Clinical intelligent scheduling method and system based on big data analysis

Abstract

The invention relates to a clinical intelligent scheduling method and system based on big data analysis, and relates to the technical field of data processing.The method comprises the steps that a data set of patient data, nurse data and environment data is obtained, and the data set is preprocessed to generate a standardized data set; processing the standardized data set, and predicting nursing requirements of future scheduling; generating a hierarchical nurse distribution scheme through a recommendation algorithm based on the skills of nurses, the current workload and the historical work data; generating a scheduling table meeting hierarchical nurse allocation schemes and nursing requirements by adopting a genetic algorithm in combination with dynamic constraint conditions; based on the scheduling data and the physiological indexes of the scheduling table, dynamically adjusting the workload of each nurse through a supervised learning model, and generating a first scheduling strategy; and automatically matching a substitute nurse and optimizing the first scheduling strategy through the reinforcement learning model in an emergency, and generating a second scheduling strategy. The efficiency and accuracy of clinical scheduling can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a clinical intelligent scheduling method, system, electronic device and non-transitory computer-readable storage medium based on big data analysis. Background Art

[0002] At present, clinical nursing scheduling mainly relies on manual experience for deployment. Head nurses need to comprehensively consider the balance of working hours, skill matching, and possible emergencies to ensure the smooth progress of nursing work.

[0003] However, this method is highly dependent on personal experience, which not only has a large workload and low efficiency, but is also prone to unreasonable scheduling due to human negligence, affecting the quality of care. In addition, traditional scheduling methods mostly use fixed shifts, which are difficult to flexibly adapt to fluctuations in patient needs, such as increased nighttime care needs for postoperative patients or centralized arrangements for rehabilitation training, which can easily lead to waste or shortage of nursing resources. Summary of the invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a clinical intelligent scheduling method, system, electronic device and non-transitory computer-readable storage medium based on big data analysis, which can improve the efficiency and accuracy of scheduling.

[0005] The technical solution of the present invention to solve the above technical problems is as follows: The present invention provides a clinical intelligent scheduling method based on big data analysis, the method comprising: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set; Processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; Generate a hierarchical nurse allocation plan through a recommendation algorithm based on nurses' skills, current workload, and historical work data; Genetic algorithms are used in combination with dynamic constraints to generate a shift schedule that meets the hierarchical nurse allocation plan and the nursing needs; Based on the scheduling data and physiological indicators of the schedule, dynamically adjusting the workload of each nurse through a supervised learning model to generate a first scheduling strategy; Through the reinforcement learning model, substitute nurses are automatically matched in emergency situations and the first scheduling strategy is optimized to generate a second scheduling strategy.

[0006] Optionally, preprocessing the data set to generate a standardized data set includes: The nurse data are processed using a K-means clustering algorithm, and the nurses are grouped according to their skill level and work intensity to obtain grouped nurse data; The PCA dimension reduction algorithm is used to reduce the dimension of the multi-dimensional feature data of the grouped nurse data, the patient data and the environmental data and extract key features to obtain the standardized data set.

[0007] Optionally, the time series prediction model is an LSTM network, and the decision tree model is an XGBoost algorithm; The method of processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts includes: The standardized data set is processed using the LSTM network to predict fluctuations in nursing needs of postoperative patients at night; Using the XGBoost algorithm to process the standardized data set to predict the nursing demand level for future shifts; The nursing demand for the future shift is determined according to the nursing demand fluctuation and the nursing demand level.

[0008] Optionally, the hierarchical nurse allocation scheme is generated by a recommendation algorithm based on the nurses' skills, current workload and historical work data, including: Collaborative filtering recommendation algorithm, matching each nurse’s fitness with the scheduled shift based on historical scheduling data; KNN algorithm,calculates the workload similarity among nurses; The hierarchical nurse allocation plan is generated according to the fitness and the workload similarity.

[0009] Optionally, the dynamic constraints of the genetic algorithm include daily working hours constraints, weekly shift days constraints and consecutive rest days constraints.

[0010] Optionally, the matching of each nurse's adaptability to the scheduled shift based on historical scheduling data includes: A first fitness function for maximizing the weighted sum of nurse satisfaction scores and nursing needs coverage is constructed; constructing a second fitness function for minimizing the product of the sum of the nurse fatigue index and the number of shift adjustments; The fitness is determined according to a function value of the first fitness function and a function value of the second fitness function.

[0011] Optionally, the physiological indicator includes heart rate variability data, and the supervised learning model is a random forest classifier; The scheduling data and physiological indicators based on the schedule dynamically adjust the workload of each nurse through a supervised learning model to generate a first scheduling strategy, including: Calculating the fatigue index of the nurses by using the random forest classifier in combination with the shift scheduling data and the heart rate variability data of the nurses; The workload of each nurse is dynamically adjusted according to the fatigue index of the nurse to generate the first scheduling strategy.

[0012] Optionally, the reinforcement learning model is a Deep Q-Learning model; the method of automatically matching a substitute nurse in an emergency and optimizing the first scheduling strategy by the reinforcement learning model to generate a second scheduling strategy includes: Obtaining a linear combination of the skill matching degree, fatigue index and scheduling conflict rate of each substitute nurse; The first scheduling strategy is optimized according to a linear combination of the skill matching degree, fatigue index and scheduling time conflict rate of each substitute nurse, with the optimization goal of minimizing the decline in nursing quality and the additional workload of nurses in the current scheduling, thereby generating the second scheduling strategy.

[0013] Optionally, the nurse data includes the nurse's skill level, scheduling preference, night shift acceptance flag and historical work log, and the environmental data includes the predicted increase in nursing demand during seasonal disease peaks.

[0014] In addition, to achieve the above purpose, the present invention also proposes a clinical intelligent scheduling system based on big data analysis, the system comprising: A data acquisition module is used to acquire a data set of patient data, nurse data and environmental data, and preprocess the data set to generate a standardized data set; A demand forecasting module, used to process the standardized data set using a time series forecasting model and a decision tree model to predict nursing needs for future shifts; A hierarchical allocation module is used to generate a hierarchical nurse allocation plan through a recommendation algorithm based on the nurses' skills, current workload and historical work data; A preliminary scheduling module, for generating a schedule that meets the hierarchical nurse allocation scheme and the nursing needs by using a genetic algorithm combined with dynamic constraints; A first scheduling module, for dynamically adjusting the workload of each nurse through a supervised learning model based on the scheduling data and physiological indicators of the scheduling table, and generating a first scheduling strategy; The second scheduling module is used to automatically match substitute nurses in emergency situations and optimize the first scheduling strategy through a reinforcement learning model to generate a second scheduling strategy.

