients in taking medication.

2. According to claim 1, an intelligent medication auxiliary system for sepsis patients is characterized by: The auxiliary medication cluster generation module specifically includes the following contents: Neighborhood partitioning: calculate the average position of all data points, select the 30% of data points closest to the average position of the data points, calculate the Euclidean distance between the average position of the data points and the data point farthest from these 30% of data points, set this distance as the neighborhood width, and divide the area within the neighborhood width into the neighborhood range; Distance metrics; Generate dynamic weights so that the contribution of each data point to the cluster center is adjusted according to its local density and global distribution, as shown below: ； in, represents the dynamic weight of the i-th data point, represents an exponential function with a natural constant as base, represents the Euclidean distance from the ith data point to the nearest cluster center, represents the weight distribution parameter, and N represents the total number of data points; Cluster center update, introduces an adaptive update mechanism so that the update of cluster centers takes into account the location, weight and local density of data points, which is expressed as follows: ； Among them, v represents the index of the cluster, represents the updated position of the vth cluster center, Represents the indicator factor, when The value is 1 when it belongs to the vth cluster, otherwise it is 0; Clustering constraints are introduced. Clustering constraints based on topological structure are introduced so that the clustering process takes into account the topological relationship of data points in space, which is expressed as follows: ； in, represents the topological constraint term, V represents the total number of cluster centers, represents the location of the vth cluster center, Indicates the modulus length, represents the topological scaling factor; The objective function is designed as follows: ； in, represents the objective function value, represents the objective function weight; Clustering, set the objective function threshold, randomly select V initial cluster centers, use the distance metric function to calculate the distance between each data point and each cluster center, assign the data point to the cluster closest to it, and after all data points are assigned, update the cluster center and repeat this process until the objective function value is less than the objective function threshold and clustering is completed.

3. The intelligent medication auxiliary system for sepsis patients according to claim 1, characterized in that: The cluster structure evolution module specifically includes the following contents: Determine the structural parameters, set the total number of cluster centers, measurement weights and weight distribution parameters of the clustering process as clustering structural parameters, and set the inverse of the clustering objective function value as the fitness of the parameters; Design a parameter generation strategy, create a parameter space, introduce random numbers and nonlinear transformations, and randomly generate initial parameter points in the parameter space, as shown below: ； in, represents the generated parameter points, represents the lower limit of the parameter space, Represents a random number between 0 and 1. Indicates the upper limit of the parameter space; Calculate the information intensity and quantify the importance of the parameter points. The initial information intensity is initialized by random perturbation and information constant. The information intensity is adjusted by normalization of fitness to reflect the importance of the parameter points in the evolution process. At the same time, the reduction of information is controlled by the information adjustment coefficient, which is expressed as follows: ； Where l represents the index of the parameter point, represents the information strength of the lth parameter point at the first iteration, represents the information constant, r2 represents a number ranging from 0 to 0.1, t represents the number of parameter iterations, Indicates The information strength of the lth parameter point at the iteration, represents the information strength of the lth parameter point at the tth iteration, represents the information adjustment coefficient, Indicates The position of the lth parameter point at the iteration, Indicates The parameter fitness of the lth parameter point at the iteration; Generate driving parameters, introduce the normalization of attenuation coefficient and information intensity, generate driving parameters that can gradually weaken with the increase of iteration number, and adjust the driving intensity by the inverse of position difference, which is expressed as follows: ； in, represents the attenuation coefficient, represents the driving parameter of the lth parameter point at the tth iteration, represents the average information intensity of all parameter points at the tth iteration, Indicates The average value of the positions of all parameter points at iterations; Define the parameter movement speed, which combines inertia weight and random perturbation to define the parameter movement speed. The inertia weight gradually decreases with the increase of iteration times, so that the parameter point can move quickly in the initial stage and gradually stabilize in the later stage; Parameter movement: update the position of the parameter point according to the calculated parameter movement speed; For structural parameter iteration, first set the maximum number of parameter iterations, use the parameter generation strategy to generate the initial parameter points, calculate the parameter fitness of the initial parameter points and the average fitness value, set 1.5 times the average fitness value as the fitness target value, move the parameters, and calculate the parameter fitness after each parameter movement; when the parameter fitness is greater than the fitness target value, the evolution ends; when the number of iterations is greater than the maximum number of iterations, regenerate the initial parameter points and evolve again; otherwise, increase the number of iterations by one and continue iterating.

4. The intelligent medication auxiliary system for sepsis patients according to claim 1, characterized in that: The data acquisition module collects attribute data, clinical physiological index data, laboratory test data, medication data and patient status data of historical sepsis patients; the attribute data include the patient's age, gender and medical history; the clinical physiological index data include body temperature, heart rate, respiratory rate, blood pressure and blood oxygen saturation; the laboratory test data include white blood cell count, platelet count, lactate level, inflammatory markers, liver function and kidney function test data; the medication data includes used drugs and treatment response.

5. The intelligent medication auxiliary system for sepsis patients according to claim 1, characterized in that: The auxiliary medication module collects the patient's attribute data, clinical physiological indicators and laboratory test data, inputs the data into the auxiliary medication cluster, finds the cluster to which the input patient data belongs, provides the user with the medication data in the cluster, outputs the predicted drugs and treatment responses, and assists the patient in medication.

Images