Medical material distribution method and device

Abstract

The invention provides a medical material distribution method and device. The method comprises the following steps: counting historical state information of each department from bank scene financial data; based on a multi-output prediction strategy and the historical state information of each department, the number of predicted patients and a predicted severity value are obtained; for each department, setting a weight factor based on the predicted number of patients, and setting an importance factor based on the predicted severity value; constructing a scheduling model composed of an objective function and constraint conditions based on each weight factor, each importance factor and the existing material quantity; and solving the scheduling model to obtain a distribution result. In the scheme, the number of patients is predicted according to bank scene financial data, material weight factor estimation is carried out for different departments based on the predicted number of patients, priority division is carried out on the departments according to the urgent need degree, then refined modeling is carried out for a multi-material and cross-department scheduling model, the distribution reasonability is ensured, and the distribution efficiency is improved. The purposes of saving manpower and material resources and improving the doctor-seeing experience of the patient are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and device for distributing medical supplies. Background Art

[0002] During special periods, hospitals are facing difficulties in allocating some of the scarce shared medical supplies. Due to the differences in the number of patients and the severity of their illnesses in different departments, as well as the existence of special patients who need emergency or priority use, some medical supplies need to be used in conjunction with each other, and it is easy for each department to have excess or shortage of supplies, which requires dispatch.

[0003] If manual scheduling is frequently performed, a large amount of manpower and material resources will be consumed, and medical supplies will not be used in a timely manner, which may easily lead to poor medical experience for patients due to shortage of supplies. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a method and device for distributing medical supplies to achieve the purpose of saving manpower and material resources and improving patients' medical experience.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A first aspect of an embodiment of the present invention discloses a method for distributing medical supplies, the method comprising:

[0007] Count the historical status information corresponding to each department of the hospital from the bank scenario financial data; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' conditions;

[0008] Based on the multi-output prediction strategy and the historical status information corresponding to each department, the predicted number of patients and the predicted severity value corresponding to each department are predicted;

[0009] For each department, a weight factor is set based on the corresponding predicted number of patients, and an importance factor is set based on the corresponding predicted severity value;

[0010] Obtaining the existing quantities of various medical supplies, and constructing a scheduling model based on each of the weight factors, each of the importance factors and the existing quantities of various medical supplies; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the quantity of various medical supplies allocated to each department positively correlated with each of the weight factors, and making the similarity between the proportion of various medical supplies allocated to each department and each of the weight factors exceed a similarity threshold; the constraints are used to constrain the total amount of any medical supply allocated to each department not to exceed the corresponding existing quantity, and to constrain the quantity of various medical supplies allocated to any department to be not less than 0;

[0011] The scheduling model is solved using the simplex method to obtain the allocation result.

[0012] Preferably, the predicted number of patients and predicted severity values ​​corresponding to each department are predicted based on the multi-output prediction strategy and the historical status information corresponding to each department, including:

[0013] The historical status information of each department is input into a pre-trained neural network model to predict the future status information of each department; the future status information includes: predicted number of patients and predicted severity value; the neural network model is composed of an input layer, a hidden layer and an output layer, the hidden layer is composed of an LSTM layer and a Dense layer, and the Dense layer uses a RELU function as an activation function.

[0014] Preferably, the training process of the neural network model includes:

[0015] Constructing a neural network model, and compiling based on the neural network model, the mean square error function and the Adam optimizer to obtain a neural network model to be trained;

[0016] The neural network model to be trained is trained using the pre-acquired training set until a trained neural network model is obtained.

[0017] Preferably, the construction of a scheduling model based on each of the weight factors, each of the importance factors and the existing quantities of various medical supplies includes:

[0018] Based on each of the weight factors, each of the importance factors and the existing material quantity of various medical materials, a linear programming task including two objective items is constructed;

[0019] Based on the weighted combination method and the linear programming task, an objective function and constraint conditions are constructed;

[0020] Based on the objective function and the constraints, a scheduling model is constructed.

