Medical team automatic scheduling method, device and system

Abstract

The invention discloses a medical team automatic scheduling method, device and system. The method comprises the following steps: acquiring scheduling element data of a to-be-implemented operation; taking the scheduling element data as input of a medical team distribution model, and predicting qualification requirement information of a plurality of medical personnel posts in the medical team to be subjected to the operation by the medical team distribution model according to the input scheduling element data; the medical team allocation model is obtained by training based on the operation information of the performed operation, the patient information and the personnel qualification information of a plurality of medical personnel posts in the medical team of the performed operation. And based on the qualification requirement information predicted by the medical team distribution model, matching medical personnel for a plurality of medical personnel posts in the medical team to be subjected to the operation, and generating a medical team scheduling scheme to be subjected to the operation. According to the invention, the automatic scheduling of the medical team is realized, the scheduling processing efficiency is improved, the scheduling work burden of medical personnel is reduced, the distribution of medical resources is optimized, and the efficiency of emergency treatment is improved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular, to a method, device, and system for automatically scheduling medical teams. Background Art

[0002] In the current medical system, in the face of emergencies such as major disasters, public health emergencies, and emergency responses, corresponding mobile cabin hospitals often need to quickly mobilize a large number of medical resources to meet the rapidly increasing patient needs, especially surgical needs. However, traditional surgical scheduling methods mainly rely on manual decision-making and experience judgment, which expose many problems and challenges in emergency situations. Among them, the most severe one is the low efficiency of information acquisition and the low efficiency of medical team scheduling processing.

[0003] Specifically, in the scenario of emergency events, mobile cabin hospitals need to quickly collect and analyze key information such as the injury conditions of patients and the availability of medical staff. However, due to the special environment, medical teams and on-site resource schedulers are often not familiar with each other, making it difficult to complete complex resource information scheduling work in a short time, resulting in slow decision-making for medical team scheduling and even unreasonable allocation of medical resources. For example, some patients in urgent need of surgery may wait for a long time due to tight medical resources, while some medical staff may be idle, causing waste of medical resources.

[0004] Moreover, in emergency events, medical staff themselves already bear huge work pressures and psychological burdens. The traditional surgical scheduling method based on manual experience judgment and decision-making further increases the workload of medical staff and brings additional work burdens. Summary of the Invention

[0005] Based on the above problems, this application provides a method, device, and system for scheduling medical teams, aiming to improve the scheduling efficiency of medical teams and reduce the work burden of medical staff through an intelligent and automated scheduling method.

[0006] The embodiments of this application disclose the following technical solutions:

[0007] In the first aspect of this application, a method for automatically scheduling a medical team is provided, and the method includes:

[0008] Obtain the scheduling element data of the surgery to be performed; the scheduling element data includes surgery information and patient information;

[0009] Use the scheduling element data as the input of the medical team allocation model. The medical team allocation model predicts the qualification requirement information of multiple medical staff positions within the medical team for the surgery to be performed based on the input scheduling element data. Among them, the medical team allocation model is a model trained based on the surgery information, patient information of the surgeries already performed, and the personnel qualification information of multiple medical staff positions within the medical teams of the surgeries already performed.

[0010] Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, match medical staff for multiple medical staff positions within the medical team for the surgery to be performed, and generate a scheduling plan for the medical team of the surgery to be performed.

[0011] In an optional implementation, based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, match medical staff for multiple medical staff positions within the medical team for the surgery to be performed, and generate a scheduling plan for the medical team of the surgery to be performed, including:

[0012] Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, as well as the position information and qualification information of multiple medical staff in the hospital, match medical staff for multiple medical staff positions within the medical team for the surgery to be performed, and generate an initial scheduling plan for the medical team of the surgery to be performed;

[0013] On the scheduling plan review interface, display the initial scheduling plan for the medical team of the surgery to be performed; in response to the review and confirmation operation on the initial scheduling plan for the medical team of the surgery to be performed, use the initial scheduling plan for the medical team of the surgery to be performed as the final scheduling plan for the medical team of the surgery to be performed;

[0014] Or, in response to the scheduling modification operation on the initial scheduling plan for the medical team of the surgery to be performed, re-display the modified scheduling plan for the medical team of the surgery to be performed on the scheduling plan review interface; in response to the review and confirmation operation on the modified scheduling plan for the medical team of the surgery to be performed, use the modified scheduling plan for the medical team of the surgery to be performed as the final scheduling plan for the medical team of the surgery to be performed.

[0015] In an optional implementation, based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, match medical staff for multiple medical staff positions within the medical team for the surgery to be performed, and generate a scheduling plan for the medical team of the surgery to be performed, including:

[0016] In response to the scheduling trigger operation for the surgery to be performed on the scheduling plan review interface, display the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model;

[0017] In response to a triggering operation on a personnel selection control corresponding to the qualification requirement information for a medical staff position, display the scheduling information of medical staff who meet the corresponding qualification requirement information;

[0018] In response to a triggering operation on a medical staff card in the scheduling information, construct a matching relationship between the medical staff position and the medical staff;

[0019] After multiple medical staff positions have all been matched with medical staff, generate a scheduling plan for the medical team for the surgery to be performed.

[0020] In an optional implementation manner, after displaying the qualification requirement information for multiple medical staff positions predicted by the medical team allocation model, the method further includes:

[0021] In response to a triggering operation on an audit and modification control corresponding to the qualification requirement information for a medical staff position, display one or more alternative qualification requirement information for the medical staff position;

[0022] In response to a selection triggering operation on one piece of alternative qualification requirement information for the medical staff position, replace the displayed qualification requirement information for the medical staff position with the selected alternative qualification requirement information.

[0023] In an optional implementation manner, the method further includes:

[0024] Optimize the medical team allocation model based on the audited and modified scheduling plan for the medical team.

[0025] In an optional implementation manner, the training steps of the medical team allocation model include:

[0026] Collect training data for the model to be trained; the training data includes surgical information, patient information of surgeries already performed in the hospital, and personnel qualification information of multiple medical staff positions within the medical teams of the surgeries already performed; the model to be trained includes an input layer, a double-layer LSTM layer, a fully connected layer, and an output layer;

[0027] The input layer extracts features from the surgical information and patient information in the training data, generates a one-dimensional feature vector, and passes it to the double-layer LSTM layer;

[0028] The double-layer LSTM layer updates the state using internal memory units and gating mechanisms, and outputs a hidden state vector;

[0029] The fully connected layer performs a linear transformation on the hidden state vector and passes the transformed vector to the output layer;

[0030] The output layer outputs multiple predicted probability distribution vectors based on the transformed vector; each predicted probability distribution vector corresponds to a medical staff position; the elements of the predicted probability distribution vector correspond one-to-one with the qualification levels of the medical staff position;

[0031] Based on the qualification requirement information of multiple medical staff positions reflected by multiple prediction probability distribution vectors and the differences between the personnel qualification information of multiple medical staff positions in the medical team that has performed surgeries in the training data, adjust the parameters of the model to be trained, and iterate the training until the training cut-off condition is met to obtain a medical team allocation model.

[0032] In an alternative implementation, the multiple medical staff positions include: the surgeon, the first assistant in the surgery, the second assistant in the surgery, the anesthesiologist, and the nurse;

[0033] The multiple prediction probability distribution vectors include: the first prediction probability distribution vector corresponding to the surgeon, the second prediction probability distribution vector corresponding to the first assistant in the surgery, the third prediction probability distribution vector corresponding to the second assistant in the surgery, the fourth prediction probability distribution vector corresponding to the anesthesiologist, and the fifth prediction probability distribution vector corresponding to the nurse;

[0034] Among them, each of the first prediction probability distribution vector, the second prediction probability distribution vector, the third prediction probability distribution vector, and the fourth prediction probability distribution vector contains 4 elements, and the fifth prediction probability distribution vector contains 3 elements; the qualification levels corresponding to the 4 elements are: resident physician, attending physician, deputy chief physician, and chief physician; the qualification levels corresponding to the 3 elements are: junior, intermediate, and senior.

[0035] In an alternative implementation, the surgery information in the training data includes the surgery name, surgery level, surgery site, incision level, anesthesia method, and actual surgery duration; the patient information in the training data includes the patient's gender and age;

[0036] The surgery information in the scheduling element data includes the surgery name, surgery level, surgery site, incision level, anesthesia method, and expected surgery duration; the patient information in the scheduling element data includes the patient's gender and age.

