Big data analysis method based on hospital human resource digital intelligence simulation model
Through the digital simulation model of hospital human resources, combined with doctor data and patient traffic prediction, a reasonable scheduling plan is generated, which solves the problems of insufficient doctor resources and unreasonable scheduling, and achieves scientific and balanced scheduling optimization, improving the quality of hospital services and doctor satisfaction.
Patent Information
- Application Number
- CN202510552453.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-29
- Publication Date
- 2025-08-22
AI Technical Summary
In the prior art, doctors have insufficient resources and unreasonable scheduling methods, which leads to doctors not being able to get enough rest, work pressure is high, and the emergence of multi-skilled doctors makes scheduling problems more complex and difficult to optimize.
By using the digital simulation model of hospital human resources, doctor skills and attendance data are collected, combined with patient traffic prediction and shift scheduling needs, data preprocessing, patient traffic prediction model and balanced shift scheduling model are used to generate reasonable and balanced shift scheduling results.
It improves the scientificity and rationality of scheduling, ensures that the doctor's rest time and workload matches the department's needs, optimizes the allocation of hospital resources, and improves service quality and doctor's work satisfaction.
Smart Images
Figure CN120526952A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of data analysis technology, and specifically relates to a big data analysis method based on a digital simulation model of hospital human resources. Background Art
[0002] With the deepening of medical system reform, my country's physician workforce continues to grow, but a significant gap remains between the total number of physicians and the growing healthcare needs of the population. Faced with residents' growing demand for medical services and a relatively tight medical staff, hospitals are forced to increase the workload of their medical staff to accommodate the increased number of patients. In the current medical system, medical staff are unable to maintain their own health, facing high work pressure and overwork. In addition to the relatively limited number of physicians, irrational scheduling methods are also a major factor preventing physicians from getting adequate rest. Traditional manual scheduling is challenging to ensure adequate rest for physicians while ensuring fairness and rationality, which hinders their work enthusiasm. Furthermore, with the emergence of multi-skilled physicians, scheduling becomes increasingly complex. Different physicians may provide the skills required for different positions, making scheduling even more difficult to coordinate and optimize. Therefore, how to fully consider the characteristics of multi-skilled physicians and implement reasonable and balanced scheduling has become an urgent issue. Summary of the Invention
[0003] In order to solve the above problems existing in the prior art, the present invention provides a big data analysis method based on the digital simulation model of hospital human resources. The purpose of the present invention can be achieved through the following technical solutions: A big data analysis method based on a digital simulation model of hospital human resources, including: The hospital human resources digital simulation model is used to collect doctors' skills data and doctors' attendance data; doctors' information preprocessing data is obtained through the data preprocessing model based on the doctors' skills data and the doctors' attendance data; Acquire historical patient flow data and historical online appointment data, and obtain patient flow prediction data using a patient flow prediction model based on the historical patient flow data and the historical online appointment data; Obtaining scheduling requirements and department requirements, and converting pre-processed data into a scheduling constraint set through a logical formula based on the scheduling requirements, the department requirements, and the doctor information; A balanced scheduling result is obtained through a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
[0004] Preferably, the obtaining of doctor information preprocessing data based on the doctor's skill data and the doctor's attendance data through a data preprocessing model includes: Obtain complete data on the doctor's skills and complete data on the doctor's attendance based on the doctor's skills data and the doctor's attendance data by processing missing values; Obtaining complete coding data of the doctor's skills and complete coding data of the doctor's attendance through one-hot encoding according to the complete data of the doctor's skills and the complete data of the doctor's attendance; Obtain normal data on the mastery of the doctor's skills and normal data on the doctor's attendance by detecting abnormal values based on the complete coded data on the doctor's mastery of the skills and the complete coded data on the doctor's attendance; The doctor information preprocessing data is obtained by merging the normal data on the doctor's skills and the normal data on the doctor's attendance according to the doctor's name.