[0021] Preferably, the method of solving the scheduling model using the simplex method to obtain the allocation result includes:

[0022] Selecting a plurality of initial feasible bases from the feasible domain of the objective function;

[0023] Construct a simplex table based on each initial feasible basis;

[0024] Check whether the current simplex table is the optimal solution;

[0025] If yes, the current simplex table is used as the allocation result;

[0026] If not, the current simplex table is converted into adjacent feasible basis solutions, and the process returns to the step of checking whether the current simplex table is the optimal solution, until the current simplex table is the optimal solution.

[0027] A second aspect of an embodiment of the present invention discloses a device for distributing medical supplies, the device comprising:

[0028] A statistical unit, used to calculate historical status information corresponding to each department of the hospital from the financial data of the bank scenario; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' illness;

[0029] A prediction unit, configured to predict the number of patients and the predicted severity value corresponding to each department based on a multi-output prediction strategy and the historical status information corresponding to each department;

[0030] A setting unit, configured to set a weight factor based on the corresponding predicted number of patients and an importance factor based on the corresponding predicted severity value for each department;

[0031] A modeling unit, used for obtaining the existing quantity of various medical supplies, and constructing a scheduling model based on each of the weight factors, each of the importance factors and the existing quantity of various medical supplies; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the quantity of various medical supplies allocated to each department positively correlated with each of the weight factors, and making the similarity between the proportion of various medical supplies allocated to each department and each of the weight factors exceed a similarity threshold; the constraints are used to constrain the total amount of any medical supply allocated to each department not to exceed the corresponding existing quantity, and constrain the quantity of various medical supplies allocated to any department to be not less than 0;

[0032] The solving unit is used to solve the scheduling model using the simplex method to obtain the allocation result.

[0033] Preferably, the prediction unit is specifically used for:

[0034] The historical status information of each department is input into a pre-trained neural network model to predict the future status information of each department; the future status information includes: predicted number of patients and predicted severity value; the neural network model is composed of an input layer, a hidden layer and an output layer, the hidden layer is composed of an LSTM layer and a Dense layer, and the Dense layer uses a RELU function as an activation function.

[0035] Preferably, the device further comprises:

[0036] The training unit is used to construct a neural network model, and compile it based on the neural network model, the mean square error function and the Adam optimizer to obtain the neural network model to be trained; the neural network model to be trained is trained using a pre-acquired training set until a trained neural network model is obtained.

[0037] Preferably, the modeling unit is specifically used for:

[0038] Based on each of the weight factors, each of the importance factors and the existing material quantity of various medical materials, a linear programming task including two objective items is constructed;

[0039] Based on the weighted combination method and the linear programming task, an objective function and constraint conditions are constructed;

[0040] Based on the objective function and the constraints, a scheduling model is constructed.

[0041] Preferably, the solving unit is specifically used for:

[0042] Selecting a plurality of initial feasible bases from the feasible domain of the objective function;

[0043] Construct a simplex table based on each initial feasible basis;

[0044] Check whether the current simplex table is the optimal solution;

[0045] If yes, the current simplex table is used as the allocation result;

[0046] If not, the current simplex table is converted into adjacent feasible basis solutions, and the process returns to the step of checking whether the current simplex table is the optimal solution, until the current simplex table is the optimal solution.

