Machine Learning-Based Method for Allocating Hospitalization Resources under the Weekend Effect

Abstract

The present invention provides a method for allocating hospital resources under the weekend effect based on machine learning. A prediction framework for the discharge mode of inpatients under the weekend effect is constructed using machine learning methods, so as to complete the allocation of hospital resources. First, the random forest algorithm is used to identify the key factors affecting the discharge mode; the Bayesian causal network and causal forest model are used to explore the causal relationships between the key factors; then, the vector autoregressive model is used to analyze time series data to capture the dynamic relationships between variables, and the dynamic time warping technology is applied to measure the similarity of the admission and surgical processes of different patients, generating time series similarity features; the key features, causal relationship features, and time series features are combined and input into the long short-term memory network model to output the discharge mode of patients. The long short-term memory network model can process complex time series data, improve the accuracy of prediction, and provide a basis for improving the allocation of hospital resources.

Description

Technical Field

[0001] The present invention relates to machine learning technology, and particularly to inpatient resource allocation technology based on machine learning. Background Art

[0002] The weekend effect is caused by differences in the severity of patients' conditions, treatment plans, and medical resource allocation during non-working hours. Currently, domestic and foreign medical system policies lead to the widespread existence of the weekend effect. However, there are still many unexplored aspects in existing research, especially the specific impact of the weekend effect in specific diseases, patient groups, or medical service fields. Although traditional epidemiological statistical analysis methods and meta-analyses have been used to study the weekend effect, they still have deficiencies in identifying the causes and have low utilization efficiency of medical resources. With the in-depth understanding of the weekend effect and its potential causes, traditional statistical analysis methods may not be able to fully capture the dynamic changes and causal relationships behind this complex phenomenon. The weekend effect involves multiple complex factors, including changes in patients' conditions, hospital resource allocation, and the working status of medical staff. These factors show significant heterogeneity in time and space. To more comprehensively understand and predict the impact of the weekend effect on inpatient resource allocation, more advanced analysis methods are needed.

[0003] Machine learning and deep learning technologies have shown great potential in the field of medical data analysis. These technologies can not only process a large amount of complex medical data but also automatically capture potential patterns and causal relationships in the data through model training. However, existing research mainly focuses on the impact of specific diseases or single factors, lacking a systematic study of the comprehensive problem of the weekend effect. By using machine learning technology and combining causal inference and time series analysis methods, the key factors affecting the clinical outcomes of inpatients and their interaction relationships can be more accurately identified, thus providing more accurate predictions and more guiding inpatient resource allocation support. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for improving inpatient resource allocation by predicting the way patients leave the hospital, aiming at the problem that the impact of the weekend effect on the clinical outcomes of inpatients has not been clarified.

[0005] The technical solution adopted by the present invention to solve the above technical problem is an inpatient resource allocation method under the weekend effect based on machine learning, including the steps:

[0006] Data collection and preprocessing step: Collect inpatient data from the medical service database in chronological order, and mark the date variables in the inpatient data as working days and non-working days; the inpatient data includes patients' personal information, data on the admission period, data on the surgery period, and data on the discharge period.

[0007] Key feature selection step: Using the random forest algorithm, evaluate the importance of each variable in the inpatient data by constructing decision trees for the patients in the inpatient data, and identify the key variables affecting the discharge method as key features;

[0008] Causal relationship feature selection step: Based on the selected key feature - discharge method, construct a Bayesian causal network to explore the causal relationship between the key features including the weekend effect and the discharge method, and use the causal forest model based on decision trees for the non - linear relationship and interaction between key features to obtain the causal relationship features for predicting the discharge method of inpatients under the weekend effect;

[0009] Time - series feature generation step: Use the time - series data composed of the admission time data, surgery time data, and discharge time data in the inpatient data to train a vector autoregressive model. The vector autoregressive model learns to capture the dynamic change relationships of each variable at each time point during the admission time, surgery time, and discharge time. Apply the dynamic time warping technique to measure the similarity of the admission and surgery processes of different patients. The trained vector autoregressive model is used to receive the input time - series data of the admission time and surgery time of inpatients, calculate the similarity degree of each inpatient in the time series, and output a set of time - series features of admission - surgery - discharge method corresponding to each patient;

[0010] Discharge method prediction step: Combine the key features, causal relationship features, and time - series features, and pre - process the combined features. Then input the pre - processed features into a long short - term memory network. The long short - term memory network outputs the time - series features of each time step to a fully - connected layer; The fully - connected layer outputs the predicted discharge method for each time step based on the input time - series features;

[0011] Inpatient resource allocation step: Allocate inpatient resources according to the predicted discharge method.