[0005] Preferably, obtaining the patient flow prediction data by using a patient flow prediction model according to the historical patient flow data and the historical online appointment data includes: The patient flow model includes an offline flow prediction model and an online no-show prediction model, and the patient flow prediction data includes offline patient flow prediction data and online patient flow prediction data; Obtaining the offline patient flow prediction data through the offline flow prediction model according to the historical patient flow data; The online patient flow prediction data is obtained based on the historical online appointment data through the online no-show prediction model.
[0006] Preferably, obtaining the offline patient flow prediction data through the offline flow prediction model according to the historical patient flow data includes: Obtaining historical patient flow time series data by adding a timestamp according to the historical patient flow data; According to the historical patient flow time series data, historical patient flow training time series data and historical patient flow test time series data are obtained by dividing the data set; Presetting an initial offline traffic prediction model, wherein the initial offline traffic prediction model includes a convolutional neural network layer, a long short-term memory network layer, and an attention layer; Extracting features from the historical patient flow training time series data through the convolutional neural network layer to obtain historical patient flow training data features; According to the characteristics of the historical patient flow training data, time series modeling is performed through the long short-term memory network layer to obtain the time series characteristics of the historical patient flow training data; Performing weighted processing on the temporal features of the historical patient flow training data through the attention layer to obtain temporal enhancement features of the historical patient flow training data; Obtaining predicted patient flow data using the initial offline flow prediction model based on the time series enhancement features of the historical patient flow training data; Acquire real patient flow data, and train the initial offline flow prediction model using a patient flow loss function according to the real patient flow data and the predicted patient flow data to obtain the offline flow prediction model; The patient flow loss function expression is: , in, L(y,`y) represents the patient flow loss function, y is the actual patient flow data, y is the predicted patient flow data, w t is the weight, d t To dynamically adjust the error threshold, l To control the decay rate factor, α is a hyperparameter used to balance the error range. median Indicates the median operation. c is the time decay factor, t represents the time step, e represents the logarithmic function; The offline patient flow prediction data is obtained by using the offline flow prediction model according to the historical patient flow test time series data.
[0007] Preferably, obtaining the online patient flow prediction data through the online no-show prediction model based on the historical online appointment data includes: Obtaining historical online reservation time series data by adding timestamps according to the historical online reservation data; According to the historical online reservation time series data, historical online reservation training time series data and historical online reservation test time series data are obtained by dividing the data set; A basic classification result is obtained by performing a three-fold cross validation calculation on the basic classifier based on the historical online reservation training time series data; A meta-classifier is trained by using the basic classification results and the historical online appointment training time series data to obtain a no-show meta-classifier; The online patient flow prediction data is obtained by using the no-show meta-classifier based on the historical online appointment test time series data.
[0008] Preferably, the basic classifiers include a Logistics regression classifier, an XGBoost classifier, and a multi-layer perceptron classifier; and the meta-classifier is an XGBoost classifier.
[0009] Preferably, the scheduling constraint set obtained by converting the pre-processed data according to the scheduling requirements, the department requirements, and the doctor information through a logical formula includes: The logic formula conversion is expressed as: , in, minutes Indicates minimum value calculation, z represents the objective function, i represents the i-th doctor, t represents the t-th doctor, I represents the set of all doctors, s i represents the total work intensity of the i-th doctor in the scheduling cycle, s t represents the total work intensity of the t-th doctor in the scheduling cycle, k represents the kth position, K Indicates the number of positions, x ijk Represents a decision variable, indicating that when the i-th doctor is assigned to take the k-th position on the j-th day, xijk is 1, otherwise it is 0. j represents the jth day in the scheduling cycle, J Indicates the set of all shifts in the scheduling cycle. I k represents the set of doctors who are qualified for the k-th job, S i Indicates the total workload of each doctor during the scheduling period, R Indicates the number of full days of rest, Indicates the lower limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle, Indicates the upper limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle. G represents the workload difference threshold, 、 、 is the coefficient.