[0037] In an alternative implementation, after collecting the training data of the model to be trained, the method further includes:

[0038] Preprocess the training data;

[0039] The input layer extracts features from the surgery information and patient information in the training data, specifically: the input layer extracts features from the surgery information and patient information in the preprocessed training data;

[0040] The preprocessing methods include one or more of the following:

[0041] Duplicate removal, format unification, error correction, or missing value processing.

[0042] In an alternative implementation, after determining the final schedule plan for the medical team of the surgery to be performed, the method further includes:

[0043] Sending a notification message of the surgery to be performed to the medical staff related to the final schedule plan of the medical team, where the notification message includes medical team personnel information, patient information, and surgery time information.

[0044] A second aspect of the present application provides a medical team automatic scheduling device, which includes:

[0045] A data acquisition module, configured to acquire scheduling element data of the surgery to be performed; the scheduling element data includes surgery information and patient information;

[0046] A qualification requirement prediction module, configured to use the scheduling element data as an input to a medical team allocation model, and the medical team allocation model predicts qualification requirement information for multiple medical staff positions in the medical team of the surgery to be performed according to the input scheduling element data; wherein, the medical team allocation model is a model trained based on the surgery information, patient information of the surgeries already performed, and the personnel qualification information of multiple medical staff positions in the medical teams of the surgeries already performed;

[0047] A personnel matching module, configured to match medical staff for multiple medical staff positions in the medical team of the surgery to be performed based on the qualification requirement information for multiple medical staff positions predicted by the medical team allocation model, and generate a schedule plan for the medical team of the surgery to be performed.

[0048] A third aspect of the present application provides a medical team automatic scheduling system, which includes: a scheduling device and a display device; the scheduling device is communicatively connected to the display device; the scheduling device includes a processor and a memory;

[0049] The memory is configured to store a computer program;

[0050] The processor is configured to run the computer program, and when the computer program runs, it executes the steps of the medical team automatic scheduling method introduced in any implementation manner of the first aspect, and controls the display device to display the generated schedule plan for the medical team of the surgery to be performed.

[0051] Compared with the prior art, the present application has the following beneficial effects:

[0052] In the technical solution of this application, in order to improve the scheduling efficiency of the medical team, first, the surgical information and patient information of the surgery to be performed are obtained as scheduling element data. Then, the scheduling element data is used as the input of the medical team allocation model. According to the input scheduling element data, the medical team allocation model predicts the qualification requirement information of multiple medical staff positions within the medical team for the surgery to be performed. Since this medical team allocation model is a model trained based on the surgical information, patient information of the surgeries that have been performed, and the personnel qualification information of multiple medical staff positions within the medical teams of the surgeries that have been performed, this model is capable of predicting, based on the surgical information and patient information of the surgery, the personnel qualifications that should be configured for each medical staff position in the medical team for performing this surgery. For example, it is predicted that the position of the surgeon should be staffed with attending physicians, deputy chief physicians, or chief physicians. Finally, based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, medical staff are matched to multiple medical staff positions within the medical team for the surgery to be performed, and a scheduling plan for the medical team for the surgery to be performed is generated. Since the model has predicted the qualification requirement information of the positions, and the qualifications of each medical staff are determined, convenient to query and index, therefore, the processing efficiency of the medical team scheduling is greatly improved, the scheduling time is saved, and the medical team scheduling for each surgery to be performed can be carried out intelligently and automatically. The technical solution of this application improves the timeliness and effectiveness of the mobilization of medical resources in the face of scenarios such as major disasters, public health emergencies, and emergency responses, and can help to reasonably allocate medical resources and reduce the workload of medical staff in medical team scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0054] Figure 1 It is a schematic flowchart of a method for automatic scheduling of a medical team provided by an embodiment of the present application;

[0055] Figure 2 It is an overall implementation flowchart of a method for automatic scheduling of a medical team provided by an embodiment of the present application;

[0056] Figure 3 It is a schematic diagram of the predicted output of a medical team allocation model provided by an embodiment of the present application;

[0057] Figure 4 It is a schematic flowchart of another method for automatic scheduling of a medical team provided by an embodiment of the present application;

[0058] Figure 5 It is a schematic flowchart of another medical team automatic scheduling method provided by an embodiment of the present application;

[0059] Figure 6A It is a schematic diagram of a scheduling plan review interface provided by an embodiment of the present application;

[0060] Figure 6B It is a schematic diagram of an editing area in the scheduling plan review interface;

[0061] Figure 7 It is a schematic structural diagram of a medical team automatic scheduling device provided by an embodiment of the present application;

[0062] Figure 8 It is a schematic structural diagram of a medical team automatic scheduling system provided by an embodiment of the present application. Detailed implementation manners

[0063] As described above, currently, when dealing with some emergency medical events, it is often necessary to mobilize a large number of medical resources to serve a large number of urgent medical needs. For example, multiple patients respectively need to undergo surgeries of different levels or even surgeries on different parts. Currently, it often relies on medical staff to make manual decisions based on understanding the patient's injuries and combining the specific situations of the known medical resources (such as the qualifications of each doctor and nurse) to schedule the surgeries for the patients and allocate corresponding medical teams. However, the manual decision-making method is often inefficient, relying on the timely acquisition and accurate judgment of information, which poses a great challenge to the medical staff responsible for resource scheduling. In some cases, it is very likely to cause problems such as slow decision-making or unreasonable resource allocation, affecting the timeliness and effectiveness of medical treatment. In addition, for the medical staff responsible for resource scheduling, the workload is large and the burden is heavy.

[0064] Regarding the deficiencies of the traditional manual scheduling method introduced above, the inventors have proposed a medical team automatic scheduling method, device and system through research. Based on the surgical information and patient information of the surgery to be performed, a medical team allocation model trained in advance is used to predict the qualification requirement information for multiple medical staff positions within the medical team. Thus, it is not necessary to spend too much manpower to analyze which level of surgery, which part of the surgery, and which condition of the patient should be performed by medical staff with what qualifications. In the technical solution of the present application, on the premise that the qualification requirement information for multiple medical staff positions has been predicted, only by combining the specific qualifications of the relevant personnel in each medical staff position, a suitable medical team can be matched for the surgery to be performed to generate a scheduling plan. Thus, the above-mentioned problems are effectively solved.

[0065] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0066] See Figure 1 , which is a schematic flowchart of a method for automatically scheduling a medical team provided by an embodiment of this application. As Figure 1 shown, the method for automatically scheduling a medical team includes:

[0067] S101. Obtain the scheduling element data of the surgery to be performed.

[0068] In the embodiments of this application, the scheduling element data is the necessary data material for realizing the scheduling of the medical team. The completeness and accuracy of the scheduling element data directly determine the effect of the scheduling. Specifically, the scheduling element data includes surgical information and patient information. In the scheduling element data, as an example, the surgical information includes the name of the surgery, the surgical grade, the surgical site, the incision grade, the anesthesia method, and the expected duration of the surgery. In the scheduling element data, the patient information includes the patient's gender and age.

[0069] Among them, the expected duration of the surgery can be the duration estimated by medical staff based on the specific difficulty of the surgery and the physical condition of the patient. In addition, some intelligent means can also be used to estimate the expected duration of the patient's surgery, such as predicting the expected duration of the surgery based on information such as the patient's age, physical examination and test indexes, surgical site, and surgical grade.

[0070] In a possible implementation manner, the expected duration of the surgery is represented by a fixed time length. For example, the expected duration of a certain patient's surgery is 60 minutes. In some other possible implementation manners, the expected duration of the surgery can also be a time interval. For example, the expected duration of a certain patient's surgery is 30-60 minutes. The provision of the expected duration of the surgery facilitates predicting the medical team configuration that meets this duration expectation, and avoids the situation where the personnel qualifications and capabilities do not match the duration expectation of this surgery.

[0071] S102. Use the scheduling element data as the input of the medical team allocation model, and the medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed according to the input scheduling element data.

[0072] In the embodiments of the present application, the medical team allocation model is a model trained based on the surgical information, patient information of the implemented surgeries, and the qualification information of multiple medical staff positions within the medical teams of the implemented surgeries. The qualification information of multiple medical staff positions within the medical teams of the implemented surgeries is used as labels, and the surgical information and patient information of the implemented surgeries are used as the inputs in the model prediction stage, so that the trained medical team allocation model has the ability to accurately predict the qualification requirement information of medical staff positions based on the surgical information and patient information.

[0073] To facilitate the understanding of the medical team allocation model, the training process of this model will be introduced below.