[0012] Through the comprehensive application of the random forest algorithm, Bayesian causal network, causal forest, vector autoregressive VAR model, dynamic time warping DTW technology, and long short - term memory network LSTM model, the prediction scheme for the discharge method constructed by the present invention performs excellently in aspects such as feature selection, causal relationship analysis, and time - series feature extraction, significantly improving the prediction accuracy, precision, recall rate, and F1 score, providing a reliable basis for subsequent inpatient resource allocation.

[0013] The beneficial effects of the present invention are as follows: it shows excellent effectiveness in predicting the discharge mode of patients, can effectively extract and identify key features from a large amount of clinical data. It not only has a more accurate ability to predict the discharge mode, but also can quickly adapt to new data situations. It maintains high accuracy when dealing with unknown data in the verification stage, demonstrating good adaptability and generalization ability. On diverse inpatient data, the present invention can maintain accurate and stable prediction ability, effectively improving the efficiency and quality of medical services. Especially in the case of dealing with the weekend effect, it reduces the risk of predicting the discharge mode of patients, supports more scientific clinical decision-making and medical management, reduces waste of medical resources, and lowers costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flowchart for predicting the discharge mode of inpatients under the weekend effect constructed by a machine learning method. DETAILED DESCRIPTION OF THE INVENTION

[0015] Based on the admission date of inpatients, with the surgery date as the intermediate time node and the discharge mode as the end point, the present invention constructs the clinical path for each patient, that is, constructs the specific medical diagnosis and treatment items that the patient needs to receive every day. To implement the method of the present invention, a novel model for predicting the discharge mode of patients is proposed. This model comprehensively utilizes the random forest algorithm, Bayesian causal network, causal forest, vector autoregressive model, dynamic time warping technology, and long short-term memory network model, and specifically includes the following steps:

[0016] Data collection and preprocessing step: Collect the clinical data of inpatients from the medical service database in chronological order, including relevant data such as the admission period, surgery period, and discharge mode, mark the working days and non-working days for the date data, and the non-working days include weekends and holidays, providing weekend effect time features for subsequent analysis.

[0017] Feature selection and key factor identification step: Use the random forest algorithm to evaluate the importance of each variable in the inpatient data by constructing multiple decision trees, and identify and select the key variables affecting the discharge mode as features by analyzing various variables of the patient, such as the admission date, surgery date, admission method, etc.

[0018] Causality analysis steps: Based on the selected key feature - the way of leaving the hospital, construct a Bayesian causal network to explore the causal relationship between the features including the impact of the weekend effect on the way of leaving the hospital. Use a causal forest model based on decision trees to mine the non-linear relationships and interactions between these features, and determine the features that have an important impact on the way of leaving the hospital under the weekend effect, such as the admission date, admission route, and surgery date. Combine the causal relationship features that can accurately predict the way of leaving the hospital for inpatients under the weekend effect. The features affected by the weekend effect can be variables such as the admission date, admission route, admission type, admission situation, case classification, surgery date, and elective surgery.

[0019] Time series feature steps: Use a vector autoregressive model to train the multi-dimensional time series data of inpatients. The surgery period is used as an intermediate variable between the admission period and the way of leaving the hospital to capture the dynamic relationships of each variable at each time point in the admission period, surgery period, and discharge period. Through multiple regression analysis, predict the changes of each variable. Apply dynamic time warping technology to measure the time series similarity of different patients during the admission and surgery periods, and generate a set of similarity features for the admission-surgery time series for each patient, reflecting the similarity degree of different patients in the time series. Integrate the time series dynamic change features generated by the vector autoregressive model and the time series similarity features generated by the dynamic time warping technology to form the final time series features.

[0020] Steps for predicting the way of leaving the hospital: First, merge the key variable features, causal relationship features, and time series features to ensure that all feature dimensions are in the same format and complete. Then, align the data in each feature according to the admission time point, surgery time point, and discharge time point to ensure that the data at all time points are correctly matched. Next, perform standardization processing on the feature data to unify the dimension of the features and ensure that the data can be compared and analyzed on the same scale. After that, input the processed feature data into a long short-term memory network, which can handle complex time series data and capture the dependencies between time steps. The long short-term memory network outputs the time series features of each time step to the fully connected layer for further processing. The fully connected layer predicts the way of leaving the hospital at each time step based on the time series features output by the long short-term memory network. Finally, obtain a more accurate prediction result for the way of leaving the hospital of inpatients.