[0015] In addition, to achieve the above-mentioned purpose, the present invention also proposes an electronic device, comprising: a memory for storing a computer software program; a processor for reading and executing the computer software program, thereby realizing a clinical intelligent scheduling method based on big data analysis as described above.

[0016] In addition, to achieve the above-mentioned purpose, the present invention also proposes a non-transitory computer-readable storage medium, in which a computer software program is stored. When the computer software program is executed by a processor, a clinical intelligent scheduling method based on big data analysis as described above is implemented.

[0017] The beneficial effects of the present invention are: (1) The present invention intelligently matches nurse resources and assigns appropriate nurses to appropriate patient care tasks based on the nurses' skills, experience, and workload, ensuring that patients can receive the most professional nursing services. It can predict nursing needs in the future based on historical data and real-time data, dynamically adjust the number of nurses, avoid waste or shortage of nursing resources, and ensure that patients' nursing needs can be met at different time periods.

[0018] (2) The present invention generates the optimal shift schedule through genetic algorithm, taking into account the nurses' work preferences, fatigue index and working time constraints, ensuring that each nurse's workload is balanced, avoiding overwork of individual nurses, and meeting the nurses' rest needs as much as possible. The goal of the genetic algorithm is "minimum work conflict and minimum adjustment cost". By dynamically adjusting the number of nurses and combining the results of nursing demand prediction, it can effectively reduce work conflicts caused by unreasonable scheduling and improve nurses' job satisfaction.

[0019] (3) The present invention can automatically calculate the nurses’ working hours and the number of night shifts, automatically identify statutory holidays and nurses’ birthdays, nurses’ scheduled vacation time, and optimize attendance management. The random forest classification model is used to predict whether nurses need to adjust their shifts due to fatigue or health reasons, which reduces the workload of manual attendance management and improves management efficiency and accuracy.

[0020] In summary, the present invention realizes efficient management and optimized scheduling of nursing resources through intelligent and data-driven methods, which not only improves the quality and efficiency of nursing, but also enhances the professional satisfaction of nurses and patient satisfaction, while reducing management costs and enhancing the overall operational capacity of the hospital. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A scene diagram of a clinical intelligent scheduling method based on big data analysis provided by the present invention; Figure 2 A flowchart of a clinical intelligent scheduling method based on big data analysis provided by the present invention; Figure 3 A schematic diagram of the structure of a clinical intelligent scheduling system based on big data analysis provided by the present invention; Figure 4 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention; Figure 5 A schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0022] 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0023] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0024] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0025] See also Figure 1 , Figure 1 This is a scene diagram of a clinical intelligent scheduling method based on big data analysis provided by the present invention. Figure 1As shown, the terminal and the server are connected via a network, such as a wired or wireless network connection. The terminal may include but is not limited to portable terminals such as mobile phones and tablets installed with various network platform applications, as well as fixed terminals such as computers, query machines, and advertising machines. The server provides users with various business services, including service push servers, user recommendation servers, etc.

[0026] It should be noted that Figure 1 The scenario diagram of a clinical intelligent scheduling method based on big data analysis shown is only an example. The terminal, server and application scenario described in the embodiment of the present invention are for more clearly illustrating the technical solution of the embodiment of the present invention, and do not generate limitations on the technical solution provided by the embodiment of the present invention. A person of ordinary skill in the art can know that with the evolution of the system and the emergence of new business scenarios, the technical solution provided by the embodiment of the present invention is also applicable to similar technical problems.

[0027] Among them, the terminal can be used for: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set; Processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; Generate a hierarchical nurse allocation plan through a recommendation algorithm based on nurses' skills, current workload, and historical work data; Genetic algorithms are used in combination with dynamic constraints to generate a shift schedule that meets the hierarchical nurse allocation plan and the nursing needs; Based on the scheduling data and physiological indicators of the schedule, dynamically adjusting the workload of each nurse through a supervised learning model to generate a first scheduling strategy; Through the reinforcement learning model, substitute nurses are automatically matched in emergency situations and the first scheduling strategy is optimized to generate a second scheduling strategy.

[0028] See also Figure 2 , provides a flowchart of a clinical intelligent scheduling method based on big data analysis of the present invention, comprising the following steps: Step 201: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set.

[0029] In some embodiments, step 201 may include: The nurse data are processed using a K-means clustering algorithm, and the nurses are grouped according to their skill level and work intensity to obtain grouped nurse data; The PCA dimension reduction algorithm is used to reduce the dimension of the multi-dimensional feature data of the grouped nurse data, the patient data and the environmental data and extract key features to obtain the standardized data set.

[0030] Among them, nurse data can include nurses' skill levels, scheduling preferences, night shift acceptance flags, and historical work logs, and environmental data can include predicted increases in nursing demand during seasonal disease peaks.

[0031] Specifically, the K-means clustering algorithm can be used to group nurse data and classify them according to nurses' skill level, work experience, work intensity, etc., to ensure that nurses at all levels are reasonably matched during scheduling and achieve efficient collaboration of the nursing team.

[0032] Input data can include: Nurse skill level: such as junior nurse, intermediate nurse, senior nurse, etc., reflects the professional and technical level of nurses.

[0033] Work experience: The number of years a nurse has worked reflects his / her accumulation and experience in nursing work.

[0034] Work intensity: It is calculated through working hours, frequency of night shifts, number of patients nursed, etc., reflecting the current workload of nurses.

[0035] Specifically, K nurse data points can be randomly selected as the initial cluster centers. The K value can be selected based on actual needs and experience, for example, nurses can be divided into 3 groups (primary, intermediate, and senior) or more groups. The distance from each nurse data point to each cluster center is calculated, and the nurse is assigned to the group represented by the nearest cluster center. The distance is usually calculated using the Euclidean distance. Each cluster center is then recalculated, and the mean of all nurse data points in each group is used as the new cluster center. The allocation phase and the update phase are repeated until the cluster center no longer changes or the set number of iterations is reached.

[0036] The output is the grouped nurse data. The nurses are divided into different groups, and the nurses in each group have similarities in skill level, work experience, and work intensity. For example, nurses in the junior nurse group may have lower skill levels and less work experience, but relatively higher work intensity; nurses in the senior nurse group have higher skill levels, rich experience, and moderate work intensity.

[0037] The PCA dimensionality reduction algorithm reduces the dimensionality of the multi-dimensional feature data of grouped nurse data, patient data, and environmental data, extracts key features, reduces data dimensions, improves subsequent calculation efficiency, and retains the main information of the data.