[0074] In an optional implementation manner, the deep learning framework TensorFlow is used to build the model. The model to be trained includes an input layer, a double long short-term memory (LSTM) layer, a fully connected layer, and an output layer connected in sequence. Before training, the training data of the model to be trained needs to be collected first. The training data mainly comes from the operating room anesthesia information system (abbreviation: anesthesia system) within the hospital and the electronic medical record (EMR) system within the hospital. Among them, the anesthesia system is responsible for managing and storing patient surgery-related data. It should be noted that to schedule the medical team for surgeries, a stable data synchronization mechanism needs to be established with the emergency system to ensure the timeliness and accuracy of the data. Therefore, in the embodiments of the present application, it is proposed that this can be achieved through methods such as API interfaces, database triggers, or scheduled tasks, ensuring that whenever a new surgery application is generated in the emergency system, this data can be synchronized automatically or on demand to optimize the data quality for model training.

[0075] The training data includes the data input into the input layer of the model during training and the data used as the data labels for model training. Specifically, the training data includes the surgical information and patient information of the implemented surgeries within the hospital, which are used as the inputs to the input layer of the model. The training data also includes the qualification information of multiple medical staff positions within the medical teams of the implemented surgeries as data labels. Among them, the former is similar to the data structure of the model input data when actually using the medical team allocation model. The surgical information in the training data includes the surgery name, surgery level, surgical site, incision level, anesthesia method, and actual surgery duration, and the patient information in the training data includes the patient's gender and age.

[0076] To optimize the effect of model training, after collecting the training data, it can be preprocessed to improve the data quality of the training data. In one possible implementation, after collecting the training data of the model to be trained, the training data is preprocessed. The preprocessing methods include one or more of the following: deduplication, format unification, error correction, or missing value handling. These operations can all achieve data cleaning.

[0077] The reason for data cleaning is that the data synchronized from the hospital's internal anesthesia system and the in-hospital EMR system often contains redundant, incorrect, or inconsistent information, so data cleaning is required. The following introduces the specific operations of several data cleaning methods. Deduplication: Check and delete duplicate records, especially for redundant master tables, to ensure the uniqueness of the data. Format unification: Unify the data formats from different sources, such as date and time formats, unit conversions, etc., for subsequent processing and analysis. Error correction: Identify and correct obvious data errors, such as logical errors, range errors, etc. Missing value handling: For missing key information, fill it according to business rules or mark it as missing.

[0078] Through the above data cleaning operations, the data quality is effectively improved, and the interference of redundant or incorrect data on model training during the model training process is reduced. The model convergence speed is improved. For the above implementation method of preprocessing the training data, correspondingly, the input layer extracts features from the surgical information and patient information in the training data, specifically: the input layer extracts features from the surgical information and patient information in the preprocessed training data.

[0079] In an alternative implementation, multiple medical staff positions include: the surgeon, the first assistant in the operation, the second assistant in the operation, the anesthesiologist, and the nurse. In some specific implementation scenarios, there may be more doctors or fewer doctors in the medical team of a surgery. For example, in the case of more doctors, the medical team may also include the third assistant in the operation. In the case of fewer doctors, for example, the medical team only includes the surgeon, the anesthesiologist, and the nurse. The above are only examples and do not limit the medical staff positions.

[0080] In one example, the data form of the training data is [Surgery Name: SURGERY_NAME; Surgery Level: SURGERY_LEVEL_NAME; Operation Site: OPERATION_TYPE; Incision Level: AGGLU_LEVEL_NAME; Anesthesia Method: ANESTHESIA_TYPE_NAME; Actual Surgery Duration: DURATION_TIME; Patient Gender: IDENTIFICATION_SEX; Patient Age: IDENTIFICATION_AGE; Surgeon Level: OPERATOR_LEVEL; First Assistant Level of Surgery: FIRST_ASSISTANT_LEVEL; Second Assistant Level of Surgery: SECOND_ASSISTANT_LEVEL; Anesthesiologist Level: ANAESTHETIST_LEVEL; Nurse Level: NURSE_LEVEL;].

[0081] The possible qualification levels of doctors and nurses are introduced below. The qualification levels of surgeons such as the surgeon, the first assistant in surgery, and the second assistant in surgery all include four levels: resident doctor, attending doctor, deputy chief doctor, and chief doctor. The qualification levels of anesthesiologists include four levels: resident doctor, attending doctor, deputy chief doctor, and chief doctor. The qualification levels of nurses include three levels in total: junior, intermediate, and senior.

[0082] After the training data in the above data form enters the input layer, the input layer extracts features from the surgery information and patient information in the training data, generates a one-dimensional feature vector, and passes it to the double-layer LSTM layer. In the double-layer LSTM layer, each layer contains 129 LSTM units. The double-layer LSTM layer updates the state using internal memory units and gating mechanisms, and outputs a hidden state vector. The fully connected layer performs a linear transformation on the hidden state vector and passes the transformed vector to the output layer. The output layer outputs multiple predicted probability distribution vectors based on the transformed vector.

[0083] Before training, it is necessary to digitize non-numerical data. For example, the data representing the qualification levels of each medical staff position in the training data. As an example, for the chief surgeon, the first assistant in the operation, the second assistant in the operation, and the anesthesiologist, their respective qualification levels can all be represented as binary vectors after one-hot encoding with a length of 4. Among them, if the qualification level is resident physician, the vector is represented as 1000; if the qualification level is attending physician, the vector is represented as 0100; if the qualification level is deputy chief physician, the vector is represented as 0010; if the qualification level is chief physician, the vector is represented as 0001. Similarly, the qualification level of nurses can be represented as a binary vector after one-hot encoding with a length of 3. Among them, if the qualification level is junior, the vector is represented as 100; if the qualification level is intermediate, the vector is represented as 010; if the qualification level is senior, the vector is represented as 001. For other data in the training data except for the qualification level, if it is non-numerical, it also needs to be digitized. For example, the operation name can be represented by the operation ICD code; the operation site can be uniquely indicated by a specific code or a random code; the patient's gender can be distinguished by the numbers 0 and 1. The anesthesia method can be represented by a code. After digitization, the data form is as follows:

[0084]

Surgery ICD code: SURGERY_NAME; Surgery level: SURGERY_LEVEL_NAME; Operation site code: OPERATION_TYPE; Incision level: AGGLU_LEVEL_NAME; Anesthesia method: ANESTHESIA_TYPE_CODE; Actual duration of the operation: DURATION_TIME; Patient gender (0, 1): IDENTIFICATION_SEX; Patient age: IDENTIFICATION_AGE; Chief surgeon level: OPERATOR_LEVEL; First assistant level in the operation: FIRST_ASSISTANT_LEVEL; Second assistant level in the operation: SECOND_ASSISTANT_LEVEL; Anesthesiologist level: ANAESTHETIST_LEVEL; Nurse level: NURSE_LEVEL;

[0085] The above data after processing is converted into a one-dimensional vector [SURGERY_NAME; SURGERY_LEVEL_NAME; OPERATION_TYPE; AGGLU_LEVEL_NAME; ANESTHESIA_TYPE_CODE; DURATION_TIME; IDENTIFICATION_SEX; IDENTIFICATION_AGE]. After receiving this vector, the input layer of the model passes it to the double-layer LSTM layer. The double-layer LSTM layer updates the state according to the internal memory units and gating mechanisms and outputs a hidden state vector. The fully connected layer receives the hidden state vector from the upper layer and then performs a linear transformation on this vector: z = Wx + b. Here, z is the transformed vector, W is the weight matrix, x is the hidden state vector received from the upper layer, and b represents the bias term.

[0086] The fully connected layer outputs the transformed vector z to the output layer. Since it is expected that the model outputs the qualification requirement information for five medical staff positions, the structure of the output layer can be set in advance. In practical applications, the output layer can be set to include 5 output vectors, among which 4 output vectors each include 4 neurons, and 1 output vector includes 3 neurons. As an example, in the output vector containing 4 neurons, the first neuron corresponds to the prediction probability of "Chief Physician", the second neuron corresponds to the prediction probability of "Deputy Chief Physician", the third neuron corresponds to the prediction probability of "Attending Physician", and the fourth neuron corresponds to the prediction probability of "Resident Physician". In the above-mentioned output vector containing 3 neurons, the first neuron corresponds to the prediction probability of "Junior", the second neuron corresponds to the prediction probability of "Intermediate", and the third neuron corresponds to the prediction probability of "Senior".