[0021] Steps for inpatient resource allocation: Allocate inpatient resources according to the predicted way of leaving the hospital.

[0022] Embodiment

[0023] The allocation of inpatient resources is based on reliable prediction results of the way of leaving the hospital. Take the construction of a prediction model for the way of leaving the hospital of inpatients with ischemic heart disease under the weekend effect in the ICD range I20 - I25 of diagnostic codes as an example. The specific implementation steps are as Figure 1 described.

[0024] S1. Data collection and preprocessing steps:

[0025] First, obtain the electronic medical records of inpatients with ischemic heart disease from the medical database, and collect the clinical data of inpatients from the patients' electronic medical records according to the date. In the preprocessing stage, data screening was carried out, and only the data directly related to three time points were selected. The data at the admission stage were collected, including the patient's admission date, admission route, admission type, whether it is ST-segment elevation myocardial infarction (STEMI), admission diagnosis, admission situation, case classification, as well as the patient's personal information such as age, gender, marital status, etc.; the data at the surgery stage were collected, including the surgery date, elective surgery, surgery level, and surgery-related information; the data at the discharge stage were collected, including the discharge date and discharge method. To improve the data processing efficiency and optimize the data structure of inpatients, including data merging and deduplication, the working days and non-working days (including weekends and holidays) in the date data were marked to consider the weekend effect, and other data were also processed with integer value labeling to simplify the dataset structure and improve the efficiency of subsequent analysis. After data screening and structure optimization, the data of inpatients for analyzing the weekend effect were obtained.

[0026] S2. Feature selection and key factor identification steps:

[0027] First, perform feature engineering and conduct preliminary feature selection on all variables that may affect the discharge method, including the patient's basic information, which includes age, gender, marital status, etc., admission information, and surgery information. Then, divide the dataset into a training set and a test set, with 80% for training and 20% for testing; ensure that the distribution of various discharge methods in the training set and the test set is roughly the same to avoid data bias. Use the training set data to build a random forest model and set the parameters of the random forest model, such as the number of decision trees, maximum depth, etc. After model training, use the test set data to evaluate the performance of the model. Commonly used evaluation indicators include accuracy, precision, recall, and F1 value.

[0028] The variable importance assessment is carried out through the feature importance attribute in the random forest model. Evaluate the importance of each variable for predicting the discharge method, and arrange the importance of the variables in descending order to identify the variables that have the greatest impact on the prediction results. According to the variable importance assessment results, select several of the most important variables as the final features. A threshold can be set to only select the variables with importance scores exceeding this threshold. Retrain the random forest model with the selected key features and perform parameter adjustment, such as using grid search to optimize the hyperparameters. Repeatedly train and validate to ensure the stability and prediction performance of the model. Through cross-validation, such as k-fold cross-validation, further verify the performance of the model to ensure its consistent performance on different data subsets.

[0029] S3. Causality analysis steps:

[0030] Based on the key features selected in the feature selection and key factor identification steps, construct the initial structure of the Bayesian causal network. Set the nodes representing features and the edges representing causal relationships, and preliminarily determine the possible causal relationships. Then use the data of the key features for training. Through methods such as maximum likelihood estimation or Bayesian estimation, learn the parameters in the network, and through structure learning algorithms, such as the greedy algorithm, optimize the network structure to ensure that the network can best represent the causal relationships in the data. Use the cross-validation method to evaluate the performance of the Bayesian causal network to ensure that it accurately reflects the causal probability of features on the discharge method.

[0031] After the Bayesian causal network determines the causal relationships, use the causal forest model to further explore the non-linear relationships and interactions in these causal relationships. According to the causal relationship features determined by the Bayesian causal network, construct the causal forest model. The causal forest model uses a decision-tree-based method to deeply explore the complex relationships and interactions between features. When training the causal forest model, input the key features and the discharge method as the target variables into the model, and the model will learn the non-linear relationships between features. The output of the causal forest model is the specific causal effect estimation of each feature on the discharge method, including the average causal effect and the conditional causal effect, and reveals the non-linear relationships and interactions between features. Pay special attention to the features containing the weekend effect, such as the impact of the admission date, admission route, and surgery date on the discharge method, and evaluate the effect differences of these features on weekdays or non-weekdays.