[0038] Input data can include: Grouped nurse data: including nurses’ skill level, work experience, work intensity and other characteristics.

[0039] Patient data: such as fracture type, surgical method, postoperative care needs, real-time vital signs, etc.

[0040] Environmental data: such as national statutory holidays, hospital workday schedules, etc.

[0041] Specifically, the input data can be standardized first so that the mean of each feature is 0 and the variance is 1. Then the covariance matrix is ​​calculated, which reflects the correlation between the features. Then the eigenvalues ​​and eigenvectors of the covariance matrix are solved. The eigenvalues ​​represent the importance of each principal component, and the eigenvectors represent the direction of the principal component. Sort the eigenvalues ​​from large to small, and select the first k eigenvectors as the principal components. Finally, the original data is projected onto the direction of the principal component to obtain the reduced-dimensional data.

[0042] The output result can be a standardized data set, a data set after dimensionality reduction, which retains the main information of the original data, but reduces the dimension and reduces the computational complexity. For example, the original data may have dozens of features, but after PCA dimensionality reduction, only a few key features may remain, which can better reflect the internal structure and laws of the data.

[0043] In summary, the present invention groups nurse data through K-means clustering algorithm, realizes the reasonable classification of nurses, and provides a basis for subsequent scheduling. The PCA dimensionality reduction algorithm performs dimensionality reduction processing on the grouped nurse data, patient data, and environmental data, extracts key features, improves data processing efficiency, and provides a more concise and efficient data basis for subsequent machine learning model training and optimization.

[0044] Step 202: Process the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts.

[0045] In some embodiments, the time series prediction model is an LSTM network, the decision tree model is an XGBoost algorithm, and step 202 may include: The standardized data set is processed using the LSTM network to predict fluctuations in nursing needs of postoperative patients at night; Using the XGBoost algorithm to process the standardized data set to predict the nursing demand level for future shifts; The nursing demand for the future shift is determined according to the nursing demand fluctuation and the nursing demand level.

[0046] In the specific implementation, the LSTM network and XGBoost algorithm are used to predict the fluctuation of nursing needs of postoperative patients at night and the nursing demand level of future scheduling, respectively.

[0047] The fluctuation of postoperative nighttime nursing needs of patients is predicted through LSTM network, so that sufficient nursing resources can be arranged in advance. The input data is a standardized data set, which is processed by PCA dimensionality reduction and contains time series features such as the patient's historical nursing needs data, real-time vital signs, and postoperative nursing needs change trend.

[0048] LSTM (Long Short-Term Memory) is a special type of recurrent neural network (RNN) that is specifically designed to process time series data and can capture long-term dependencies. Its core is to control the flow of information through a series of gates (input gate, forget gate, and output gate), avoiding the gradient vanishing problem of traditional RNN.

[0049] In the data preparation stage, time series data (such as historical data on the nighttime care needs of postoperative patients) can be divided into training sets and test sets. Convert the data into the format required by LSTM, usually a three-dimensional array (number of samples, time steps, number of features). Build an LSTM network, including an input layer, an LSTM layer, a fully connected layer, and an output layer. Select a suitable activation function (such as ReLU or Sigmoid) and an optimizer (such as Adam). The training process uses the training set data to train the LSTM model, and updates the network parameters through back propagation to minimize the prediction error. During the training process, the validation set is used to adjust hyperparameters, such as learning rate, number of hidden units, time steps, etc. The trained LSTM model can be used to predict the test set data to obtain the predicted value of the nighttime care needs of postoperative patients. Evaluate the performance of the model, such as calculating indicators such as mean square error (MSE) and mean absolute error (MAE).

[0050] The output result is the prediction of nursing demand fluctuation. The LSTM model outputs the predicted value of the nighttime nursing demand of postoperative patients, reflecting the changing trend of nursing demand at different time points. For example, it predicts the number of nursing times and nursing duration required on a certain night.

[0051] The present invention predicts the nursing demand level of future scheduling through the XGBoost algorithm, and predicts the nursing demand level of future scheduling (such as primary, secondary, and tertiary care) through the XGBoost algorithm, so as to reasonably allocate nursing resources.

[0052] Specifically, the input data is a standardized data set, and the data set after PCA dimensionality reduction processing contains characteristics such as patient medical record data (such as fracture type, surgical method), postoperative care needs, and changes in the condition of hospitalized patients.

[0053] XGBoost (eXtreme Gradient Boosting) is an ensemble learning algorithm based on gradient boosting, which builds a strong learner by combining multiple weak learners (usually decision trees). It has efficient computing performance and good prediction effect.

[0054] In the data preparation phase, the data set can be divided into a training set and a test set. The data is preprocessed, such as feature encoding and standardization. The XGBoost model is built and hyperparameters are set, such as the tree depth, learning rate, and regularization parameters. The model is trained using the training set data, and the model performance is gradually optimized using the gradient boosting algorithm.

[0055] During the training process, XGBoost automatically handles missing values, feature selection, and other issues to improve the robustness of the model. Use methods such as cross-validation to evaluate model performance and adjust hyperparameters to obtain the best model. Then use the trained XGBoost model to predict the test set data to obtain the level of care needs for each patient. Evaluate model performance, such as calculating accuracy, recall, F1 score, and other indicators.

[0056] The output can be a prediction of the level of care needed, where the XGBoost model outputs the level of care needed for each patient (e.g., primary, secondary, or tertiary care). For example, it is predicted that a certain patient will need moderate care, while another patient will need severe care.

[0057] It can be understood that the nursing needs for future scheduling are determined based on the nursing demand fluctuations predicted by LSTM and the nursing demand level predicted by XGBoost, so as to reasonably arrange nursing resources.

[0058] The data integration stage can integrate the nighttime nursing demand fluctuations predicted by LSTM and the nursing demand level predicted by XGBoost. For example, for a certain night, LSTM predicts that 10 nursing operations are needed, and XGBoost predicts that 5 patients need secondary nursing and 3 patients need primary nursing. The total nursing demand can be calculated based on the nursing demand level and fluctuation. For example, suppose that tertiary nursing requires 1 nurse, secondary nursing requires 2 nurses, and primary nursing requires 3 nurses. If the number of nursing operations predicted by LSTM is 10, and the nursing demand level distribution predicted by XGBoost is: tertiary nursing: 2 times, secondary nursing: 5 times, primary nursing: 3 times, then the total nursing demand is: 2+10+9=21 (times).

[0059] Then, based on the calculated nursing needs, combined with the nurses’ skills, experience and workload, the optimal schedule can be generated through intelligent matching of nurse resources and flexible scheduling optimization. For example, according to the level and fluctuation of nursing needs, the combination of senior nurses and junior nurses can be reasonably arranged to ensure the efficient implementation of nursing work.