[0087] At the output layer, the softmax function is used to calculate the vector z to obtain the prediction probability of the qualification level corresponding to each dimension in the prediction probability distribution vector. Among them, the softmax function can be expressed as:

[0088]

[0089] In the above formula, i represents a specific dimension in the vector, and j takes values in turn among all dimensions in the vector. Among them, p i represents the probability of the i-th dimension, and z i is the element corresponding to the i-th dimension in the output vector. After being processed by the output layer, 5 prediction probability distribution vectors will be output, which are: the first prediction probability distribution vector z 0 ' = [p 0 , p 1 , p 2 , p 3 , the second prediction probability distribution vector z1 ' = [p 0 , p 1 , p 2 , p 3 , the third predicted probability distribution vector z 2 ' = [p 0 , p 1 , p 2 , p 3 , the fourth predicted probability distribution vector z 3 ' = [p 0 , p 1 , p 2 , p 3 , and the fifth predicted probability distribution vector z 4 ' = [p 0 , p 1 , p 2 . Each predicted probability distribution vector output by the output layer corresponds to a medical staff position. For example, the first predicted probability distribution vector corresponds to the surgeon, z 0 ' = [p 0 , p 1 , p 2 , p 3 , z 0 ' represents the predicted probability distribution of the surgeon. According to the probability values of p 0 , p 1 , p 2 , p 3 , the level of the surgeon to be configured for the current patient to undergo the current operation can be determined. The second predicted probability distribution vector corresponds to the first assistant in the operation, z 1 ' = [p 0 , p 1 , p 2 , p 3 , z 1 ' represents the predicted probability distribution of the first assistant in the operation. According to the probability values of p 0 , p 1 , p 2 , p 3 , the level of the first assistant in the operation to be configured for the current patient to undergo the current operation can be determined. The third predicted probability distribution vector corresponds to the second assistant in the operation, z 2 ' = [p 0 , p 1 , p 2 , p 3 , z 2 ' represents the predicted probability distribution of the second assistant in the operation. According to the probability values of p 0 , p 1 , p 2 , p 3The probability value can determine the level of the second surgical assistant to be configured for the current patient undergoing the current surgery. The fourth predicted probability distribution vector corresponds to anesthesiologists, z 3 ' =

p 0 ,p 1 ,p 2 ,p 3

p 0 ,p 1 ,p 2

[0090] Combined with the above examples, the elements of the predicted probability distribution vector correspond one-to-one with the qualification levels of medical staff positions. Among them, the first, second, third, and fourth predicted probability distribution vectors each contain 4 elements, and the qualification levels corresponding to these 4 elements are: resident physician, attending physician, deputy chief physician, and chief physician. Similarly, the fifth predicted probability distribution vector contains 3 elements, and the qualification levels corresponding to these 3 elements are: junior, intermediate, and senior.

[0091] In the model training stage, according to the differences between the qualification requirement information of multiple medical staff positions reflected by multiple predicted probability distribution vectors and the personnel qualification information of multiple medical staff positions in the medical team that has performed surgeries in the training data, the parameters of the model to be trained are adjusted, and iterative training is carried out until the training cut-off condition is met to obtain a medical team allocation model.

[0092] As introduced above, in the training data, the personnel qualification information of multiple medical staff positions in the medical team that has performed surgeries is provided, for example:

[0093] {"Surgeon": "Chief Physician",

[0094] "First Surgical Assistant": "Deputy Chief Physician",

[0095] "Second assistant in surgery": "Resident physician",

[0096] "Anesthesiologist": "Attending physician",

[0097] "Nurse": "Junior".

[0098] As an example, the predicted qualification requirement information is as follows:

[0099] {"Surgeon in charge": "Associate chief physician",

[0100] "First assistant in surgery": "Associate chief physician",

[0101] "Second assistant in surgery": "Resident physician",

[0102] "Anesthesiologist": "Resident physician",

[0103] "Nurse": "Junior".

[0104] If one or more of the five predicted probability distribution vectors output by the model are different from the corresponding personnel qualification information in the training data, it means that the current prediction situation does not match the actual situation, and the prediction performance of the model still needs to be improved. In this regard, it is necessary to continue to adjust the model parameters to optimize the performance of the model. For the training termination condition of the model, it can be set according to one or more of the performance indicators of the model, the number of iterative trainings, and the loss function value. Here, the training termination condition is not specifically limited.

[0105] Through the process introduced above, the model is trained, and finally the medical team allocation model required by the technical solution of this application is obtained. Furthermore, the scheduling element data obtained in S101 is used as the input of the medical team allocation model. The medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed according to the input scheduling element data. According to the above introduction, the output of the model is multiple predicted probability distribution vectors, and each predicted probability distribution vector can uniquely represent a qualification level of the corresponding medical staff position. The medical team allocation model precisely represents the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed through the predicted probability distribution vectors.

[0106] To automatically schedule the medical team for the surgery to be performed, the following specific team personnel matching is also carried out through step S103.

[0107] S103. Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, match medical staff for multiple medical staff positions in the medical team for the surgery to be performed, and generate a medical team scheduling plan for the surgery to be performed.

[0108] In the embodiments of the present application, the qualification information and position information of different doctors and nurses can be pre-connected. Thus, it is convenient to perform matching and comparison after obtaining the qualification requirement information of each medical staff position predicted by the model. For example, the qualification requirement information of 5 medical staff positions predicted by the medical team allocation model for the surgery to be performed is as follows:

[0109] {"Surgeon": "Chief Physician",

[0110] "First Assistant in Surgery": "Associate Chief Physician",

[0111] "Second Assistant in Surgery": "Attending Physician",

[0112] "Anesthesiologist": "Resident Physician",

[0113] "Nurse": "Junior"}.

[0114] Since the qualification requirement information of each medical staff position for the surgery to be performed has been determined, the information of medical staff with corresponding qualification levels can be locked for specific matching and generating a scheduling plan. It can be seen that the medical team allocation model saves the manual analysis time for the qualification requirements of medical staff and intelligently realizes the medical team scheduling process. Combining the embodiments introduced above, in the technical solution of the present application, in order to improve the scheduling efficiency of the medical team, first, the surgery information and patient information of the surgery to be performed are obtained as scheduling element data. Then, the scheduling element data is used as the input of the medical team allocation model, and the medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed according to the input scheduling element data. Since the medical team allocation model is a model trained based on the surgery information, patient information of the performed surgeries and the personnel qualification information of multiple medical staff positions in the medical teams of the performed surgeries, this model is capable of predicting the personnel qualifications that should be configured for each medical staff position in the medical team for performing the surgery according to the surgery information and patient information of the surgery. For example, it is predicted that the position of the surgeon should be configured with an attending physician, an associate chief physician or a chief physician. Finally, based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, medical staff are matched for multiple medical staff positions in the medical team for the surgery to be performed, and a medical team scheduling plan for the surgery to be performed is generated.

[0115] Since the model has predicted the qualification requirement information for the positions, and the qualifications of each medical staff are determined, convenient to query and index, the processing efficiency of the medical team scheduling is greatly improved, the scheduling time is saved, and the medical team scheduling for each operation to be performed can be carried out intelligently and automatically. The technical solution of this application improves the timeliness and effectiveness of the mobilization of medical resources in the face of major disasters, public health emergencies, emergency response and other scenarios, and can help to reasonably allocate medical resources and reduce the workload of medical staff in medical team scheduling.

[0116] Figure 2 This is the overall implementation flowchart of a medical team automatic scheduling method provided by an embodiment of this application. In Figure 2 it vividly presents the data input process and the process of model prediction output. And in Figure 2 it also shows the process of matching the qualification requirement information of multiple medical staff positions output by the model prediction with the position information and qualification information of multiple medical staff in the hospital, and finally generating the medical team scheduling plan for the operation to be performed. Figure 3 This is the prediction output schematic diagram of a medical team allocation model provided by an embodiment of this application. The model predicts the respective qualification levels of the surgeon, the first assistant in the operation, the second assistant in the operation, the anesthesiologist and the nurse, so that according to the predicted output content, the qualification requirements for 5 medical staff positions for the operation to be performed can be determined.

[0117] In the embodiment of this application, it is further proposed that the medical team scheduling plan can be determined or modified through an audit method. The following combines Figure 4 and Figure 5 to introduce two implementation processes.

[0118] Figure 4 This is another medical team automatic scheduling method provided by an embodiment of this application. As Figure 4 shown, the process of this method includes:

[0119] S401. Obtain the scheduling element data of the operation to be performed.

[0120] S402. Use the scheduling element data as the input of the medical team allocation model. The medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the operation to be performed according to the input scheduling element data.