[0032] The admission date and the surgery date belong to the features containing the weekend effect because they contain the markers of date data. The admission route usually refers to the way patients enter the hospital, such as through the emergency department, outpatient clinic, etc. During weekends, regular outpatient services may not be available, and more patients may enter the hospital through the emergency department. Also, the medical resources on weekends, such as the number of available medical devices, the configuration of doctors and nursing staff, etc., will be different from weekdays. All these will affect the discharge method through the differences in the admission routes of patients or the available admission methods, and may affect the discharge method due to differences in variables such as treatment plans, medical outcomes, and length of hospital stay. To a certain extent, it can reflect the differences in the medical service contact and usage patterns between weekends and weekdays, so it belongs to the features containing the weekend effect.

[0033] S4. Time series features:

[0034] Taking the surgical period as an intermediate variable between the admission period and the discharge method, a vector autoregressive model is constructed. Multidimensional time series data of inpatients is input, including various variable data of the admission period, surgical period, and discharge period. The optimal order (lag period) of the vector autoregressive model is selected through information criteria. During the model training process, the parameters of the vector autoregressive model are estimated by the least squares method, and the performance of the vector autoregressive model is estimated using the cross-validation method to ensure that the model can capture the dynamic change characteristics of each variable at the admission time point, surgical time point, and discharge time point in the time series data. Then, multiple regression analysis is used to predict the values of each variable at each time step except the initial time step - admission time step, and a time series dynamic change feature is constructed, which consists of the original data of each variable at the admission time step, the predicted data of each variable at the surgical time step, and the discharge method as the predicted data.

[0035] The dynamic time warping technique is applied to measure the similarity of time series. The time series data of the admission period and surgical period of inpatients is input, and the similarity degree of patients on the time series is calculated. A group of similarity features of the admission-surgery time-discharge method sequence is generated for each patient.

[0036] The dynamic change features extracted by the vector autoregressive model and the similarity features calculated by the dynamic time warping technique are merged. The two feature vectors are spliced into a comprehensive feature vector through feature connection, and the merged features are standardized to ensure that different features have the same dimension and avoid bias caused by some features in the subsequent prediction model training.

[0037] S5. Steps for predicting the discharge method:

[0038] 5-1. Steps for integrating causal relationship feature data of weekend effects:

[0039] The key variable features, causal relationship features, and time series features are merged to ensure that all feature dimension formats are consistent and complete. The key variable features include basic information such as admission date, surgical date, and admission method. The causal relationship features come from Bayesian causal networks and causal forest models, including the causal probability and causal effect of the features on the discharge method. The time series features include the dynamic change features and similarity features generated by the vector autoregressive model and the dynamic time warping technique. The data in each feature is aligned according to the admission time point, surgical time point, and discharge time point to ensure that all time point data is correctly matched and avoid time point misalignment of data. The alignment process needs to consider the timestamp information of the data to ensure that the feature data of each time step corresponds to the corresponding time point. The feature data is standardized to unify the dimension of the features and ensure that the data is compared and analyzed on the same scale. The standardized process ensures that the model training process is not affected by the difference in feature dimensions and improves the training effect and prediction accuracy of the model.

[0040] 5-2. Long short-term memory network processing steps:

[0041] Recurrent neural networks are neural networks that process sequence data and are able to process sequences of arbitrary lengths, with information being passed from one unit to the next over time. Standard recurrent neural networks face the problems of gradient vanishing and gradient exploding, which limit their ability to learn on long sequences. Long short-term memory networks regulate the flow of information by introducing three gating structures, effectively preserving long-term dependencies and avoiding gradient problems. Long short-term memory networks are a special type of recurrent neural network with memory units and gating units that can better capture long-term dependencies in sequences. In a long short-term memory network, the output at each time point depends not only on the input and hidden state at the current time point, but also on the hidden state and cell state at the previous time point. This enables long short-term memory networks to better process sequence data, especially on long sequences.

[0042] In these gating mechanisms, the forget gate determines how much of the cell state information at the previous moment is retained. The output gate determines how much new information content is updated to the cell state. The candidate memory cell generates new information candidates. The new cell state in the state update is updated based on the output of the forget gate and the input gate, including the forgotten part of the information from the previous cell state and adding new information, and the output is guaranteed to be between -1 and 1 through the tanh function. The final hidden state is calculated based on the output gate and the current cell state, and determines the information content of the output. These operations ensure that the hidden state contains the information of the current time step and will be used as the input part of the next time step.

[0043] The long short-term memory network model processing enhances the understanding and memory of the time series of hospitalized patients during the admission period, surgery period and discharge mode, and combines the weekend effect characteristic data to improve the accuracy and efficiency of predicting the discharge mode of hospitalized patients. Through this method, the present invention ensures the high accuracy and robustness of analyzing the discharge mode under the weekend effect, and further provides a suitable treatment time plan for hospital management and patient treatment.