[0060] In summary, the present invention can accurately predict the fluctuation of nursing needs of postoperative patients at night and the nursing needs level of future shifts through the combination of LSTM network and XGBoost algorithm. It not only takes into account the changing trend of time series, but also combines the individual characteristics of patients and changes in their condition, providing a scientific basis for the rational arrangement of nursing resources. Finally, through intelligent matching and scheduling optimization, the efficiency, fairness and quality of nursing work are ensured.

[0061] Step 203: Generate a hierarchical nurse allocation plan through a recommendation algorithm based on the nurses' skills, current workload, and historical work data.

[0062] In some embodiments, step 203 may include: A collaborative filtering recommendation algorithm is used to match the fitness of each nurse with the scheduled shift based on historical scheduling data; The KNN algorithm was used to calculate the workload similarity among nurses; The hierarchical nurse allocation plan is generated according to the fitness and the workload similarity.

[0063] Specifically, the collaborative filtering recommendation algorithm and KNN algorithm are used to evaluate the adaptability of nurses to the scheduled shifts and the workload similarity between nurses, respectively, and then generate a hierarchical nurse allocation plan. The collaborative filtering recommendation algorithm is based on historical scheduling data to evaluate the adaptability of each nurse to the scheduled shift and predict the nurse's adaptability to a specific shift.

[0064] The input data is historical scheduling data, including the nurse's past scheduling records, nursing task completion status, patient feedback and other information. Collaborative filtering recommendation algorithms are mainly divided into two categories, user-based collaborative filtering and item-based collaborative filtering. In the present invention, a user-based collaborative filtering method can be used.

[0065] Specifically, the similarity between nurses can be calculated, usually using cosine similarity or Pearson correlation coefficient. For example, for nurse A and nurse B, the similarity of their performance in past shifts (such as nursing task completion, patient satisfaction, etc.) is calculated.

[0066] Fitness prediction can predict the current nurse's fitness for a certain shift based on the historical scheduling data of similar nurses. For example, if nurse A has cared for fracture surgery patients many times in the past and performed well, the system will think that nurse A has a high fitness for similar shifts.

[0067] The output result can be a fitness score, where each nurse has a fitness score for different shifts, reflecting the nurse's adaptability to a specific shift. For example, nurse A's fitness score for night shift is 0.8, indicating that she has a high adaptability to night shift.

[0068] The purpose of the KNN algorithm is to calculate the workload similarity between nurses and ensure the balance of scheduling.

[0069] The input data can be nurse workload data, including working hours, frequency of night shifts, number of nursing tasks, fatigue index, etc.

[0070] The KNN (K-Nearest Neighbors) algorithm finds the nearest neighbors by calculating the distance between data points. In this scheme, the Euclidean distance can be used to calculate the workload similarity between nurses.

[0071] First, the workload data of nurses can be standardized to make different features comparable. Then the Euclidean distance between each nurse and other nurses can be calculated. For example, for nurse A and nurse B, their distance in features such as working hours and frequency of night shifts can be calculated. Then, based on the distance calculation results, the workload similarity between nurses can be evaluated. The smaller the distance, the higher the similarity.

[0072] The output result can be a similarity matrix: a matrix that represents the workload similarity between each nurse and other nurses. For example, the workload similarity between nurse A and nurse B is 0.7, indicating that their workloads are relatively similar. The purpose of generating a hierarchical nurse allocation plan is to generate a reasonable hierarchical nurse allocation plan based on the adaptability of nurses to shifts and the workload similarity between nurses.

[0073] Specifically, each nurse can be comprehensively evaluated by combining the fitness score obtained by the collaborative filtering recommendation algorithm and the workload similarity obtained by the KNN algorithm. For example, for a certain shift, nurses with high fitness and low workload similarity are given priority. Then, hierarchical allocation is carried out according to the nurses' skill level and work experience. For example, senior nurses are assigned to shifts that require high skills, and junior nurses are assigned to auxiliary shifts. Ensure that the nurse combination of each shift is complementary in skills and experience, and the workload is balanced. Finally, based on the comprehensive evaluation results, a preliminary scheduling plan is generated. The scheduling plan is optimized and adjusted to ensure that it meets the constraints (such as daily working hours, night shift frequency, etc.).

[0074] The output result can be a hierarchical nurse allocation plan, such as a detailed shift schedule that clearly specifies the allocation of each nurse in different shifts to ensure the efficiency, fairness and quality of nursing work.

[0075] In summary, the present invention can generate a reasonable hierarchical nurse allocation scheme by combining the collaborative filtering recommendation algorithm and the KNN algorithm, taking into account the adaptability of nurses to shifts and the workload similarity between nurses. This scheme not only improves the job satisfaction of nurses, but also ensures the efficient use of nursing resources and the improvement of nursing quality.

[0076] In some embodiments, fitness is determined by: A first fitness function for maximizing the weighted sum of nurse satisfaction scores and nursing needs coverage is constructed; constructing a second fitness function for minimizing the product of the sum of the nurse fatigue index and the number of shift adjustments; The fitness is determined according to a function value of the first fitness function and a function value of the second fitness function.

[0077] Specifically, in order to optimize the shift scheduling, two fitness functions need to be constructed: the first fitness function and the second fitness function. These two functions are used to evaluate the performance of the shift scheduling scheme in terms of nurse satisfaction and nursing demand coverage, as well as the performance in terms of nurse fatigue index and the number of shift scheduling adjustments.

[0078] The first fitness function aims to construct a fitness function for maximizing the weighted sum of nurse satisfaction scores and nursing needs coverage.

[0079] The input data of the first fitness function can be the nurse satisfaction score: the nurse satisfaction score calculated based on the nurse's scheduling preference, working hours, night shift frequency, etc. And the nursing demand coverage rate: the nursing demand coverage rate calculated based on the scheduling plan reflects the degree to which the scheduling plan meets the nursing needs.

[0080] Specifically, the weights of the nurse satisfaction score and the nursing demand coverage rate can be set, and then the weighted sum of the nurse satisfaction score and the nursing demand coverage rate can be calculated as the value of the first fitness function. The output result can be the first fitness function value, which represents the comprehensive performance of the scheduling plan in terms of nurse satisfaction and nursing demand coverage rate.

[0081] The second fitness function can construct a fitness function for minimizing the product of the sum of the nurse fatigue index and the number of shift adjustments. The input data can be the nurse fatigue index: the nurse fatigue index calculated according to the nurse's working hours, night shift frequency, rest time, etc. and the number of shift adjustments: the number of shift adjustments calculated according to the shift plan, reflecting the stability of the shift plan.