[0121] Among them, the implementation manners of S401 and S402 are basically the same as those of S101 and S102 in the previous embodiment, and can be referred to the introduction of the previous embodiment, and will not be elaborated here.

[0122] Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model in S401 and S402, the process of matching medical staff for multiple medical staff positions within the medical team for the surgery to be performed and generating a scheduling plan for the medical team for the surgery to be performed is described in this embodiment through S403 to S409.

[0123] S403. Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, as well as the position information and qualification information of multiple medical staff in the hospital, match medical staff for multiple medical staff positions within the medical team for the surgery to be performed, and generate an initial scheduling plan for the medical team for the surgery to be performed.

[0124] In this embodiment, the initial scheduling plan for the medical team may be a scheduling plan initially generated for the surgery to be performed based on the matching of position information and the matching of qualification information. Since the prediction of qualification requirement information is achieved through the medical team allocation model, and the positions of each medical staff in the hospital are known and their qualifications are determined, the matching of position information and qualification information can be realized through an automated method, thereby determining the personnel information of the medical team positions and generating an initial scheduling plan for the medical team.

[0125] Considering that there may still be some uncontrollable factors in the initial scheduling plan for the medical team, such as changes in the duty dates of medical staff, or medical staff being unable to cooperate with the scheduled surgery due to other surgeries or outpatient reasons, or shortages in the resources of medical staff with the predicted qualification levels, or the qualifications of the medical staff predicted for the surgery to be performed being too high and slightly lower qualifications being able to complete the surgery, etc. For a more reasonable and accurate medical team scheduling, it is proposed in this embodiment that a scheduling plan review interface can be presented to the user through a front-end interaction method. The user (such as the medical staff responsible for review) can view the initial scheduling plan for the medical team for the surgery to be performed in the scheduling plan review interface, and then conveniently interact in this interface to achieve the review.

[0126] S404. On the scheduling plan review interface, display the initial scheduling plan for the medical team for the surgery to be performed.

[0127] In a possible implementation manner, controls supporting the user to perform review confirmation and controls supporting the user to modify the scheduling are displayed on the scheduling plan review interface. By triggering the relevant controls, the user can express their intention. By detecting the situation where the controls are triggered, it can be determined whether the user's operation intention is to review and confirm or modify the scheduling.

[0128] In addition to the above-mentioned controls, a schedule plan display area for displaying the schedule plan may also be included on the schedule plan review interface. By browsing the content in the schedule plan display area, the user can learn about the current schedule plan for the surgery to be performed. For example, after forming the initial schedule plan for the medical team, the content of the plan is displayed in the above-mentioned schedule plan display area.

[0129] In addition, a duty information display area may also be included on the schedule plan review interface. This area is used to display the duty dates of medical staff in the hospital within a certain period in the future. As an example, the duty information display area shows the specific duty dates and qualification levels of the medical staff on duty in the next two weeks.

[0130] S405. Determine whether to review and confirm the initial schedule plan of the medical team. If yes, go to S406; if no, go to S407.

[0131] In specific implementation, it can be determined whether the user reviews and confirms the current initial schedule plan of the medical team by detecting the triggering situation of the review confirmation control on the schedule plan review interface. If the control is triggered, it means that the user has reviewed and confirmed the plan, and then go to S406. On the contrary, if the control is not triggered, further detect the triggering situation of the schedule modification control on the schedule plan review interface. If the schedule modification control is triggered, go to S407.

[0132] S406. In response to the review and confirmation operation of the initial schedule plan of the medical team for the surgery to be performed, use the initial schedule plan of the medical team for the surgery to be performed as the final schedule plan of the medical team for the surgery to be performed.

[0133] S407. In response to the schedule modification operation of the initial schedule plan of the medical team for the surgery to be performed, redisplay the modified schedule plan of the medical team for the surgery to be performed on the schedule plan review interface.

[0134] In practical applications, the form of the schedule modification control can be various. For example, it can be in the form of a drop-down menu or a text editing box. When the user triggers the schedule modification control, it means that the current schedule plan needs to be adjusted. When the user finishes triggering the schedule modification control, it means that this modification is completed. The modified plan can be redisplayed on the schedule plan review interface. For example, in the schedule plan display area exemplified above, display the modified schedule plan of the medical team for the surgery to be performed to facilitate the user to view the modified effect in a timely manner.

[0135] S408. Determine whether to review and confirm the modified schedule plan of the medical team. If yes, go to S409.

[0136] After the user finishes the modification, the embodiments of the present application can continue to detect the triggering situation of the review confirmation control to determine whether the user reviews and confirms the medical team scheduling modification plan currently displayed. If the review confirmation control is triggered after the user modification, it indicates that the user has reviewed and confirmed the plan. Refer to S409.

[0137] S409. In response to the review confirmation operation on the medical team scheduling modification plan for the surgery to be performed, use the medical team scheduling modification plan for the surgery to be performed as the final medical team scheduling plan for the surgery to be performed.

[0138] Through the above steps S403 - S409, the review of the medical team scheduling plan for the surgery to be performed on the scheduling plan review interface is completed. Through human-computer interaction, while simplifying the scheduling work, the requirement for flexible adjustment of the scheduling plan is also met.

[0139] Figure 5 Another medical team automatic scheduling method provided by the embodiments of the present application. As Figure 5 shown, the process of this method includes:

[0140] S501. Obtain the scheduling element data of the surgery to be performed.

[0141] S502. Use the scheduling element data as the input of the medical team allocation model. The medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed according to the input scheduling element data.

[0142] Among them, the implementation manners of S501 and S502 are basically the same as those of S101 and S102 in the previous embodiments. The introduction in the previous embodiments can be referred to and will not be elaborated here.

[0143] Based on S501 and S502, on the basis of the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, the process of matching medical staff for multiple medical staff positions in the medical team for the surgery to be performed to generate the medical team scheduling plan for the surgery to be performed is described through S503 - S506 in this embodiment.

[0144] S503. In response to the scheduling trigger operation for the surgery to be performed on the scheduling plan review interface, display the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model.

[0145] Figure 6A A schematic diagram of a scheduling plan review interface provided by the embodiments of the present application. As Figure 6A shown, the upper left area is the information display area for the reviewing doctor, which exemplarily displays the avatar, department, hospital area, and medical insurance doctor code of the reviewing doctor on this system interface.Figure 6A The lower left area of is the schedule display area. Users can select to create a new schedule in this area, or display the annual calendar or monthly calendar after selecting the year and month. Thus, users can understand the relationship between dates and weeks through this schedule display area. Figure 6A The upper right of is the display area for surgeries awaiting review. In Figure 6A 's example, surgeries awaiting review at level 1 and level 2 are combined for display, and surgeries awaiting review at level 3 and level 4 are combined for display. In this example, there are a total of 9 surgeries at level 1 and level 2 awaiting review, corresponding to 9 patients; there are a total of 4 surgeries at level 3 and level 4 awaiting review, corresponding to 4 patients.

[0146] Exemplarily, Figure 6A shows the cards of 4 patients related to surgeries at level 3 and level 4, such as the 4 cards of "Example Patient 1", "Example Patient 2", "Example Patient 3", and "Example Patient 4" in the figure. Exemplarily, Figure 6A also shows the cards of 4 patients related to surgeries at level 1 and level 2, such as the 4 cards of "Example Patient 5", "Example Patient 6", "Example Patient 7", and "Example Patient 8" in the figure. If a user needs to know the information of other patients or surgeries at level 1 or level 2 that are not shown, they can trigger the card with the words "Level 1 / Level 2 Surgeries Awaiting Review". Immediately, more cards of patients undergoing surgeries at level 1 or level 2 will be displayed on the interface.

[0147] In Figure 6A the lower right is the editing area, which can display the prediction results of the qualification requirement information of the medical team for each surgery to be performed. The scheduling trigger operation for surgeries to be performed can be a click operation on the patient card. Furthermore, the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model is displayed in the editing area. Users can select and match medical team members through this area. Also, users can modify the medical team scheduling plan through the trigger operation in this area.