[0010] Preferably, obtaining a balanced scheduling result through a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information pre-processing data includes: The patient flow prediction data and the doctor information pre-processing data are used as states, and the shift adjustment is used as an action; Calculating a reward by weighted summation according to the scheduling constraint set; The balanced scheduling result is obtained by optimizing the state, the action, and the reward through the DDQN algorithm.
[0011] A big data analysis system based on a digital simulation model of hospital human resources, including a data preprocessing module, a traffic prediction module, a scheduling constraint module, and a balanced scheduling module, including: The data preprocessing module is used to collect doctor skill mastery data and doctor attendance data through the hospital human resources digital simulation model; and obtain doctor information preprocessing data based on the doctor skill mastery data and the doctor attendance data through the data preprocessing model; The flow prediction module is used to obtain historical patient flow data and historical online appointment data, and obtain patient flow prediction data through a patient flow prediction model based on the historical patient flow data and the historical online appointment data; The scheduling constraint module is used to obtain scheduling requirements and department requirements, and to convert the pre-processed data of the scheduling requirements, department requirements, and doctor information into a scheduling constraint set through a logical formula; The balanced scheduling module is used to obtain a balanced scheduling result through a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
[0012] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the big data analysis method based on the digital simulation model of hospital human resources is implemented.
[0013] A storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the above-mentioned big data analysis method based on the digital simulation model of hospital human resources.
[0014] The beneficial effects of the present invention are: (1) The patient flow is predicted through the offline flow prediction model and the online no-show flow prediction model respectively, taking into account the situation of patients no-showing online, improving the accuracy of the prediction and providing a basis for subsequent scheduling.
[0015] (2) By generating scheduling constraints based on scheduling needs, department needs, and doctor information data, and fully considering doctors' rest time and the characteristics of multi-skilled doctors, the rationality and scientificity of scheduling are improved, ensuring that doctors' workloads match department needs.
[0016] Through patient flow prediction, scheduling constraints, and reinforcement learning models, on the basis of accurately predicting future patient flow, scheduling is adaptively adjusted according to scheduling constraints through reinforcement learning models, optimizing hospital resource allocation, and improving hospital service quality and doctor job satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.
[0018] Figure 1 This is a flow chart of a big data analysis method based on a digital simulation model of hospital human resources according to the present invention. DETAILED DESCRIPTION
[0019] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.
[0020] See also Figure 1 , a big data analysis method based on the digital simulation model of hospital human resources, including: S1: Collecting doctors' skills data and doctors' attendance data through the hospital human resources digital simulation model; obtaining doctors' information preprocessing data based on the doctors' skills data and the doctors' attendance data through the data preprocessing model; S2: Acquire historical patient flow data and historical online appointment data, and obtain patient flow prediction data using a patient flow prediction model based on the historical patient flow data and the historical online appointment data; S3: Obtaining scheduling requirements and department requirements, and converting the pre-processed data into a scheduling constraint set through a logical formula based on the scheduling requirements, the department requirements, and the doctor information; S4: Obtaining a balanced scheduling result through a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
[0021] Specifically, the department demand data includes the functions of doctors required by the department and the number of doctors required by the department.
[0022] Specifically, the hospital human resources digital simulation model includes a recruitment management module, a daily business management module, a salary management module, a training management module, a performance appraisal module, and a talent evaluation module; the recruitment management module includes a position posting module, an application management module, a recruitment management module, and a talent pool module; the daily business management module includes a personnel file module, a position management module, a daily business module for entry, transfer and departure, an attendance management module, and a professional title management module; the salary management module includes a basic setting module, a salary payment module, a retention management module, a tax processing module, and a salary report module; the doctor's attendance data is obtained through the attendance management module; the training management module includes a training plan module, a courseware module, a learning plan module, a question bank management module, an examination management module, a grade certificate management module, and a statistical query module; the doctor's skill mastery data is obtained through the grade certificate management module; the performance appraisal module includes a performance plan setting module, a performance indicator collection module, a performance bonus accounting module, and a performance bonus distribution module; the talent evaluation module includes an evaluation indicator setting module, an evaluation data collection module, an evaluation result generation module, and an evaluation result application module.