[0082] Specifically, the product of the sum of the nurse fatigue index and the number of shift adjustments can be calculated as the value of the second fitness function. The output result can be the second fitness function value, which represents the comprehensive performance of the shift scheduling scheme in terms of the nurse fatigue index and the number of shift adjustments.

[0083] The purpose of determining the final fitness is to determine the final fitness of the scheduling plan based on the function values ​​of the first fitness function and the second fitness function. The first fitness function value and the second fitness function value can be comprehensively evaluated. Weighted sum, product or other combination methods can be used to select according to actual needs and preferences. Then, based on the comprehensive evaluation results, the final fitness of the scheduling plan is determined. For example, the weighted sum of the first fitness function value and the second fitness function value can be used as the final fitness. The output result can be the final fitness, which represents the comprehensive performance of the scheduling plan and is used to guide the optimization and selection of the scheduling plan.

[0084] In summary, the present invention can comprehensively evaluate the performance of the scheduling scheme in terms of nurse satisfaction, nursing demand coverage, nurse fatigue index and scheduling adjustment times by constructing the first fitness function and the second fitness function. The final fitness combines the values ​​of the two fitness functions and provides a scientific evaluation basis for the optimization of the scheduling scheme. By maximizing the final fitness, an optimal scheduling scheme that satisfies both nurse satisfaction and nursing demand coverage and minimizes nurse fatigue index and scheduling adjustment times can be obtained.

[0085] Step 204: Genetic algorithm is used in combination with dynamic constraints to generate a shift schedule that satisfies the hierarchical nurse allocation plan and the nursing needs.

[0086] Among them, the dynamic constraints of the genetic algorithm include daily working hours constraints, weekly shift days constraints and consecutive rest days constraints.

[0087] In the present invention, a genetic algorithm (GA) is used to generate a schedule that meets the hierarchical nurse allocation plan and nursing needs. A genetic algorithm is a search algorithm based on natural selection and genetics principles, which solves optimization problems by simulating the biological evolution process.

[0088] Specifically, a set of possible scheduling schemes can be randomly generated, each of which is called an "individual". Each individual is composed of multiple "genes", and each gene represents the scheduling information of a nurse in a certain shift.

[0089] The fitness function can be used to evaluate the quality of each individual. The fitness function gives a value based on the quality of the scheduling plan (such as nurse satisfaction, nursing demand coverage, fatigue index, etc.). The higher the fitness, the better the scheduling plan. Then, some individuals can be selected as the "parents" of the next generation based on the fitness. Roulette selection, tournament selection and other methods are usually used, and individuals with high fitness have a higher probability of being selected. Then part of the genes of the two parent individuals can be exchanged to generate new offspring individuals. For example, the shifts of some nurses in the two scheduling plans are exchanged to generate a new scheduling plan. Finally, the newly generated offspring individuals are randomly mutated to change the value of a gene with a certain probability. For example, randomly changing the shift of a nurse increases the diversity of the population and prevents the algorithm from falling into a local optimum.

[0090] In the process of generating the shift schedule, the following dynamic constraints need to be met: Daily working hours constraint: Each nurse’s daily working hours shall not exceed 8 hours. When generating the shift schedule, ensure that each nurse’s shift schedule does not exceed the prescribed daily working hours.

[0091] Constraints on the number of days per week: Each nurse is scheduled to work no more than 5 days per week. During the scheduling process, the number of working days per week for each nurse is counted to ensure that it does not exceed the prescribed upper limit.

[0092] Consecutive rest days constraint: Try to ensure that each nurse has at least 2 consecutive rest days per week. When scheduling, give priority to nurses' rest days to ensure that the continuous rest days requirement is met.

[0093] The process of generating a shift schedule by genetic algorithm first randomly generates a set of initial shift plans that meet dynamic constraints. Each plan is a possible shift schedule, which contains the schedule information of all nurses in different shifts. Then, the fitness function is used to evaluate the pros and cons of each shift plan. The fitness function comprehensively considers factors such as nurse satisfaction, nursing demand coverage, and nurse fatigue index. Then, according to the fitness, some shift plans are selected as the "parents" of the next generation. The roulette selection method can be used, and the shift plan with high fitness has a higher probability of being selected. Then, some genes of the two parent shift plans are exchanged to generate a new child shift plan. For example, the shift arrangements of some nurses in the two shift plans are exchanged. Then, the newly generated child shift plan is randomly mutated to change the shift of a certain nurse with a certain probability. This step helps to increase the diversity of the population and prevent the algorithm from falling into a local optimum. Then check whether the newly generated shift plan meets the dynamic constraints (daily working hours, weekly shift days, and consecutive rest days). If not, adjust or regenerate. Repeat the above steps until the preset number of iterations is reached or the fitness is no longer significantly improved. In each iteration, the population gradually evolves to a better scheduling scheme. Finally, the scheduling scheme with the highest fitness is selected as the final schedule. While satisfying the dynamic constraints, this scheme maximizes nurse satisfaction and nursing demand coverage, and minimizes the nurse fatigue index.

[0094] In practical applications, nursing needs may change over time, so the schedule needs to be updated regularly. Genetic algorithms can dynamically adjust the schedule based on the latest nursing demand forecasts (such as the output of the LSTM model) and nurse data (such as fatigue index, work preferences, etc.). For example, if the nursing demand suddenly increases on a certain day, the algorithm can re-optimize the schedule to ensure that the demand is met while minimizing the impact on the nurse's work schedule.

[0095] By combining genetic algorithms with dynamic constraints, the present invention can generate a schedule that meets the hierarchical nurse allocation plan and nursing needs. The genetic algorithm gradually optimizes the scheduling plan by simulating natural selection and genetic processes, ensuring that the comprehensive benefits of scheduling are maximized while meeting the constraints. This not only improves the scientificity and rationality of scheduling, but also enhances the flexibility and adaptability of nursing resources.

[0096] Step 205: Based on the shift data and physiological indicators of the shift schedule, dynamically adjust the workload of each nurse through a supervised learning model to generate a first shift scheduling strategy.

[0097] In some embodiments, the physiological indicator includes heart rate variability data, and the supervised learning model is a random forest classifier. Step 205 may include: Calculating the fatigue index of the nurses by using the random forest classifier in combination with the shift scheduling data and the heart rate variability data of the nurses; The workload of each nurse is dynamically adjusted according to the fatigue index of the nurse to generate the first scheduling strategy.