[0047] Based on a method and device for distributing medical supplies provided by the above-mentioned embodiment of the present invention, historical status information corresponding to each department of the hospital is statistically obtained from the financial data of the bank scenario; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' conditions; based on the multi-output prediction strategy and the historical status information corresponding to each department, the predicted number of patients and the predicted severity value corresponding to each department are predicted; for each department, a weight factor is set based on the corresponding predicted number of patients, and an importance factor is set based on the corresponding predicted severity value; the existing quantity of various medical supplies is obtained, and based on each of the weight factors The scheduling model is constructed based on the importance factors and the existing material quantity of various medical materials; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the material quantity of various medical materials allocated to each department positively correlated with each weight factor, and making the similarity between the proportion of various medical materials allocated to each department and each weight factor exceed the similarity threshold; the constraints are used to constrain the total amount of any medical material allocated to each department not to exceed the corresponding existing material quantity, and constrain the material quantity of various medical materials allocated to any department to be not less than 0; the scheduling model is solved by the simplex method to obtain the allocation result. In this solution, the number of patients is predicted according to the financial data of the bank scenario, and the material weight factor is estimated based on the predicted number of patients for different departments, and the departments are prioritized according to the degree of urgency, and then the scheduling model for multiple materials and cross-department is refined to ensure the rationality of the allocation, so as to achieve the purpose of saving manpower and material resources and improving the patient's medical experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0049] Figure 1 A flowchart of a method for distributing medical supplies disclosed in an embodiment of the present invention;

[0050] Figure 2 An internal structure diagram of an LSTM disclosed in an embodiment of the present invention;

[0051] Figure 3 A structural diagram of a three-layer neural network model disclosed in an embodiment of the present invention;

[0052] Figure 4 A flowchart of a simplex method for solving a scheduling model disclosed in an embodiment of the present invention;

[0053] Figure 5 This is a structural diagram of a medical material distribution device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0056] As can be seen from the background technology, if manual scheduling is frequently performed, a large amount of manpower and material resources will be consumed, and medical supplies will not be used in a timely manner, which may easily lead to a poor medical experience for patients due to shortage of supplies.

[0057] Therefore, an embodiment of the present invention discloses a method and device for distributing medical supplies. In this solution, the number of patients is predicted based on the financial data of the bank scenario, and the material weight factors of different departments are estimated based on the predicted number of patients. The departments are prioritized according to the degree of urgency, and then a refined model is created for the scheduling model of multiple materials and cross-departments to ensure the rationality of the distribution, so as to achieve the purpose of saving manpower and material resources and improving the patient's medical experience.

[0058] like Figure 1 FIG. 1 is a flow chart of a method for distributing medical supplies disclosed in an embodiment of the present invention, comprising the following steps:

[0059] Step S101: Calculate the historical status information corresponding to each department of the hospital from the bank scenario financial data.

[0060] In step S101 , the historical status information includes: the number of historical patients and historical severity values ​​used to characterize the average severity of the historical patients' conditions.

[0061] It should be noted that the financial data of banking scenarios in smart medical scenarios can provide banks with in-depth insights, helping them develop and provide financial products and services that are more in line with the needs of the medical industry.

[0062] In order to achieve the reasonable allocation of medical supplies, the number of patients, patient severity, number of doctors, number of nurses, number of beds and other characteristics of each department can be counted on the basic financial data of the bank scenario for prediction. For example, the number of patients and patient severity of each department can be counted through medical payment data. Medical payment data includes the amount and items of payment, which can reflect the severity of patients to a certain extent.

[0063] Although such a large number of features can provide rich guidance information, it increases the workload of information collection, increases the complexity of the problem, and brings inconvenience to the analysis. Moreover, the accuracy of predicting so many variables will naturally be discounted. However, if each variable is analyzed separately, the correlation between variables is often ignored, and there is no effect of global analysis, which is inaccurate.

[0064] Therefore, for multivariate dimensionality reduction, a feature preprocessing algorithm, namely the principal component analysis algorithm (PCA), is used. PCA is a linear dimensionality reduction algorithm that can project high-dimensional data onto a low-dimensional hyperplane through a certain linear projection. The hyperplane has the following characteristics: nearest reconstruction, the distance between the sample points and the hyperplane is close enough; maximum separability, the projections of the sample points on the hyperplane can be separated as much as possible. This can maximize the amount of data information on the projection plane, and then combine the eigenvalue decomposition of the covariance matrix to perform feature compression, and finally obtain low-dimensional feature variables.