[0148] Figure 6B is a schematic diagram of the editing area in the scheduling plan review interface. In Figure 6BIn the shown editing area, patient information is displayed in the upper left corner. As shown in the figure, the patient information is: Example Patient 5, male, 48 years old. Information such as the name of the operation, the operation level, the surgeon in charge, the first assistant of the operation, the second assistant of the operation, the anesthesiologist, the nurse, and the estimated duration is displayed in the lower left corner. In the example of this figure, the medical team allocation model predicts that the qualification level of the surgeon in charge is deputy chief physician, the qualification level of the first assistant of the operation is attending physician, the qualification level of the second assistant of the operation is resident physician, the qualification level of the anesthesiologist is attending physician, and the qualification level of the nurse is intermediate. In the predicted qualification requirement information, a control named "Select" is displayed on the right side of each element. In this application, it is referred to as a selection control. The user can trigger a certain selection control in this editing area and then trigger the medical staff card on the right to achieve the selection and matching of specific personnel for specific positions. See the introductions of S504 and S505 below for details.

[0149] S504. In response to the triggering operation of the personnel selection control corresponding to the qualification requirement information of the medical staff position, display the scheduling information of the medical staff who meet the corresponding qualification requirement information.

[0150] In specific implementation, since the triggered selection control corresponds one-to-one to the predicted qualification requirement information of the specific medical staff position, when the user triggers the selection control, the scheduling information of the medical staff who meet the qualification requirement information is immediately displayed on the right.

[0151] S505. In response to the triggering operation of the medical staff card in the scheduling information, construct the matching relationship between the medical staff position and the medical staff.

[0152] For example, when the user triggers the selection control on the right side of the deputy chief physician, the scheduling information in Figure 6B is displayed. If the user triggers the medical staff card, it means that the medical staff corresponding to the card is matched with the position of the surgeon in charge of the cystoscopy to be performed on Example Patient 5. Similarly, if the user triggers the selection control on the right side of the qualification requirement information of the first assistant of the operation, the scheduling information of each medical staff who meets the qualification requirement information of "attending physician" can also be displayed in the editing area, facilitating the user to match specific medical staff for the position of the first assistant of the operation. The selection and matching of other medical staff positions are similar and will not be elaborated here.

[0153] S506. Wait until medical staff have been matched for multiple medical staff positions, and generate a scheduling plan for the medical team of the operation to be performed.

[0154] When personnel matching has been completed for all medical staff positions in the medical team of this operation, a scheduling plan for the medical team of the operation can be generated. In the scheduling plan, there is not only information about specific positions, qualifications, and medical staff, but also through Figure 6A and Figure 6BThe shown interface also facilitates comparing the scheduling dates of each medical staff. Such a visual display is friendly and user-friendly for users, saving time and effort.

[0155] In some other possible implementation manners, not only can the user achieve personnel matching through the interaction in the scheduling plan review interface, but also the user can modify the qualification requirement information of one or more predicted medical staff positions. After presenting the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, the automatic medical team scheduling method may further include: in response to a triggering operation on a review and modification control corresponding to the qualification requirement information of a medical staff position, presenting one or more alternative qualification requirement information of this medical staff position; in response to a selection triggering operation on one piece of alternative qualification requirement information of this medical staff position, replacing the presented qualification requirement information of this medical staff position with the selected alternative qualification requirement information. For example, changing the qualification requirement information of the surgeon from "chief physician" to "associate chief physician". Through the modification, the correction of the team configuration is achieved.

[0156] In order to further deeply apply the review results to the latest prediction, in the embodiments of the present application, it is also proposed that the automatic medical team scheduling method may further include: optimizing the medical team allocation model based on the reviewed and modified medical team scheduling plan. In this way, it is equivalent to correcting the labels in the training data. This implementation manner of further optimizing the medical team allocation model by using the reviewed and modified medical team scheduling plan can improve the problem of inaccurate prediction, enhance the utility of the trained medical team allocation model, and further reduce the human burden.

[0157] In the present application, through innovative user interface design, smooth interaction logic, and efficient feedback collection mechanism, a good user experience is provided for medical staff, and the continuous progress of the system is driven. The interface reduces unnecessary input and page switching operations. After the transmitted patients are intelligently queued in the background, the queuing data is clearly listed on the front-end interface for doctors to review or modify. Once the doctor modifies and clicks to save, the device records the scheduled data of the current patient after manual optimization and feeds it back to the model side for optimization and learning. The scheduling plan review interface in the present application has the following multiple advantages:

[0158] (1) Intuitiveness and ease of use: Adopting a simple and clear interface layout, placing key functions in prominent positions to ensure that doctors can quickly find the required operations. At the same time, through details such as color matching, icon design, and font selection, the beauty and readability of the interface are improved, and visual fatigue is reduced.

[0159] (2) Personalized customization: Providing customizable user interface options, allowing doctors to set view modes, data display methods, etc. according to personal preferences to meet the working habits and needs of different doctors.

[0160] (3) Interactive guidance: Clear interactive guidance, such as floating tips, step-by-step instructions, etc., is set at key operation links to help doctors get started quickly and reduce misoperations.

[0161] In the embodiments of the present application, it is also proposed that in an alternative implementation, after determining the final medical team scheduling plan for the operation to be performed, the medical team automatic scheduling method further includes: sending a notification message of the operation to be performed to the medical staff related to the final medical team scheduling plan. Among them, the notification message includes medical team personnel information, patient information, and operation time information. The form of the notification message can be a text message, a mini-program, or a message on a mobile client, etc. The form of the message is not limited here.

[0162] By means of notification after automatic scheduling, the communication efficiency of the scheduling plan is improved, enabling the medical staff within the medical team to obtain operation-related information in a timely manner. Thus, time arrangements can be made more effectively and reasonably. Especially in the case of schedule modification, timely notification conveys the operation information and avoids delaying the pre-operation communication and operation arrangement of the team.

[0163] In the past, for urgent or sudden medical events, due to the lack of effective intelligent tools, there was a problem of uneven distribution of key resources such as medical staff in mobile cabin hospitals in case of emergency. Some patients in urgent need of surgery may wait for a long time due to resource shortage, while some medical staff may be idle, resulting in waste of resources. In case of emergency, medical staff need to bear huge work pressure and psychological burden. The traditional manual operation scheduling method not only increases the workload of medical staff, but also may lead to work mistakes and patient dissatisfaction due to decision-making errors or poor communication. After the model training and optimization are completed, by accessing different physician qualification information in the mobile cabin hospital in the early stage, information matching and comparison are completed, a comparison table is set up and relevant docking interfaces are set. Once an emergency occurs, the on-site dispatcher inputs the patient's injury condition, the level of the operation to be performed, and the operation name, and the device outputs the qualification requirements of the team members required for the current operation, and further matches the optimal scheduling plan according to the information comparison table and feedbacks it to the on-site dispatcher for review. After the review is completed, a notification is sent to the mobile phones of the medical team members to guide the medical staff and the wounded to the corresponding best operation positions, ensuring timely preparation for the operation, shortening the handover and interaction time, and completing the rescue as quickly as possible to the greatest extent. The technical solution provided by the embodiments of the present application can automatically implement the medical team scheduling, improve the treatment efficiency and resource management ability of the mobile cabin hospital in case of emergency, and at the same time support the review and modification of the scheduling plan to ensure that the scheduling plan can operate reasonably and quickly.

[0164] The technical solution of the present application has multiple technological innovation breakthroughs. Specifically:

[0165] 1) Innovative Application of Deep Learning Algorithm in Surgical Scheduling:

[0166] In this application, for the first time, a deep learning algorithm is applied to the field of surgical scheduling. By training a model to learn the rules and patterns of surgical scheduling from historical data, accurate prediction of surgical requirements and optimal allocation of the medical staff team are achieved. This innovative application not only improves the accuracy and efficiency of scheduling, but also opens up a new way for the intelligent development of surgical scheduling, and plays a great advantage in relevant special scenarios.

[0167] 2) Multi-dimensional Information Fusion and Comprehensive Analysis:

[0168] During the scheduling process, not only surgical requirements and the availability of the medical team are considered, but also multi-dimensional information such as the basic conditions of patients and anesthesia methods is integrated. By comprehensively analyzing this information, a more comprehensive and reasonable scheduling plan can be generated, improving the overall utilization efficiency of medical resources and the treatment effect.

[0169] 3) Construction of an Intelligent Decision Support System:

[0170] Automatically complete the decision-making process of surgical scheduling, reducing the complexity and uncertainty of manual decision-making. Through intelligent decision support, it can quickly respond to emergencies, provide more accurate and timely scheduling information for medical staff, and improve the level and quality of medical services.