[0098] In the present application, a random forest classifier is used to combine the scheduling data and the nurses' heart rate variability data to calculate the nurses' fatigue index, and dynamically adjust the workload of each nurse according to the fatigue index to generate a first scheduling strategy.

[0099] Random forest classifier calculates nurses’ fatigue index The purpose is to accurately assess nurses’ fatigue level by combining the scheduling data and nurses’ heart rate variability data through random forest classifier.

[0100] The input data can be shift data: including nurses’ working hours, night shift frequency, consecutive working days, rest days, etc. And heart rate variability data: nurses’ heart rate variability data collected through devices such as smart bracelets reflects the nurses’ physiological fatigue status.

[0101] Random Forest is an ensemble learning algorithm that improves the accuracy of classification or regression by building multiple decision trees. In this scenario, Random Forest is used to classify the fatigue status of nurses.

[0102] In the data preparation stage, the shift data and heart rate variability data can be integrated into a feature vector. For example, the feature vector may include: working hours, night shift frequency, consecutive working days, heart rate variability indicators (such as SDNN, RMSSD, etc.).

[0103] During the model training phase, historical data (nurse data with known fatigue status) can be used to train the random forest classifier. The training data should contain the above features and corresponding fatigue labels (such as low fatigue, moderate fatigue, and high fatigue).

[0104] In the fatigue index calculation stage, the trained random forest classifier can be used to predict the fatigue status of each nurse. Based on the prediction results, the fatigue index is calculated. For example, the fatigue status can be mapped to a numerical value: low fatigue: 1, medium fatigue: 2, high fatigue: 3. The output result can be a fatigue index, and the fatigue index of each nurse reflects the current fatigue level.

[0105] The workload of each nurse is dynamically adjusted according to the fatigue index. The purpose is to ensure that the nurse's work arrangement is reasonable and avoid excessive fatigue.

[0106] Workload assessment is to evaluate the current workload of each nurse based on the fatigue index. For example, nurses with a high fatigue index should reduce their workload, while nurses with a low fatigue index can increase their workload appropriately. The dynamic adjustment strategy is to adjust the nurses' shift schedule based on the fatigue index. Adjustment strategies may include: reducing the number of night shifts for high-fatigue nurses, increasing the working hours for low-fatigue nurses, arranging rest days for high-fatigue nurses, and optimizing the shift distribution of nurses to ensure a balanced workload.

[0107] Finally, the first scheduling strategy can be generated based on the adjusted nurse workload. The strategy should meet the following conditions: meet dynamic constraints (such as daily working hours, weekly scheduling days, consecutive rest days), maximize nurse satisfaction, and ensure coverage of nursing needs.

[0108] The output result can be the first scheduling strategy, which is a detailed schedule that clearly specifies the allocation of each nurse in different shifts, while taking into account the nurses' fatigue index to ensure a reasonable workload.

[0109] For example, assume the following data and model: Input data is shift scheduling data: Nurse A: working hours = 8 hours, night shift frequency = 2 times / week, consecutive working days = 5 days; Nurse B: Working hours = 7 hours, night shift frequency = 1 time / week, continuous working days = 4 days.

[0110] Heart rate variability data: Nurse A: Heart rate variability index SDNN=50ms, RMSSD=30ms; Nurse B: Heart rate variability index SDNN=60ms, RMSSD=40ms.

[0111] Random forest classifier, the trained random forest model predicts fatigue status based on input features: Nurse A: Fatigue status = high fatigue (fatigue index = 3); Nurse B: Fatigue status = low fatigue (fatigue index = 1).

[0112] Dynamically adjust strategies to adjust nurses’ workload based on fatigue index: Nurse A: Reduce the number of night shifts and arrange 1 day of rest; Nurse B: Increase working hours appropriately and arrange one night shift.

[0113] The first scheduling strategy generates the following schedule: Nurse A: working hours = 7 hours, night shift frequency = 1 time / week, consecutive working days = 4 days, rest days = 2 days; Nurse B: Working hours = 8 hours, night shift frequency = 2 times / week, consecutive working days = 5 days, rest days = 2 days.

[0114] In some embodiments, the method of the present invention can also automatically identify statutory holidays, nurses' birthdays, and nurses' scheduled vacation time. Specifically, by automatically identifying statutory holidays, nurses' birthdays, and nurses' scheduled vacation time, the scheduling plan can be further optimized to ensure that while meeting nursing needs, the personal time of nurses is fully respected, thereby improving the job satisfaction and quality of life of nurses.

[0115] In the specific implementation, the national statutory holiday information, including the date and type of the holiday (such as Spring Festival, National Day, etc.), can be obtained from the hospital information system or external calendar service. The birthday information of nurses can be extracted from the hospital's human resources management system to ensure that the system can recognize the birthday of each nurse. Nurses are allowed to submit leave reservation applications through the hospital's internal system, and the system automatically records and manages these leave reservations.

[0116] In some embodiments, statutory holidays, nurses' birthdays, and scheduled vacation time can be integrated into the data model of the scheduling system to ensure that these special dates can be considered in real time when generating the schedule. For statutory holidays, the system automatically marks these dates and makes special treatments based on the hospital's scheduling policies (such as statutory holiday work arrangements, overtime compensation, etc.). For nurses' birthdays, the system gives priority to giving nurses rest or arranging easier shifts when scheduling, reflecting care for nurses. For scheduled vacation time, the system automatically avoids these dates when generating the schedule to ensure that nurses can take their scheduled vacations.

[0117] When generating a shift schedule, the system can automatically check each nurse's statutory holidays, birthdays, and scheduled vacations to ensure that the schedule meets these constraints. If a nurse has a shift requirement on a statutory holiday or birthday, the system will prioritize other nurses or adjust the schedule to ensure that the nurse can get proper rest on these special days. For scheduled vacations, the system will adjust the schedule in advance to ensure that nursing work can proceed smoothly during the nurse's vacation, while avoiding confusion caused by temporary adjustments to the schedule.

[0118] By automatically identifying and processing these special dates, the system not only improves the scientificity and rationality of scheduling, but also shows respect and care for nurses' personal time, which helps to improve nurses' job satisfaction and loyalty. The system can generate reports regularly to remind managers to pay attention to nurses' statutory holidays, birthdays, and scheduled vacations, further optimizing human resource management.

[0119] In summary, by combining the random forest classifier with the scheduling data and heart rate variability data, the present invention can accurately evaluate the fatigue index of nurses, dynamically adjust the workload of each nurse according to the fatigue index, and generate a reasonable first scheduling strategy. It not only takes into account the physiological and psychological state of nurses, but also ensures the efficiency and quality of nursing work, and improves the job satisfaction of nurses and the nursing experience of patients.