[0023] In step S1, the data preprocessing model includes: S101: Obtaining complete data on the doctor's skills and complete data on the doctor's attendance by processing missing values according to the doctor's skills data and the doctor's attendance data; S102: Obtaining complete coding data of the doctor's skills and complete coding data of the doctor's attendance by one-hot encoding according to the complete data of the doctor's skills and the complete data of the doctor's attendance; S103: Obtaining normal data on the mastery of the doctor's skills and normal data on the doctor's attendance through abnormal value detection based on the complete coded data on the doctor's mastery of the skills and the complete coded data on the doctor's attendance; S104: The doctor information pre-processing data is obtained by merging the normal data on the doctor's skills and the normal data on the doctor's attendance according to the doctor's name.
[0024] In step S2, the patient flow prediction model includes: S201: The patient flow model includes an offline flow prediction model and an online no-show prediction model, and the patient flow prediction data includes offline patient flow prediction data and online patient flow prediction data; S202: Obtaining offline patient flow prediction data using the offline flow prediction model according to the historical patient flow data; S203: Obtain the online patient flow prediction data according to the historical online appointment data through the online no-show prediction model.
[0025] Specifically, the offline traffic prediction model training process includes: S202-1: Obtaining historical patient flow time series data by adding timestamps according to the historical patient flow data; S202-2: obtaining historical patient flow rate training time series data and historical patient flow rate test time series data by dividing the data set according to the historical patient flow rate time series data; S202-3: Preset an initial offline traffic prediction model, where the initial offline traffic prediction model includes a convolutional neural network layer, a long short-term memory network layer, and an attention layer; S202-4: Extracting features from the historical patient flow training time series data through the convolutional neural network layer to obtain historical patient flow training data features; S202-5: Performing time series modeling through the long short-term memory network layer according to the historical patient flow training data characteristics to obtain the historical patient flow training data time series characteristics; S202-6: Performing weighted processing on the time series features of the historical patient flow training data through the attention layer to obtain enhanced time series features of the historical patient flow training data; S202-7: Obtaining predicted patient flow data using the initial offline flow prediction model based on the time series enhancement features of the historical patient flow training data; S202-8: Acquire real patient flow data, and train the initial offline flow prediction model using a patient flow loss function according to the real patient flow data and the predicted patient flow data to obtain the offline flow prediction model; The patient flow loss function expression is: , in, L(y,`y) represents the patient flow loss function, y is the actual patient flow data, y is the predicted patient flow data, w t is the weight, d t To dynamically adjust the error threshold, l To control the decay rate factor, α is a hyperparameter used to balance the error range. median Indicates the median operation. c is the time decay factor, t represents the time step, e represents the logarithmic function; S202-8: Obtain the offline patient flow prediction data through the offline flow prediction model according to the historical patient flow test time series data.
[0026] In this embodiment, the offline traffic prediction model includes a 3×3 convolution kernel, 32 filters, a 3×3 pooling kernel, the convolutional neural network layer contains 20 neurons, and the long short-term memory network layer contains 30 neurons. Each layer contains a dropout layer with a value of 0.2, indicating that approximately 20% of the neurons in each layer are automatically and randomly inactivated during neural network training. The neural network hyperparameters are: a learning rate of 0.01, a batch size of 512, and 200 iterations.
[0027] Specifically, the online no-show prediction model includes: S203-1: Obtaining historical online reservation time series data by adding timestamps to the historical online reservation data; S203-2: obtaining historical online reservation training time series data and historical online reservation test time series data by dividing the data set according to the historical online reservation time series data; S203-3: Obtaining a basic classification result by performing a three-fold cross validation calculation on a basic classifier based on the historical online appointment training time series data; S203-4: training a meta-classifier using the basic classification results and the historical online appointment training time series data to obtain a no-show meta-classifier; S203-5: Obtain the online patient flow prediction data through the missed appointment meta-classifier based on the historical online appointment test time series data, where the online patient flow prediction data is the data of patients who have not missed their appointments.