[0044] The application of LSTM networks is to effectively capture and analyze the time series data of hospitalized patients during the period of admission, surgery, and discharge. The uniqueness of LSTM networks lies in their ability to establish forward and backward dependencies for each time step in the time series. Specifically, when the LSTM network analyzes the data of a specific treatment day, it not only takes into account the characteristics related to the weekend effect during the admission period, but also comprehensively considers the characteristics related to the weekend effect during the surgery period. This comprehensive consideration of the context enables the LSTM network to more comprehensively and accurately grasp the contextual relationship of the key moments in the treatment process of hospitalized patients under the influence of the weekend effect when dealing with and analyzing the discharge patterns of hospitalized patients with the weekend effect.

[0045] During the implementation process, the processed feature data is input into the long short-term memory network. The long short-term memory network can process complex time-series data, capture the dependencies between the time steps of the admission time and the surgery time, and is suitable for processing data with time-series characteristics. Through the long short-term memory network, we can better understand the time-series data of the admission time and the surgery time, and combine the weekend effect feature data to improve the accuracy and efficiency of predicting the discharge mode of inpatients.

[0046] 5-3. Processing steps of the time distribution fully connected layer:

[0047] The time fully connected layer is used to output the predicted discharge mode of inpatients, achieving accurate prediction of the discharge mode.

[0048] Using the training data, the network parameters are adjusted by the backpropagation algorithm to minimize the loss function. The long short-term memory network outputs the time-series features of each time step, and then these features are input into the fully connected layer for further processing. The fully connected layer processes the time-series features output by the long short-term memory network through linear transformation and activation functions to extract high-level feature information. Based on the time-series features output by the long short-term memory network, the fully connected layer predicts the discharge mode of the patient. The prediction result can obtain the specific discharge mode category through the Softmax function or other classifiers. The model performance is evaluated by metrics such as accuracy and F1 score. The validation dataset is used to evaluate the prediction performance of the model, and the model is further optimized through cross-validation and hyperparameter tuning. According to the model evaluation results, the network structure and parameters are adjusted to improve the prediction effect of the model, and finally the best-performing long short-term memory network model is obtained.

[0049] Combining the application of the long short-term memory network structure and the fully connected layer, the model we proposed can effectively process and analyze the time-series data of inpatients in terms of the admission time, surgery time, and discharge mode. This method has important significance for the analysis of time-series data in the medical field, can help medical professionals better understand and predict the discharge mode of patients under the weekend effect, so as to optimize the treatment plan and provide better medical services for patients. Through the present invention, we can provide a more accurate and reliable tool for analyzing the impact of the weekend effect in the medical field, provide more powerful support for medical management decisions to solve the negative impact of the weekend effect, and provide more appropriate diagnosis and treatment time suggestions for patients.

[0050] A model for predicting the discharge mode of inpatients under the influence of the weekend effect, which implements the above steps, includes a random forest, a Bayesian causal network, a vector autoregressive model, a dynamic time warping technique, and a long short-term memory network. When the data of the admission and surgery periods of historical inpatients is input into the discharge mode prediction model, after the above steps, the discharge mode prediction model can output the predicted discharge mode.

[0051] The embodiment selects the electronic medical record information of inpatients in a certain municipal tertiary hospital and performs a number of preprocessing steps to ensure the accuracy and relevance of the data.

[0052] The embodiment selects the electronic medical record data of inpatients with ischemic heart disease in a certain municipal tertiary hospital and performs a number of preprocessing steps to ensure the accuracy and relevance of the data. Then, the random forest algorithm is used for feature selection to identify the key factors affecting the discharge mode in the weekend effect. The Bayesian causal network and the causal forest model are applied to explore the causal relationships between the key factors. Then, the surgery period is regarded as an intermediate variable between the admission period and the discharge mode, and the vector autoregressive model is used to analyze the time series data to capture the dynamic relationships between variables; the dynamic time warping technique is applied to measure the similarity of the admission and surgery processes of different patients and generate time series similarity features; the key factors are combined with the causal features, the features generated by the vector autoregressive model and the dynamic time warping technique and input into the long short-term memory network model to better predict the discharge mode of patients, and cross-validation is used to evaluate the generalization ability of the model, predict the discharge mode of patients under the weekend effect in medical services, provide a scientific basis for clinical decision-making, and support more accurate medical management and decision-making.