[0120] Step 206: Automatically match substitute nurses in emergency situations and optimize the first scheduling strategy through a reinforcement learning model to generate a second scheduling strategy.

[0121] In some embodiments, the reinforcement learning model is a Deep Q-Learning model, and step 206 may include: Obtaining a linear combination of the skill matching degree, fatigue index and scheduling conflict rate of each substitute nurse; The first scheduling strategy is optimized according to a linear combination of the skill matching degree, fatigue index and scheduling time conflict rate of each substitute nurse, with the optimization goal of minimizing the decline in nursing quality and the additional workload of nurses in the current scheduling, thereby generating the second scheduling strategy.

[0122] Specifically, in order to optimize the first scheduling strategy and generate the second scheduling strategy, it is necessary to comprehensively consider the skill matching degree, fatigue index and scheduling time conflict rate of substitute nurses.

[0123] The purpose of obtaining the linear combination is to evaluate the comprehensive adaptability of each nurse as a substitute by calculating the linear combination of each substitute nurse's skill matching, fatigue index and scheduling conflict rate.

[0124] Input data can be skill matching: a measure of how well the substitute nurse's skills match the current nursing needs. For example, orthopedic nurses have a high matching degree of nursing skills for patients after fracture surgery. And fatigue index: calculated by combining the random forest classifier with the scheduling data and heart rate variability data, reflecting the current fatigue level of the nurse. And scheduling conflict rate: a measure of how well the substitute nurse's scheduling time conflicts with the current scheduling needs. For example, if the substitute nurse already has a scheduled shift, the conflict rate will increase.

[0125] First, the weights of skill matching, fatigue index, and scheduling conflict rate can be set. For each substitute nurse, a linear combination of its skill matching, fatigue index, and scheduling conflict rate can be calculated. The output result can be a comprehensive adaptability score, and the comprehensive adaptability score of each substitute nurse reflects its comprehensive adaptability as a substitute.

[0126] The purpose of optimizing the first shift scheduling strategy is to optimize the first shift scheduling strategy based on the comprehensive adaptability score, and minimize the decline in nursing quality and the additional workload of nurses in the current shift. The optimization goals can include minimizing the decline in nursing quality: ensuring that the skills of substitute nurses are highly matched with nursing needs, and reducing the decline in nursing quality caused by skill mismatch. And minimizing the additional workload of nurses: avoiding scheduling nurses who are overly tired or have scheduling conflicts, and reducing the additional workload of nurses.

[0127] First, the substitute nurses can be sorted according to the comprehensive adaptability score, and the nurses with high comprehensive adaptability scores can be selected as substitutes. For each shift that needs a substitute, the nurse with the highest comprehensive adaptability is selected as a substitute, while ensuring that the scheduling time does not conflict. If the comprehensive adaptability scores of multiple nurses are similar, other factors (such as nurses' work preferences, consecutive working days, etc.) can be further considered for fine-tuning. Then, the adjusted scheduling plan can be evaluated to ensure that dynamic constraints (such as daily working hours, weekly scheduling days, and consecutive rest days) are met. If the substitute selection for a certain shift leads to a decline in nursing quality or an increase in the additional workload of nurses, the selection is readjusted to find a suboptimal solution. The output result can be the second scheduling strategy, that is, the optimized scheduling table, which clarifies the allocation of each nurse in different shifts, while minimizing the decline in nursing quality and the additional workload of nurses.

[0128] For example, assume the following data and model: Enter data for substitute nurse information: Nurse A: Skill matching degree = 0.9, fatigue index = 2, scheduling conflict rate = 0.1; Nurse B: skill matching degree = 0.7, fatigue index = 1, scheduling conflict rate = 0.2; Nurse C: Skill matching degree = 0.8, fatigue index = 3, scheduling conflict rate = 0.1.

[0129] Weight settings: w1=0.5, w2=0.3, w3=0.2.

[0130] Linear combination calculation: Nurse A's comprehensive adaptability: 0.5×0.9−0.3×2−0.2×0.1=0.45−0.6−0.02=−0.17; Nurse B's comprehensive adaptability: 0.5×0.7−0.3×1−0.2×0.2=0.35−0.3−0.04=0.01; Nurse C's comprehensive adaptability: 0.5×0.8−0.3×3−0.2×0.1=0.4−0.9−0.02=−0.52.

[0131] Optimize the scheduling results: Sort by comprehensive adaptability score: Nurse B (0.01); Nurse A (-0.17); Nurse C (-0.52).

[0132] Nurse B was selected as the substitute because she had the highest overall adaptability score.

[0133] Second shift scheduling strategy: The generated shift schedule is as follows: Nurse A: Keep the original schedule; Nurse B: Replaces on shifts that need to be adjusted; Nurse C: Keep the original schedule.

[0134] By calculating the linear combination of skill matching, fatigue index and scheduling time conflict rate of each substitute nurse, the present invention can comprehensively evaluate the adaptability of each nurse as a substitute. The first scheduling strategy is optimized according to the comprehensive adaptability score, and the second scheduling strategy is generated to ensure that while meeting the nursing needs, the decline in nursing quality and the additional workload of nurses are minimized. This not only improves the scientificity and rationality of scheduling, but also enhances the flexibility and adaptability of nursing resources.

[0135] See also Figure 3 , Figure 3 A schematic diagram of the structure of a clinical intelligent scheduling system based on big data analysis provided by the present invention.

[0136] like Figure 3 As shown, a clinical intelligent scheduling system based on big data analysis proposed in an embodiment of the present invention includes: The data acquisition module 301 is used to acquire a data set of patient data, nurse data and environmental data, and pre-process the data set to generate a standardized data set; A demand prediction module 302 is used to process the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; A hierarchical allocation module 303 is used to generate a hierarchical nurse allocation plan through a recommendation algorithm based on the nurses' skills, current workload and historical work data; A preliminary scheduling module 304 is used to generate a schedule that meets the hierarchical nurse allocation plan and the nursing needs by using a genetic algorithm combined with dynamic constraints; A first scheduling module 305, for dynamically adjusting the workload of each nurse through a supervised learning model based on the scheduling data and physiological indicators of the scheduling table, and generating a first scheduling strategy; The second scheduling module 306 is used to automatically match substitute nurses in emergency situations and optimize the first scheduling strategy through a reinforcement learning model to generate a second scheduling strategy.