[0028] Specifically, the basic classifiers include a Logistics regression classifier, an XGBoost classifier, and a multi-layer perceptron classifier; and the meta-classifier is an XGBoost classifier.
[0029] In step S3, the pre-processed data according to the scheduling requirements, the department requirements, and the doctor information is converted into a scheduling constraint set through a logical formula, including: The logic formula conversion is expressed as: , in, minutes Indicates minimum value calculation, z represents the objective function, i represents the i-th doctor, t represents the t-th doctor, I represents the set of all doctors, s i represents the total work intensity of the i-th doctor in the scheduling cycle, s t represents the total work intensity of the t-th doctor in the scheduling cycle, krepresents the kth position, K Indicates the number of positions, x ijk Represents a decision variable, indicating that when the i-th doctor is assigned to take the k-th position on the j-th day, xijk is 1, otherwise it is 0. j represents the jth day in the scheduling cycle, J Indicates the set of all shifts in the scheduling cycle. I k represents the set of doctors who are qualified for the k-th job, S i Indicates the total workload of each doctor during the scheduling period, R Indicates the number of full days of rest, Indicates the lower limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle, Indicates the upper limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle. G represents the workload difference threshold, 、 、 is the coefficient.
[0030] In this embodiment, the first formula is the objective function, which means minimizing the difference in work intensity between doctors; the second formula means that each doctor can only take on the work of one position in a shift; the third formula means ensuring that there are no vacancies in each position in each shift and that doctors who do not have the skills for that position are not assigned to that position; the fourth formula means that after a doctor finishes the night shift, he or she can only rest during the morning and afternoon shifts of the next day; the fifth formula means that within a scheduling cycle, each doctor can rest for at least R full days; the sixth formula means that within a scheduling cycle, the number of times a doctor who is competent for the k-th position is assigned to the k-th position is no more than the upper limit and no less than the lower limit of the department's scheduling regulations; the seventh formula means that within a scheduling cycle, the workload difference between doctors who take on the same position is less than the workload difference threshold; the eighth formula means the total intensity of each doctor's work within a scheduling cycle.
[0031] In step S4, the balanced scheduling model includes: S401: taking the patient flow prediction data and the doctor information pre-processing data as states and taking shift adjustment as actions; The shift adjustment is expressed as: , Wherein, Action represents the action, which is to schedule doctor a to department c in the bth time period, where a is the doctor number, b is the time period number, and c is the department number; S402: Obtaining a reward by weighted sum calculation according to the scheduling constraint set; S403: Optimizing the state, the action, and the reward using the DDQN algorithm to obtain the balanced scheduling result.
[0032] A big data analysis system based on a digital simulation model of hospital human resources, including a data preprocessing module, a traffic prediction module, a scheduling constraint module, and a balanced scheduling module, including: The data preprocessing module is used to collect doctor skill mastery data and doctor attendance data through the hospital human resources digital simulation model; and obtain doctor information preprocessing data based on the doctor skill mastery data and the doctor attendance data through the data preprocessing model; The flow prediction module is used to obtain historical patient flow data and historical online appointment data, and obtain patient flow prediction data through a patient flow prediction model based on the historical patient flow data and the historical online appointment data; The scheduling constraint module is used to obtain scheduling requirements and department requirements, and to convert the pre-processed data of the scheduling requirements, department requirements, and doctor information into a scheduling constraint set through a logical formula; The balanced scheduling module is used to obtain a balanced scheduling result through a balanced scheduling model based on the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
[0033] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.
[0034] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0035] The program code included in the computer-readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF or the like, or any suitable combination thereof. The computer program code for performing the operation of the present invention can be written in one or more programming languages or a combination thereof, and the programming language includes an object-oriented programming language such as Java, Smalltalk, C++, and also includes a conventional procedural programming language such as "C" language or similar programming language. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, utilizing an Internet service provider to connect through the Internet).