[0171] 4) Instant Medical Team Resource Scheduling, Review and Adjustment, and Real-time Notification Technology:

[0172] Based on cloud computing and high-speed data exchange technology, the scheduling of the surgical medical team is realized instantly. With the help of this technology, on-site schedulers can obtain the latest scheduling results of specific patients in real time, without the need for on-site manual docking with doctors from various cities, hospitals, and departments to match whether they can support relevant surgeries. At the same time, the solution supports the functions of scheduling result review, adjustment, and real-time notification. Once a surgery application and scheduling result are generated, doctors and on-site schedulers involved are immediately notified by means of text messages, APP push, etc., ensuring the immediacy and accuracy of information.

[0173] Based on the above introduction, the technical solution of this application has outstanding technical advantages:

[0174] 1. Significantly improve scheduling efficiency and accuracy: Through deep learning technology, it is possible to automatically generate an optimal surgical medical team scheduling plan by analyzing surgical historical data and real-time patient injuries and resource information. This not only greatly shortens the time required for scheduling, but also improves the accuracy and rationality of scheduling, ensuring the rapid and effective scheduling of medical resources in case of emergencies.

[0175] 2. Optimize resource allocation and reduce resource waste: It can adjust the schedule of the surgical physician team according to the actual situation, effectively avoiding the uneven distribution of key resources such as medical staff. This ensures that patients in urgent need of surgery can receive treatment promptly, while reducing the idle time of medical staff, significantly improving resource utilization rate, and reducing resource waste.

[0176] 3. Reduce the workload of medical staff: Through automated scheduling, it undertakes the cumbersome information collection, analysis, and decision-making work in traditional manual scheduling, thus greatly reducing the workload of medical staff. Medical staff can focus more on the treatment of patients, improving work efficiency and medical quality.

[0177] 4. Improve the efficiency of emergency treatment: In emergency situations such as major disasters and public health emergencies, the device can respond quickly, generate and adjust the scheduling plan of the surgical medical team rapidly, ensuring that the mobile cabin hospital can quickly mobilize medical resources to meet the rapidly increasing patient needs, especially surgical needs. This significantly improves the efficiency of emergency treatment and buys precious treatment time for patients.

[0178] 5. Enhance the resilience and response ability of the medical system: It not only improves the treatment efficiency and resource management ability of the mobile cabin hospital in emergency situations, but also enhances the resilience and response ability of the entire medical system in dealing with major disasters, public health emergencies, etc. This helps to build a safer, more efficient, and reliable medical system to ensure the life safety and physical health of the people.

[0179] Based on the medical team automatic scheduling method provided in the foregoing embodiments, correspondingly, the present application also provides a medical team automatic scheduling device. The implementation of this device will be described below with reference to the accompanying drawings.

[0180] Figure 7 It is a schematic structural diagram of the medical team automatic scheduling device. As Figure 7 shown, the device includes:

[0181] A data acquisition module 71, configured to acquire scheduling element data of the surgery to be performed; the scheduling element data includes surgery information and patient information;

[0182] A qualification requirement prediction module 72, configured to use the scheduling element data as the input of the medical team allocation model, and the medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team for the surgery to be performed according to the input scheduling element data; wherein, the medical team allocation model is a model trained based on the surgery information, patient information of the surgeries that have been performed, and the personnel qualification information of multiple medical staff positions in the medical teams of the surgeries that have been performed;

[0183] The personnel matching module 73 is used to match medical staff for multiple medical staff positions within the medical team for the to-be-performed surgery based on the qualification requirement information of the multiple medical staff positions predicted by the medical team allocation model, and generate a scheduling plan for the medical team of the to-be-performed surgery.

[0184] Optionally, the personnel matching module 73 includes:

[0185] An initial plan generation unit is used to match medical staff for multiple medical staff positions within the medical team for the to-be-performed surgery based on the qualification requirement information of the multiple medical staff positions predicted by the medical team allocation model, as well as the position information and qualification information of multiple medical staff within the hospital, and generate an initial scheduling plan for the medical team of the to-be-performed surgery;

[0186] A plan display and review unit is used to display the initial scheduling plan for the medical team of the to-be-performed surgery on the scheduling plan review interface; in response to the review confirmation operation on the initial scheduling plan for the medical team of the to-be-performed surgery, use the initial scheduling plan for the medical team of the to-be-performed surgery as the final scheduling plan for the medical team of the to-be-performed surgery;

[0187] Alternatively, the plan display and review unit is used to, in response to a scheduling modification operation on the initial scheduling plan for the medical team of the to-be-performed surgery, re-display the modified scheduling plan for the medical team of the to-be-performed surgery on the scheduling plan review interface; in response to the review confirmation operation on the modified scheduling plan for the medical team of the to-be-performed surgery, use the modified scheduling plan for the medical team of the to-be-performed surgery as the final scheduling plan for the medical team of the to-be-performed surgery.

[0188] Optionally, the personnel matching module 73 includes:

[0189] A qualification requirement display unit is used to display the qualification requirement information of the multiple medical staff positions predicted by the medical team allocation model in response to a scheduling trigger operation on the to-be-performed surgery on the scheduling plan review interface;

[0190] A scheduling information display unit is used to display the scheduling information of medical staff who meet the corresponding qualification requirement information in response to a trigger operation on the personnel selection control corresponding to the qualification requirement information of the medical staff position;

[0191] A personnel matching unit is used to construct a matching relationship between the medical staff position and the medical staff in response to a trigger operation on the medical staff card in the scheduling information;

[0192] A plan generation unit is used to generate a scheduling plan for the medical team of the to-be-performed surgery when all the multiple medical staff positions have been matched with medical staff.

[0193] Optionally, the medical team automatic scheduling device further includes:

[0194] A to-be-selected information display unit, configured to display one or more to-be-selected qualification requirement information of the medical staff position in response to a trigger operation on an audit modification control corresponding to the qualification requirement information of the medical staff position;

[0195] An information replacement display unit, configured to replace and display the selected to-be-selected qualification requirement information into the qualification requirement information of the medical staff position in response to a selection trigger operation on one piece of to-be-selected qualification requirement information of the medical staff position.

[0196] Optionally, the medical team automatic scheduling device further includes: a model training module 74; the model training module 74 is configured to train a medical team allocation model in the following manner:

[0197] Collect training data for the to-be-trained model; the training data includes surgical information, patient information of surgeries that have been performed in the hospital, and personnel qualification information of multiple medical staff positions in the medical team for the surgeries that have been performed; the to-be-trained model includes an input layer, a double-layer LSTM layer, a fully connected layer, and an output layer;

[0198] The input layer extracts features from the surgical information and patient information in the training data, generates a one-dimensional feature vector, and transmits it to the double-layer LSTM layer;

[0199] The double-layer LSTM layer updates the state by using internal memory units and gating mechanisms, and outputs a hidden state vector;

[0200] The fully connected layer performs a linear transformation on the hidden state vector, and transmits the transformed vector to the output layer;

[0201] The output layer outputs multiple predicted probability distribution vectors according to the transformed vector; each predicted probability distribution vector corresponds to a medical staff position; the elements of the predicted probability distribution vector correspond one-to-one to the qualification levels of the medical staff position;

[0202] According to the differences between the qualification requirement information of multiple medical staff positions reflected by the multiple predicted probability distribution vectors and the personnel qualification information of multiple medical staff positions in the medical team for the surgeries that have been performed in the training data, adjust the parameters of the to-be-trained model, and iteratively train until the training cutoff condition is met to obtain the medical team allocation model.

[0203] Optionally, the multiple medical staff positions include: the surgeon, the first assistant in the surgery, the second assistant in the surgery, the anesthesiologist, and the nurse;

[0204] The multiple predicted probability distribution vectors include: a first predicted probability distribution vector corresponding to the operating surgeon, a second predicted probability distribution vector corresponding to the first assistant in the operation, a third predicted probability distribution vector corresponding to the second assistant in the operation, a fourth predicted probability distribution vector corresponding to the anesthesiologist, and a fifth predicted probability distribution vector corresponding to the nurse;

[0205] Among them, the first predicted probability distribution vector, the second predicted probability distribution vector, the third predicted probability distribution vector, and the fourth predicted probability distribution vector each contain 4 elements, and the fifth predicted probability distribution vector contains 3 elements; the qualification levels corresponding to the 4 elements are: resident physician, attending physician, deputy chief physician, and chief physician; the qualification levels corresponding to the 3 elements are: junior, intermediate, and senior.

[0206] Optionally, the medical team automatic scheduling device further includes:

[0207] A data preprocessing module 75 for preprocessing the training data;

[0208] The model training module 74 is specifically configured to: extract features from the operation information and patient information in the preprocessed training data by the input layer;

[0209] The preprocessing method includes one or more of the following:

[0210] Duplicate removal, format unification, error correction, or missing value processing.

