CN114283929A

CN114283929A - Medical task allocation method and device, computer equipment and medium

Info

Publication number: CN114283929A
Application number: CN202111146350.5A
Authority: CN
Inventors: 刘梓轩; 申田
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2021-09-28
Filing date: 2021-09-28
Publication date: 2022-04-05

Abstract

The application discloses a medical task allocation method, a medical task allocation device, computer equipment and a medium, and relates to the technical field of intelligent scheduling. The method comprises the following steps: determining a target medical role corresponding to the target medical task from candidate medical roles based on a target task capability portrait corresponding to the target medical task, wherein the target task capability portrait is used for representing medical task execution capabilities required for executing the target medical task, and different candidate medical roles correspond to different medical task execution capabilities; and determining a target doctor object from at least one candidate doctor object corresponding to the target medical role based on the doctor allocation rule, and allocating the target medical task to the target doctor object. Therefore, each medical task to be distributed does not need to be distributed manually, the target medical task can be intelligently analyzed by the computer equipment, the target medical objects corresponding to each medical task are further intelligently matched, and the distribution efficiency of the medical tasks is improved.

Description

Medical task allocation method and device, computer equipment and medium

Technical Field

The embodiment of the application relates to the technical field of intelligent scheduling, in particular to a medical task allocation method, a medical task allocation device, computer equipment and a medium.

Background

In the working scene of a hospital, there are various medical tasks, such as a ward round task, a work in duty and a visit task, an operation task, a teaching task, a research task, etc., and various medical care roles that undertake different tasks, such as an attending doctor, a nurse, an anesthesiologist, etc.

In the related art, in order to maintain the working scene of the hospital, a manual scheduling mode is generally adopted, that is, a scheduling manager manually allocates each medical task to each doctor one by one according to the task sequence, so that the doctors can work according to the scheduling result.

Obviously, by adopting the manual scheduling mode, if the task amount is large, the scheduling efficiency is low.

Disclosure of Invention

The embodiment of the application provides a medical task allocation method, a medical task allocation device, computer equipment and a medium. The technical scheme is as follows:

according to an aspect of the present application, there is provided a medical task assigning method, the method comprising:

determining a target medical role corresponding to a target medical task from candidate medical roles based on a target task capability portrait corresponding to the target medical task, wherein the target task capability portrait is used for representing medical task execution capabilities required for executing the target medical task, and different candidate medical roles correspond to different medical task execution capabilities;

determining a target doctor object from at least one candidate doctor object corresponding to the target medical role based on a doctor allocation rule, wherein the doctor allocation rule at least comprises at least one of a doctor state rule and a historical allocation rule;

assigning the target medical task to the target physician subject.

According to another aspect of the present application, there is provided a medical task assigning apparatus, the apparatus comprising:

the system comprises a first determination module, a second determination module and a third determination module, wherein the first determination module is used for determining a target medical role corresponding to a target medical task from candidate medical roles based on a target task capability portrait corresponding to the target medical task, the target task capability portrait is used for representing medical task execution capability required by the target medical task, and different candidate medical roles correspond to different medical task execution capabilities;

a second determining module, configured to determine a target doctor object from at least one candidate doctor object corresponding to the target medical role based on a doctor allocation rule, where the doctor allocation rule at least includes at least one of a doctor state rule and a historical allocation rule;

an assignment module to assign the target medical task to the target physician object.

According to another aspect of the application, a computer device is provided, comprising a processor and a memory, in which at least one program is stored, which is loaded and executed by the processor to implement the medical task assigning method as described above.

According to another aspect of the application, a computer-readable storage medium is provided, in which at least one program is stored, which is loaded and executed by a processor to implement the medical task assigning method as described above.

According to another aspect of the application, a computer program product or computer program is provided, comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the medical task assigning method provided in the above-mentioned alternative implementation.