[0036] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.
Claims
1. A big data analysis method based on a digital simulation model of hospital human resources, characterized by: The following steps are involved: Collect data on doctors’ skills and attendance through the hospital’s human resources digital simulation model; Obtaining doctor information preprocessing data based on the doctor's skill data and the doctor's attendance data through a data preprocessing model; Acquire historical patient flow data and historical online appointment data, and obtain patient flow prediction data using a patient flow prediction model based on the historical patient flow data and the historical online appointment data; Obtaining scheduling requirements and department requirements, and converting pre-processed data into a scheduling constraint set through a logical formula based on the scheduling requirements, the department requirements, and the doctor information; A balanced scheduling result is obtained through a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
2. The big data analysis method based on the hospital human resources digital simulation model according to claim 1 is characterized in that: The doctor information preprocessing data obtained by using a data preprocessing model based on the doctor's skill data and the doctor's attendance data includes: Obtain complete data on the doctor's skills and complete data on the doctor's attendance based on the doctor's skills data and the doctor's attendance data by processing missing values; Obtaining complete coding data of the doctor's skills and complete coding data of the doctor's attendance through one-hot encoding according to the complete data of the doctor's skills and the complete data of the doctor's attendance; Obtain normal data on the mastery of the doctor's skills and normal data on the doctor's attendance by detecting abnormal values based on the complete coded data on the doctor's mastery of the skills and the complete coded data on the doctor's attendance; Doctor information preprocessing data is obtained by merging the normal data on the doctor's skills and the normal data on the doctor's attendance according to the doctor's name.
3. The big data analysis method based on the hospital human resources digital simulation model according to claim 1 is characterized in that: The obtaining of patient flow prediction data by a patient flow prediction model based on the historical patient flow data and the historical online appointment data includes: The patient flow model includes an offline flow prediction model and an online no-show prediction model, and the patient flow prediction data includes offline patient flow prediction data and online patient flow prediction data; Obtaining the offline patient flow prediction data through the offline flow prediction model according to the historical patient flow data; The online patient flow prediction data is obtained based on the historical online appointment data through the online no-show prediction model.
4. The big data analysis method based on the hospital human resources digital simulation model according to claim 3 is characterized in that: Obtaining the offline patient flow prediction data through the offline flow prediction model according to the historical patient flow data includes: Obtaining historical patient flow time series data by adding a timestamp according to the historical patient flow data; According to the historical patient flow time series data, historical patient flow training time series data and historical patient flow test time series data are obtained by dividing the data set; Presetting an initial offline traffic prediction model, wherein the initial offline traffic prediction model includes a convolutional neural network layer, a long short-term memory network layer, and an attention layer; Extracting features from the historical patient flow training time series data through the convolutional neural network layer to obtain historical patient flow training data features; According to the characteristics of the historical patient flow training data, time series modeling is performed through the long short-term memory network layer to obtain the time series characteristics of the historical patient flow training data; Performing weighted processing on the temporal features of the historical patient flow training data through the attention layer to obtain temporal enhancement features of the historical patient flow training data; Obtaining predicted patient flow data using the initial offline flow prediction model based on the time series enhancement features of the historical patient flow training data; Acquire real patient flow data, and train the initial offline flow prediction model using a patient flow loss function according to the real patient flow data and the predicted patient flow data to obtain the offline flow prediction model; The patient flow loss function expression is: ， in, L(y,`y) represents the patient flow loss function, y is the actual patient flow data, y is the predicted patient flow data, w t is the weight, δ t To dynamically adjust the error threshold, λ To control the decay rate factor, α is a hyperparameter used to balance the error range. median Indicates the median operation. γ is the time decay factor, t represents the time step, e represents the logarithmic function; The offline patient flow prediction data is obtained by using the offline flow prediction model according to the historical patient flow test time series data.