[0065] In the specific implementation process of step S101, in order to predict the changes in the number of patients in each department in the future and the severity values ​​of the patients in each department, the principal component analysis algorithm can be used to count the number of patients Number and the patient severity value Level of each department of the hospital in the past day from the financial data of the banking scenario as the experimental data set, and define the historical status information of the department every day in the past as S (Number, Level).

[0066] Step S102: Based on the multi-output prediction strategy and the historical status information corresponding to each department, the predicted number of patients and the predicted severity value corresponding to each department are predicted.

[0067] In step S102, a multi-output prediction strategy is used to predict the department status S' in the next F days. That is, the historical status information of the previous H days is used to predict the most likely status S of each department in the next F days. The formula is expressed as:

[0068] S t '+1 ,...,S t ' +F = argmaxp(S t+1 ,...,S t+F |S t-H+1 ,...,S t )

[0069] Among them, S t It represents the status of the department at time point t, including the number of patients in the department and the severity value. The left side of the equation represents the predicted status of the department in the next F days.

[0070] In the specific implementation process of step S102, the historical status information of each department is input into the pre-trained neural network model to predict the future status information of each department.

[0071] Among them, the future status information includes: predicted number of patients and predicted severity value.

[0072] The neural network model consists of an input layer, a hidden layer, and an output layer. The hidden layer consists of an LSTM layer and a Dense layer. The Dense layer uses the RELU function as the activation function.

[0073] The reason for using the LSTM algorithm based on a neural network is that the daily status of each department is a time series. There are currently many technical methods for predicting time series, such as the traditional ARIMA algorithm and the LSTM algorithm based on a neural network. However, the traditional ARIMA algorithm is difficult to model for more complex nonlinear relationships, while LSTM is very powerful in processing complex nonlinear dependencies and can be trained and learn dependencies between time series through recursive neural networks. Therefore, the prediction method of the embodiment of the present invention uses LSTM for prediction, which includes the following two parts:

[0074] Multi-step prediction of the number of patients in each department:

[0075] Number_Predict(t+1),...,Number_Predict(t+F)=LSTM(Number_Observe(t-1),...,Number_Observe(tH))

[0076] Multi-step prediction of the average severity of patients in each department:

[0077] Level_Predict(t+1),...,Level_Predict(t+F)=LSTM(Level_Observe(t-1),...,Level_Observe(tH))

[0078] like Figure 2 , which is an internal structure diagram of an LSTM disclosed in an embodiment of the present invention.

[0079] The internal structure of LSTM includes three gate structures: forget gate ft, input gate it, and output gate ot. The activation functions include tanh and sigmoid functions. LSTM can selectively forget unnecessary memories through these gate structures, and add content that needs to be remembered for a long time to the neural network. These are all controlled by the activation function. Therefore, LSTM can achieve the effect of capturing long-term dependencies and building nonlinear models.

[0080] The neural network model of the embodiment of the present invention is composed of an input layer, a hidden layer and an output layer. The hidden layer is composed of an LSTM layer and a Dense layer. The Dense layer uses a RELU function as an activation function. The effect of a multi-layer neural network is always better than that of a single-layer neural network, and the number of nodes in each layer of the neural network needs to be sufficient, so that more complex nonlinear functions can be fit more accurately and a more complex model can be constructed.

[0081] like Figure 3 , which is a structural diagram of a three-layer neural network model disclosed in an embodiment of the present invention.

[0082] The neural network model contains an LSTM layer for transmitting memory information, and the other two layers are fully connected Dense layers for fitting nonlinear functions and outputting prediction information.

[0083] The activation function uses RELU. The input X is the historical status information S (Number, Level) of the department in the past H days, and the output Y is the status information S′ (Number, Level) of the department in the future F days.