[0137] See also Figure 4 , Figure 4 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention provides an electronic device 400, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, the following steps are implemented: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set; Processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; Generate a hierarchical nurse allocation plan through a recommendation algorithm based on nurses' skills, current workload, and historical work data; Genetic algorithms are used in combination with dynamic constraints to generate a shift schedule that meets the hierarchical nurse allocation plan and the nursing needs; Based on the scheduling data and physiological indicators of the schedule, dynamically adjusting the workload of each nurse through a supervised learning model to generate a first scheduling strategy; Through the reinforcement learning model, substitute nurses are automatically matched in emergency situations and the first scheduling strategy is optimized to generate a second scheduling strategy.

[0138] See also Figure 5 , Figure 5 A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 5 As shown, this embodiment provides a computer-readable storage medium 500, on which a computer program 411 is stored. When the computer program 411 is executed by a processor, the following steps are implemented: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set; Processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; Generate a hierarchical nurse allocation plan through a recommendation algorithm based on nurses' skills, current workload, and historical work data; Genetic algorithms are used in combination with dynamic constraints to generate a shift schedule that meets the hierarchical nurse allocation plan and the nursing needs; Based on the scheduling data and physiological indicators of the schedule, dynamically adjusting the workload of each nurse through a supervised learning model to generate a first scheduling strategy; Through the reinforcement learning model, substitute nurses are automatically matched in emergency situations and the first scheduling strategy is optimized to generate a second scheduling strategy.

[0139] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0140] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0141] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or multiple boxes.

[0142] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0144] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0145] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A clinical intelligent scheduling method based on big data analysis, characterized in that: The method comprises: Acquire a data set of patient data, nurse data, and environmental data, and preprocess the data set to generate a standardized data set; Processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts; Generate a hierarchical nurse allocation plan based on nurses' skills, current workload and historical work data through a recommendation algorithm; Genetic algorithms are used in combination with dynamic constraints to generate a schedule that satisfies the hierarchical nurse allocation plan and the nursing needs; Based on the scheduling data and physiological indicators of the schedule, dynamically adjusting the workload of each nurse through a supervised learning model to generate a first scheduling strategy; Through the reinforcement learning model, substitute nurses are automatically matched in emergency situations and the first scheduling strategy is optimized to generate a second scheduling strategy.

2. The clinical intelligent scheduling method based on big data analysis according to claim 1 is characterized in that: The preprocessing of the data set to generate a standardized data set includes: The nurse data are processed using a K-means clustering algorithm, and the nurses are grouped according to their skill level and work intensity to obtain grouped nurse data; The PCA dimension reduction algorithm is used to reduce the dimension of the multi-dimensional feature data of the grouped nurse data, the patient data and the environmental data and extract key features to obtain the standardized data set.

3. The clinical intelligent scheduling method based on big data analysis according to claim 2 is characterized in that: The time series prediction model is an LSTM network, and the decision tree model is an XGBoost algorithm; The method of processing the standardized data set using a time series prediction model and a decision tree model to predict nursing needs for future shifts includes: The standardized data set is processed using the LSTM network to predict fluctuations in nursing needs of postoperative patients at night; Using the XGBoost algorithm to process the standardized data set to predict the nursing demand level for future shifts; The nursing demand for the future shift is determined according to the nursing demand fluctuation and the nursing demand level.

4. The clinical intelligent scheduling method based on big data analysis according to claim 3 is characterized in that: The method generates a hierarchical nurse allocation plan based on the nurse's skills, current workload and historical work data through a recommendation algorithm, including: Collaborative filtering recommendation algorithm, matching each nurse’s fitness with the scheduled shift based on historical scheduling data; KNN algorithm,calculates the workload similarity among nurses; The hierarchical nurse allocation plan is generated according to the fitness and the workload similarity.

5. The clinical intelligent scheduling method based on big data analysis according to claim 4 is characterized in that: The dynamic constraints of the genetic algorithm include daily working hours constraints, weekly shift days constraints and consecutive rest days constraints.

6. The clinical intelligent scheduling method based on big data analysis according to claim 5 is characterized in that: The matching of each nurse's adaptability to the scheduled shift based on historical scheduling data includes: A first fitness function for maximizing the weighted sum of nurse satisfaction scores and nursing needs coverage is constructed; constructing a second fitness function for minimizing the product of the sum of the nurse fatigue index and the number of shift adjustments; The fitness is determined according to a function value of the first fitness function and a function value of the second fitness function.

7. The clinical intelligent scheduling method based on big data analysis according to claim 6 is characterized in that: The physiological index includes heart rate variability data, and the supervised learning model is a random forest classifier; The scheduling data and physiological indicators based on the schedule dynamically adjust the workload of each nurse through a supervised learning model to generate a first scheduling strategy, including: Calculating the fatigue index of the nurses by using the random forest classifier in combination with the shift scheduling data and the heart rate variability data of the nurses; The workload of each nurse is dynamically adjusted according to the fatigue index of the nurse to generate the first scheduling strategy.

8. The clinical intelligent scheduling method based on big data analysis according to claim 7 is characterized in that: The reinforcement learning model is a Deep Q-Learning model; the method of automatically matching substitute nurses in emergencies and optimizing the first scheduling strategy by the reinforcement learning model to generate a second scheduling strategy includes: Obtaining a linear combination of the skill matching degree, fatigue index and scheduling conflict rate of each substitute nurse; The first scheduling strategy is optimized according to a linear combination of the skill matching degree, fatigue index and scheduling time conflict rate of each substitute nurse, with the optimization goal of minimizing the decline in nursing quality and the additional workload of nurses in the current scheduling, thereby generating the second scheduling strategy.

9. The clinical intelligent scheduling method based on big data analysis according to claim 8 is characterized in that: The nurse data includes the nurse's skill level, scheduling preference, night shift acceptance flag and historical work log, and the environmental data includes the predicted increase in nursing demand during seasonal disease peaks.

10. A clinical intelligent scheduling system based on big data analysis, characterized in that: The system comprises: A data acquisition module is used to acquire a data set of patient data, nurse data and environmental data, and preprocess the data set to generate a standardized data set; A demand forecasting module, used to process the standardized data set using a time series forecasting model and a decision tree model to predict nursing needs for future shifts; A hierarchical allocation module is used to generate a hierarchical nurse allocation plan through a recommendation algorithm based on the nurses' skills, current workload and historical work data; A preliminary scheduling module, for generating a schedule that meets the hierarchical nurse allocation scheme and the nursing needs by using a genetic algorithm combined with dynamic constraints; A first scheduling module, for dynamically adjusting the workload of each nurse through a supervised learning model based on the scheduling data and physiological indicators of the scheduling table, and generating a first scheduling strategy; The second scheduling module is used to automatically match substitute nurses in emergency situations and optimize the first scheduling strategy through a reinforcement learning model to generate a second scheduling strategy.

Images