[0053] The embodiment scheme has been verified through a series of experiments, demonstrating its significant feasibility and advantages in predicting the discharge mode of inpatients under the influence of the weekend effect. The experimental results show that the model shows high efficiency in mining the causal features of data related to inpatients under the weekend effect, demonstrating the excellent ability of the model to extract causal features from time series in medical data. In the deep learning model training stage, the model architectures of the random forest, the Bayesian causal network, the vector autoregressive model, the dynamic time warping technique, and the long short-term memory network show their strong ability to mine causal information in medical data related to the weekend effect, and the prediction of the discharge mode is significantly accurate in key performance indicators such as accuracy, precision, recall rate, and F1 score, thus reflecting their superior learning and generalization ability.

[0054] During the validation phase, the model maintained high-accuracy predictions of the discharge method for the time series datasets of the admission period and surgical period of new and unknown inpatients, demonstrating its good adaptability and generalization ability. During the testing phase, even when dealing with diverse inpatient electronic medical record data, the model maintained a high level of accuracy and stability. These results indicate that our model can not only accurately predict the discharge method of inpatients affected by the weekend effect, but also maintain the stability and accuracy of its predictions on the electronic medical record data of users with different diagnostic types.

[0055] In summary, the experimental verification results of the random forest, Bayesian causal network, vector autoregressive model, dynamic time warping technology, and long short-term memory network model architecture strongly demonstrate the application value of the present invention in the field of predicting the discharge method of inpatients affected by the weekend effect. It is not only innovative in theory but also shows significant effects and advantages in practical applications. These advantages make the present invention an efficient and reliable tool for predicting the weekend effect on the final discharge method of inpatients in medical practice, contributing to promoting the comprehensive progress of improving medical quality and effectively utilizing medical resources associated with the weekend effect, thus making an important contribution to the management of inpatients in the medical field.

Claims

1. A method for allocating hospitalization resources under the weekend effect based on machine learning, characterized in that: Includes steps: Data collection and preprocessing steps: Inpatient data were collected from the medical service database in date order, and the date variables in the inpatient data were marked as working days and non-working days; the inpatient data included the patient's personal information, data during admission, data during surgery, and data during discharge; Key feature selection step: Use the random forest algorithm to construct a decision tree to evaluate the importance of each variable in the inpatient data, and identify the key variables that affect the discharge mode as key features; Causal feature selection step: Based on the selected key feature - discharge mode, a Bayesian causal network is constructed to explore the causal relationship between the key features including the weekend effect and the discharge mode. The nonlinear relationship and interaction between the key features of the causal forest model based on the decision tree are used to obtain the causal relationship features for predicting the discharge mode of inpatients under the weekend effect. Steps for generating time series features: Use the time series data consisting of the data of the hospitalized patients during the admission period, the data of the surgery period, and the data of the discharge period to train the vector autoregression model. The vector autoregression model learns to capture the dynamic change relationship of each variable at each time point during the admission period, the surgery period, and the discharge period. The dynamic time warping technology is used to measure the similarity of the admission and surgery processes of different patients. The trained vector autoregression model is used to receive the input time series data of the hospitalized patients during the admission period and the surgery period, calculate the similarity of each hospitalized patient in the time series, and output a set of time series features of the admission-surgery-discharge mode corresponding to each patient; Discharge mode prediction steps: merge key features, causal relationship features and time series features, preprocess the merged features, and then input the preprocessed features into the long short-term memory network. The long short-term memory network outputs the time series features of each time step to the fully connected layer; The fully connected layer outputs the predicted discharge mode at each time step based on the input time series features; Inpatient resource allocation steps: Inpatient resources are allocated based on predicted discharge patterns.

2. The method according to claim 1, characterized in that: The specific method of preprocessing the merged features is: align the variables in the merged key features, causal relationship features, and time series features according to the admission time point, surgery time point, and discharge time point; then standardize the feature data to unify the dimension of the features.

3. The method according to claim 1, characterized in that: The data during the admission period include the patient's admission date, admission route, admission type, whether it is ST-segment elevation myocardial infarction (STEMI), admission diagnosis, admission condition and case classification.

4. The method according to claim 1, characterized in that: Data from the surgical period included surgical date, elective surgery, surgical level, and surgical-related information.

5. The method according to claim 1, characterized in that: Data on discharge included the date and method of discharge.

6. The method according to claim 1, characterized in that: Key characteristics that account for the weekend effect include admission date, admission route, and surgery date.

7. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to claim 1 are implemented.

Images