[0084] The training process of the neural network model includes:

[0085] Construct a neural network model, and compile it based on the neural network model, mean square error function MSE and Adam optimizer to obtain the neural network model to be trained; use the pre-acquired training set to train the neural network model to be trained until the trained neural network model is determined to be completed using the test set.

[0086] Among them, the first 80% of the data set can be divided into a training set, and the last 20% of the data set can be divided into a test set.

[0087] The dataset includes: historical patient numbers and severity values ​​with label information. These label information corresponds to the input data and is used to guide the model to learn the correct output.

[0088] Step S103: For each department, a weight factor is set based on the corresponding predicted number of patients, and an importance factor is set based on the corresponding predicted severity value.

[0089] It should be noted that hospitals generally have multiple departments with different diagnosis and treatment directions, such as internal medicine, orthopedics, neurosurgery, infectious diseases, etc. Since the probability of infection and duration of various diseases are different, the demand quota for material dispatch in each department is usually determined according to the frequency of consultations in a recent period, but this empirical artificial value has certain inaccuracies and non-generalization.

[0090] Therefore, in the embodiment of the present invention, assuming that the number of departments to be allocated medical supplies in the hospital is N, the predicted number of patients is positively correlated with the weight factor w of the medical supplies in each department, and the weight set W is defined as:

[0091]

[0092] It should also be noted that the severity and urgency of the conditions diagnosed by each department vary significantly. Departments that are closely related to the life safety of patients often include high-risk matters such as surgery, such as organ transplantation, intensive care medicine, and cardiovascular surgery. The medical supply needs of these departments should be met as much as possible. In contrast, general internal medicine and surgery, health care, etc. can slightly postpone meeting the needs. In order to more clearly identify the urgency of the department's supplies, the importance factor corresponding to department j is set to k. σj .

[0093] It is understandable that the importance factor corresponding to department j can be set according to the historical severity value corresponding to department j. Since departments with high historical severity values ​​have correspondingly higher demands for medical supplies, the value set for the importance factor should also be higher.

[0094] Step S104: Obtain the existing material quantities of various medical materials, and construct a scheduling model based on various weight factors, various importance factors and the existing material quantities of various medical materials.

[0095] Among them, the scheduling model is composed of an objective function and constraints; the scheduling goals of the objective function include: making the quantity of various medical supplies allocated to each department positively correlated with each weight factor, and making the similarity between the proportion of various medical supplies allocated to each department and each weight factor exceed the similarity threshold.

[0096] It is understandable that the various types of materials that departments rely on, such as medical equipment, protective supplies, and medical resources, overlap with each other, and also have their own expected types. Moreover, the scheduling goals must include two items, namely, not only should departments with high weights be allocated more materials, but the allocation ratio should also be close to the weight factor w, so as to ensure a certain degree of differentiation and balance.

[0097] The objective function is as follows:

[0098]

[0099] The maximum number of types of materials is M, and the maximum supply of material type i is P i , the quantity of material type i allocated to department j is a ij , the coefficients of the two objective terms are k 1 , k 2 .

[0100] X i Represents the actual allocation ratio of material type i:

[0101] X i ={x i1 ,x i2 ,…,x ij}

[0102] Regarding the judgment of whether the similarity between the proportion of various medical supplies allocated to various departments and various weight factors exceeds the similarity threshold, the similarity is calculated using the 2-norm statistical distance, as follows:

[0103]

[0104] The constraint condition is used to constrain the total amount of any medical supplies allocated to each department to not exceed the corresponding existing supply quantity, and to constrain the quantity of various medical supplies allocated to any department to be not less than 0.

[0105] The constraints are as follows:

[0106]

[0107] In the specific implementation process of step S104, based on various weight factors, various importance factors and the existing material quantities of various medical supplies, a linear programming task containing two target items is constructed; based on the weighted combination method and the linear programming task, the objective function and constraints are constructed; based on the objective function and constraints, a scheduling model is constructed.