5. The big data analysis method based on the hospital human resources digital simulation model according to claim 3 is characterized in that: The obtaining of the online patient flow prediction data by the online no-show prediction model based on the historical online appointment data includes: Obtaining historical online reservation time series data by adding timestamps according to the historical online reservation data; According to the historical online reservation time series data, historical online reservation training time series data and historical online reservation test time series data are obtained by dividing the data set; A basic classification result is obtained by performing a three-fold cross validation calculation on the basic classifier based on the historical online reservation training time series data; A meta-classifier is trained by using the basic classification results and the historical online appointment training time series data to obtain a no-show meta-classifier; The online patient flow prediction data is obtained by using the no-show meta-classifier based on the historical online appointment test time series data.
6. The big data analysis method based on the hospital human resources digital simulation model according to claim 5 is characterized in that: The basic classifiers include Logistics regression classifier, XGBoost classifier, and Multilayer Perceptron classifier; the meta-classifier is XGBoost classifier.
7. The big data analysis method based on the hospital human resources digital simulation model according to claim 1 is characterized in that: The scheduling constraint set obtained by converting the pre-processed data according to the scheduling requirements, the department requirements, and the doctor information through a logical formula includes: The logic formula conversion is expressed as: ， in, min Indicates minimum value calculation, z represents the objective function, i represents the i-th doctor, t represents the t-th doctor, I represents the set of all doctors, s i represents the total work intensity of the i-th doctor in the scheduling cycle, s t represents the total work intensity of the t-th doctor in the scheduling cycle, k represents the kth position, K Indicates the number of positions, x ijk Represents a decision variable, indicating that when the i-th doctor is assigned to take the k-th position on the j-th day, xijk is 1, otherwise it is 0. j represents the jth day in the scheduling cycle, J Indicates the set of all shifts in the scheduling cycle. I k represents the set of doctors who are qualified for the k-th job, S i Indicates the total workload of each doctor during the scheduling period, R Indicates the number of full days of rest, Indicates the lower limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle, Indicates the upper limit of the number of times the i-th doctor is assigned to the k-th position within a scheduling cycle. G represents the workload difference threshold, 、 、 is the coefficient.
8. The big data analysis method based on the hospital human resources digital simulation model according to claim 1 is characterized in that: The balanced scheduling result obtained by using a balanced scheduling model according to the scheduling constraint set, the patient flow prediction data, and the doctor information pre-processing data includes: The patient flow prediction data and the doctor information pre-processing data are used as states, and the shift adjustment is used as an action; Calculating a reward by weighted summation according to the scheduling constraint set; The balanced scheduling result is obtained by optimizing the state, the action, and the reward through the DDQN algorithm.
9. A big data analysis system based on a digital simulation model of hospital human resources, comprising a data preprocessing module, a flow prediction module, a scheduling constraint module, and a balanced scheduling module, characterized in that: include: The data preprocessing module is used to collect doctors' skills data and doctors' attendance data through the hospital human resources digital simulation model; Obtaining doctor information preprocessing data based on the doctor's skill data and the doctor's attendance data through a data preprocessing model; The flow prediction module is used to obtain historical patient flow data and historical online appointment data, and obtain patient flow prediction data through a patient flow prediction model based on the historical patient flow data and the historical online appointment data; The scheduling constraint module is used to obtain scheduling requirements and department requirements, and to convert the pre-processed data of the scheduling requirements, department requirements, and doctor information into a scheduling constraint set through a logical formula; The balanced scheduling module is used to obtain a balanced scheduling result through a balanced scheduling model based on the scheduling constraint set, the patient flow prediction data, and the doctor information preprocessing data.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the big data analysis method based on the hospital human resources digital simulation model as described in any one of claims 1 to 8 is implemented; A storage medium containing computer-executable instructions, characterized in that: When executed by a computer processor, the computer executable instructions are used to execute the big data analysis method based on the hospital human resources digital simulation model as described in any one of claims 1-8.