[0211] Optionally, the medical team automatic scheduling device further includes:

[0212] A notification module 76 for sending a notification message of the operation to be performed to the medical staff related to the final medical team scheduling plan of the operation to be performed after determining the final medical team scheduling plan of the operation to be performed, where the notification message includes medical team personnel information, patient information, and operation time information.

[0213] Based on the medical team automatic scheduling method and the medical team automatic scheduling device introduced in the foregoing embodiments, correspondingly, the present application further provides a medical team automatic scheduling system. Figure 8 It is a schematic structural diagram of a medical team automatic scheduling system. As Figure 8 shown, the system includes: a scheduling device 81 and a display device 82; the scheduling device 81 is communicatively connected to the display device 82; the scheduling device 81 includes a processor 811 and a memory 812;

[0214] The memory 812 is used to store a computer program;

[0215] The processor 811 is configured to run the computer program. When the computer program runs, it executes the steps of the medical team automatic scheduling method introduced in the method embodiment as described above, and controls the display device 82 to display the generated medical team scheduling plan for the surgery to be performed.

[0216] It should be noted that the various embodiments in this specification are described in a progressive manner. For the same or similar parts among the various embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, they are described relatively simply. For the relevant parts, reference can be made to the partial description of the method embodiments. The device and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components indicated as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative effort.

[0217] As described above, this is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for automatic scheduling of a medical team, characterized in that: include: Acquiring scheduling element data of a surgery to be performed; the scheduling element data includes surgery information and patient information; The scheduling element data is used as an input of a medical team allocation model, and the medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team to perform the operation based on the input scheduling element data; wherein the medical team allocation model is a model trained based on the operation information of the performed operation, the patient information and the personnel qualification information of multiple medical staff positions in the medical team that has performed the operation; Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, medical staff are matched to multiple medical staff positions in the medical team to be operated on, and a scheduling plan for the medical team to be operated on is generated.

2. The method according to claim 1, characterized in that The qualification requirement information of multiple medical staff positions predicted by the medical team allocation model is used to match medical staff to multiple medical staff positions in the medical team to be operated on, and generate a scheduling plan for the medical team to be operated on, including: Based on the qualification requirement information of multiple medical staff positions predicted by the medical team allocation model, as well as the position information and qualification information of multiple medical staff in the hospital, medical staff are matched to multiple medical staff positions in the medical team to be operated on, and an initial scheduling plan for the medical team to be operated on is generated; On the scheduling plan review interface, the initial scheduling plan of the medical team for the operation to be performed is displayed; in response to the review and confirmation operation of the initial scheduling plan of the medical team for the operation to be performed, the initial scheduling plan of the medical team for the operation to be performed is used as the final scheduling plan of the medical team for the operation to be performed; Alternatively, in response to a scheduling modification operation on the initial scheduling plan of the medical team for the surgery to be performed, the modified scheduling plan of the medical team for the surgery to be performed is re-displayed on the scheduling plan review interface; in response to a review and confirmation operation on the modified scheduling plan of the medical team for the surgery to be performed, the modified scheduling plan of the medical team for the surgery to be performed is used as the final scheduling plan of the medical team for the surgery to be performed.

3. The method according to claim 1, characterized in that The qualification requirement information of multiple medical staff positions predicted by the medical team allocation model is used to match medical staff to multiple medical staff positions in the medical team to be operated on, and generate a scheduling plan for the medical team to be operated on, including: In response to a scheduling trigger operation on the to-be-performed surgery on the scheduling plan review interface, displaying qualification requirement information for multiple medical staff positions predicted by the medical team allocation model; In response to a triggering operation of a personnel selection control corresponding to qualification requirement information of a medical staff position, displaying the scheduling information of medical staff meeting the corresponding qualification requirement information; In response to the triggering operation of the medical staff card in the scheduling information, a matching relationship between the medical staff position and the medical staff is established; After the plurality of medical staff positions have been matched with medical staff, a scheduling plan for the medical team to perform the operation is generated.

4. The method according to claim 3, characterized in that After displaying the qualification requirement information of the plurality of medical staff positions predicted by the medical team allocation model, the method further includes: In response to a triggering operation of a review and modification control corresponding to qualification requirement information of a medical staff position, one or more candidate qualification requirement information of the medical staff position is displayed; In response to a selection triggering operation on a type of candidate qualification requirement information for the medical staff position, the selected candidate qualification requirement information is replaced and displayed in the qualification requirement information for the medical staff position.

5. The method according to claim 1, characterized in that The training steps of the medical team allocation model include: Collect training data for the model to be trained; the training data includes surgical information of surgeries that have been performed in the hospital, patient information, and qualification information of multiple medical staff positions in the medical team that has performed the surgery; the model to be trained includes an input layer, a double-layer LSTM layer, a fully connected layer, and an output layer; The input layer extracts features from the surgical information and patient information in the training data, generates a one-dimensional feature vector and passes it to the double-layer LSTM layer; The double-layer LSTM layer updates the state using the internal memory unit and gating mechanism and outputs a hidden state vector; The fully connected layer performs a linear transformation on the hidden state vector and transmits the transformed vector to the output layer; The output layer outputs a plurality of predicted probability distribution vectors according to the transformed vector; each predicted probability distribution vector corresponds to a medical staff position; the elements of the predicted probability distribution vector correspond one to one with the qualification level of the medical staff position; According to the difference between the qualification requirement information of multiple medical staff positions reflected by multiple predicted probability distribution vectors and the personnel qualification information of multiple medical staff positions in the medical team that has performed the operation in the training data, the parameters of the model to be trained are adjusted, and the training is iterated until the training cutoff condition is met to obtain the medical team allocation model.

6. The method according to claim 5, characterized in that The multiple medical staff positions include: chief surgeon, first assistant surgeon, second assistant surgeon, anesthesiologist and nurse; The multiple prediction probability distribution vectors include: a first prediction probability distribution vector corresponding to the surgeon, a second prediction probability distribution vector corresponding to the first assistant surgeon, a third prediction probability distribution vector corresponding to the second assistant surgeon, a fourth prediction probability distribution vector corresponding to the anesthesiologist, and a fifth prediction probability distribution vector corresponding to the nurse; Among them, the first prediction probability distribution vector, the second prediction probability distribution vector, the third prediction probability distribution vector and the fourth prediction probability distribution vector each contain 4 elements, and the fifth prediction probability distribution vector contains 3 elements; the qualification levels corresponding to the 4 elements are: resident physician, attending physician, associate chief physician and chief physician; the qualification levels corresponding to the 3 elements are: junior, intermediate and senior.

7. The method according to claim 5 or 6, characterized in that: After collecting the training data of the model to be trained, the method further includes: Preprocessing the training data; The input layer extracts features from the surgical information and patient information in the training data, specifically: the input layer extracts features from the surgical information and patient information in the preprocessed training data; The pre-processing method includes one or more of the following: Deduplication, format unification, error correction or missing value processing.

8. The method according to claim 2 or 3, characterized in that: After determining the final scheduling plan for the medical team to perform the surgery, the method further includes: A notification message of the operation to be performed is sent to the medical staff related to the final scheduling plan of the medical team, and the notification message includes the medical team personnel information, the patient information and the operation time information.

9. A medical team automatic scheduling device, characterized in that: include: A data acquisition module, used to acquire scheduling element data of a surgery to be performed; the scheduling element data includes surgery information and patient information; A qualification requirement prediction module, used to use the scheduling element data as an input of a medical team allocation model, and the medical team allocation model predicts the qualification requirement information of multiple medical staff positions in the medical team to be performed according to the input scheduling element data; wherein the medical team allocation model is a model trained based on the operation information of the performed operation, the patient information and the personnel qualification information of multiple medical staff positions in the medical team that has performed the operation; The personnel matching module is used to match medical personnel for multiple medical personnel positions within the medical team to be operated on based on the qualification requirement information of multiple medical personnel positions predicted by the medical team allocation model, and generate a scheduling plan for the medical team to be operated on.

10. A medical team automatic scheduling system, characterized in that: include: A scheduling device and a display device; the scheduling device is communicatively connected with the display device; the scheduling device includes a processor and a memory; The memory is used to store computer programs; The processor is used to run the computer program, which, when running, executes the steps of the medical team automatic scheduling method as described in any one of claims 1 to 8, and controls the display device to display the generated medical team scheduling plan for the surgery to be performed.

Images