[0108] Among them, the multi-objective and multi-constrained linear programming task can be converted into a common single-objective programming task using the weighted combination method, that is, using k 1 and k 2The objective function is obtained by performing a weighted combination of the two objective items.

[0109] Step S105: Solve the scheduling model using the simplex method to obtain the allocation result.

[0110] In step S105, the scheduling model is solved using the simplex algorithm. The algorithm first searches for a set of solutions in the feasible domain, and then transforms the feasible solutions so that the objective function value continues to decrease, thereby searching for the optimal solution to the problem. The optimal solution obtained using the simplex method is the final optimal allocation result for a variety of medical supplies.

[0111] like Figure 4 As shown, it is a flowchart of solving a scheduling model using a simplex method disclosed in an embodiment of the present invention.

[0112] Specifically, multiple initial feasible bases are selected from the feasible domain of the objective function; a simplex table is constructed based on each initial feasible base; it is checked whether the current simplex table is the optimal solution; if so, the current simplex table is used as the allocation result; if not, the current simplex table is converted to an adjacent feasible base solution, and the step of checking whether the current simplex table is the optimal solution is returned to be executed until the current simplex table is the optimal solution.

[0113] The explanation of each step in the solution process is as follows:

[0114] Start: Starts the simplex solution process.

[0115] Construct an initial feasible basis: Select an initial solution that satisfies all constraints as the starting point.

[0116] List the initial simplex table: Construct the initial simplex table based on the initial feasible basis.

[0117] Check whether it is the optimal solution: Check the coefficients of non-basic variables in the objective function row in the current simplex table. If the coefficients of all non-basic variables are non-negative, then the current solution is the optimal solution.

[0118] Conversion of adjacent feasible basis solutions: If the current solution is not the optimal solution, select a non-basis variable as the input basis variable, and then update the simplex table through row transformation to change the basis variable and find a better solution.

[0119] Repeat the test: After updating the simplex table, check again whether it is the optimal solution.

[0120] Get the allocation result: Once the optimal solution is found, read the values ​​of the basis variables from the simplex table, which are the optimal allocation results.

[0121] End: Complete the simplex method solution process.

[0122] In the simplex method, each step of the transformation is based on the optimality test result of the current solution. If the current solution is not the optimal solution, the algorithm will find a better solution by transforming to the adjacent feasible basis. This process will be repeated until the optimal solution is found.

[0123] Based on the above-mentioned method for distributing medical supplies disclosed in the embodiment of the present invention, in this solution, the number of patients is predicted according to the financial data of the bank scenario, and the material weight factor is estimated for different departments based on the predicted number of patients, and the departments are prioritized according to the degree of urgency, and then a refined model is built for the scheduling model of multiple materials and cross-departments to ensure the rationality of the distribution, so as to achieve the purpose of saving manpower and material resources and improving the patient's medical experience.

[0124] Corresponding to the method for distributing medical supplies disclosed in the above embodiment of the present invention, as Figure 5 , which is a structural diagram of a medical material distribution device disclosed in an embodiment of the present invention, including: a statistical unit 501, a prediction unit 502, a setting unit 503, a modeling unit 504 and a solution unit 505.

[0125] The statistical unit 501 is used to collect historical status information corresponding to each department of the hospital from the financial data of the banking scenario; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' illnesses.

[0126] The prediction unit 502 is used to predict the number of patients and the predicted severity value corresponding to each department based on the multi-output prediction strategy and the historical status information corresponding to each department.

[0127] The setting unit 503 is used to set a weight factor based on the corresponding predicted number of patients and an importance factor based on the corresponding predicted severity value for each department.

[0128] The modeling unit 504 is used to obtain the existing quantity of various medical supplies, and to construct a scheduling model based on various weight factors, various importance factors and the existing quantity of various medical supplies; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the quantity of various medical supplies allocated to each department positively correlated with each weight factor, and making the similarity between the proportion of various medical supplies allocated to each department and each weight factor exceed the similarity threshold; the constraints are used to constrain the total amount of any medical supply allocated to each department not to exceed the corresponding existing quantity, and to constrain the quantity of various medical supplies allocated to any department to be not less than 0.

[0129] The solving unit 505 is used to solve the scheduling model using the simplex method to obtain the allocation result.

[0130] In one embodiment, the prediction unit 502 is specifically configured to:

[0131] The historical status information of each department is input into the pre-trained neural network model to predict the future status information of each department; the future status information includes: predicted number of patients and predicted severity value; the neural network model consists of an input layer, a hidden layer and an output layer, the hidden layer consists of an LSTM layer and a Dense layer, and the Dense layer uses the RELU function as the activation function.

[0132] In one embodiment, the device further comprises:

[0133] The training unit is used to construct a neural network model, and compile it based on the neural network model, the mean square error function and the Adam optimizer to obtain the neural network model to be trained; the neural network model to be trained is trained using the pre-acquired training set until a trained neural network model is obtained.

[0134] In one embodiment, the modeling unit 504 is specifically configured to:

[0135] Based on various weight factors, various importance factors and the existing quantities of various medical supplies, a linear programming task containing two objective items is constructed; based on the weighted combination method and the linear programming task, the objective function and constraints are constructed; based on the objective function and constraints, a scheduling model is constructed.

[0136] In one embodiment, the solving unit 505 is specifically configured to:

[0137] Select multiple initial feasible bases from the feasible domain of the objective function; construct a simplex table based on each initial feasible base; check whether the current simplex table is the optimal solution; if so, use the current simplex table as the allocation result; if not, perform adjacent feasible base solution conversion on the current simplex table, and return to execute the step of checking whether the current simplex table is the optimal solution, until the current simplex table is the optimal solution.

[0138] Based on the above-mentioned embodiment of the present invention, a distribution device for medical supplies is disclosed. In this solution, the number of patients is predicted according to the financial data of the bank scenario, and the material weight factor is estimated for different departments based on the predicted number of patients, and the departments are prioritized according to the degree of urgency, and then a refined model is built for the scheduling model of multiple materials and cross-departments to ensure the rationality of the distribution, so as to achieve the purpose of saving manpower and material resources and improving the patient's medical experience.

[0139] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0140] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0141] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for distributing medical supplies, characterized in that: The method comprises: Count the historical status information corresponding to each department of the hospital from the bank scenario financial data; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' conditions; Based on the multi-output prediction strategy and the historical status information corresponding to each department, the predicted number of patients and the predicted severity value corresponding to each department are predicted; For each department, a weight factor is set based on the corresponding predicted number of patients, and an importance factor is set based on the corresponding predicted severity value; Obtaining the existing quantities of various medical supplies, and constructing a scheduling model based on each of the weight factors, each of the importance factors and the existing quantities of various medical supplies; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the quantity of various medical supplies allocated to each department positively correlated with each of the weight factors, and making the similarity between the proportion of various medical supplies allocated to each department and each of the weight factors exceed a similarity threshold; the constraints are used to constrain the total amount of any medical supply allocated to each department not to exceed the corresponding existing quantity, and to constrain the quantity of various medical supplies allocated to any department to be not less than 0; The scheduling model is solved using the simplex method to obtain the allocation result.

2. The method according to claim 1, characterized in that The method of predicting the number of patients and the predicted severity value corresponding to each department based on the multi-output prediction strategy and the historical status information corresponding to each department includes: The historical status information of each department is input into a pre-trained neural network model to predict the future status information of each department; the future status information includes: predicted number of patients and predicted severity value; the neural network model is composed of an input layer, a hidden layer and an output layer, the hidden layer is composed of an LSTM layer and a Dense layer, and the Dense layer uses a RELU function as an activation function.

3. The method according to claim 2, characterized in that The training process of the neural network model includes: Constructing a neural network model, and compiling based on the neural network model, the mean square error function and the Adam optimizer to obtain a neural network model to be trained; The neural network model to be trained is trained using the pre-acquired training set until a trained neural network model is obtained.

4. The method according to claim 1, characterized in that The scheduling model is constructed based on each of the weight factors, each of the importance factors and the existing material quantity of various medical materials, including: Based on each of the weight factors, each of the importance factors and the existing material quantity of various medical materials, a linear programming task including two objective items is constructed; Based on the weighted combination method and the linear programming task, an objective function and constraint conditions are constructed; Based on the objective function and the constraints, a scheduling model is constructed.

5. The method according to claim 1, characterized in that The method of solving the scheduling model by the simplex method to obtain the allocation result includes: Selecting a plurality of initial feasible bases from the feasible domain of the objective function; Construct a simplex table based on each initial feasible basis; Check whether the current simplex table is the optimal solution; If yes, the current simplex table is used as the allocation result; If not, the current simplex table is converted into adjacent feasible basis solutions, and the process returns to the step of checking whether the current simplex table is the optimal solution, until the current simplex table is the optimal solution.

6. A medical material distribution device, characterized in that: The device comprises: A statistical unit, used to calculate historical status information corresponding to each department of the hospital from the financial data of the bank scenario; the historical status information includes: the number of historical patients and the historical severity value used to characterize the average severity of the historical patients' illness; A prediction unit, configured to predict the number of patients and the predicted severity value corresponding to each department based on a multi-output prediction strategy and the historical status information corresponding to each department; A setting unit, configured to set a weight factor based on the corresponding predicted number of patients and an importance factor based on the corresponding predicted severity value for each department; A modeling unit, used for obtaining the existing quantity of various medical supplies, and constructing a scheduling model based on each of the weight factors, each of the importance factors and the existing quantity of various medical supplies; the scheduling model is composed of an objective function and constraints; the scheduling target of the objective function includes: making the quantity of various medical supplies allocated to each department positively correlated with each of the weight factors, and making the similarity between the proportion of various medical supplies allocated to each department and each of the weight factors exceed a similarity threshold; the constraints are used to constrain the total amount of any medical supply allocated to each department not to exceed the corresponding existing quantity, and constrain the quantity of various medical supplies allocated to any department to be not less than 0; The solving unit is used to solve the scheduling model using the simplex method to obtain the allocation result.

7. The device according to claim 6, characterized in that The prediction unit is specifically used for: The historical status information of each department is input into a pre-trained neural network model to predict the future status information of each department; the future status information includes: predicted number of patients and predicted severity value; the neural network model is composed of an input layer, a hidden layer and an output layer, the hidden layer is composed of an LSTM layer and a Dense layer, and the Dense layer uses a RELU function as an activation function.

8. The device according to claim 7, characterized in that The device also includes: The training unit is used to construct a neural network model, and compile it based on the neural network model, the mean square error function and the Adam optimizer to obtain the neural network model to be trained; the neural network model to be trained is trained using a pre-acquired training set until a trained neural network model is obtained.

9. The device according to claim 6, characterized in that The modeling unit is specifically used for: Based on each of the weight factors, each of the importance factors and the existing material quantity of various medical materials, a linear programming task including two objective items is constructed; Based on the weighted combination method and the linear programming task, an objective function and constraint conditions are constructed; Based on the objective function and the constraints, a scheduling model is constructed.

10. The device according to claim 6, characterized in that The solving unit is specifically used for: Selecting a plurality of initial feasible bases from the feasible domain of the objective function; Construct a simplex table based on each initial feasible basis; Check whether the current simplex table is the optimal solution; If yes, the current simplex table is used as the allocation result; If not, the current simplex table is converted into adjacent feasible basis solutions, and the process returns to the step of checking whether the current simplex table is the optimal solution, until the current simplex table is the optimal solution.

Images