The beneficial effects brought by the technical scheme provided by the embodiment of the application at least comprise:

in a shift scheduling scene of a hospital, a target task capability portrait corresponding to a medical task to be distributed (target medical task) is analyzed so as to find a target medical role capable of executing the target medical task from candidate medical roles, and then an allocation object (target doctor object) of the target medical task is determined from a plurality of candidate doctor objects corresponding to the medical role based on a doctor allocation rule, so that each medical task to be distributed does not need to be allocated manually, the target medical task can be intelligently analyzed by computer equipment, the target medical object corresponding to each medical task is further intelligently matched, and the allocation efficiency of the medical task is improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 illustrates a schematic diagram of an implementation environment provided by an exemplary embodiment of the present application;

FIG. 2 illustrates a flow chart of a medical task assignment method provided by an exemplary embodiment of the present application;

FIG. 3 illustrates a flow chart of a medical task assignment method provided by another exemplary embodiment of the present application;

FIG. 4 illustrates a flow chart of a medical task assignment method provided by another exemplary embodiment of the present application;

FIG. 5 illustrates a flow chart of a method of training a task assignment model in accordance with an exemplary embodiment of the present application;

FIG. 6 shows a flow chart of a medical task assignment method provided by another exemplary embodiment of the present application;

FIG. 7 illustrates a schematic diagram of a process of training and applying a task assignment model according to an exemplary embodiment of the present application;

FIG. 8 illustrates a system architecture diagram of an intelligent shift scheduling system, shown in an exemplary embodiment of the present application;

FIG. 9 illustrates a process diagram of medical task assignment shown in an exemplary embodiment of the present application;

FIG. 10 is a block diagram of a medical task assigning apparatus provided in an exemplary embodiment of the present application;

fig. 11 shows a schematic structural diagram of a computer device provided in an embodiment of the present application.

Detailed Description

To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

Compared with the manual scheduling process in the related art, the embodiment of the application provides an intelligent scheduling method, which can be applied to a hospital scheduling scene, as shown in fig. 1, which shows a schematic diagram of an implementation environment provided by an exemplary embodiment of the application. The implementation environment can comprise: a computer device 100.

The computer device 100 is a device running an intelligent shift system or an intelligent shift application, which may be a tablet computer, a notebook computer, a desktop computer, or the like. In the embodiment of the present application, a scheduling manager inputs medical tasks to be scheduled into the computer device 100, the computer device 100 intelligently allocates the medical tasks to doctor objects capable of executing the medical tasks by analyzing task capability images corresponding to the medical tasks, and displays a scheduling result of the medical tasks in a front-end interface of the computer device 100, where the scheduling result shows the medical tasks required to be executed by the doctor objects. Optionally, when the intelligent scheduling system is initialized, the scheduling manager needs to input basic doctor information of each doctor object included in a department to be scheduled (a hospital to be scheduled); optionally, the shift manager needs to input some specific shift rules, such as a task time rule, a doctor assignment rule, a task role rule, and the like. Optionally, the shift scheduling manager needs to import some historical allocation data into the computer device, so that in an actual shift scheduling process, the computer device 100 may analyze an implicit historical allocation rule from the historical allocation data, perform medical task allocation based on the historical allocation rule, and make a shift scheduling result better meet actual scene requirements.

Referring to fig. 2, a flowchart of a medical task assignment method provided by an exemplary embodiment of the present application is shown. The embodiment of the present application is described by taking an example that the method is applied to a computer device, and the method includes:

step 201, determining a target medical role corresponding to the target medical task from candidate medical roles based on a target task capability portrait corresponding to the target medical task, where the target task capability portrait is used to represent medical task execution capabilities required for executing the target medical task, and different candidate medical roles correspond to different medical task execution capabilities.

The target task capability representation is used for representing medical task execution capabilities required for executing a target medical task, and the medical task execution capabilities can be described in various aspects, for example, medical skills required for executing the target medical task, medical skill proficiency, doctor roles required for executing the target medical task, and the like.

Optionally, the target task capability representation may further include information of task time, task type, and the like of the medical task.

In a medical scheduling scene, because medical tasks to be scheduled, departments to be scheduled, or doctor objects included in a hospital to be scheduled are all predetermined, task attribute information corresponding to each medical task, doctor state information corresponding to each doctor object, or doctor basic information and the like are all input into an intelligent scheduling system, a scheduling management system analyzes the task attribute information corresponding to each medical task, and a task capacity portrait corresponding to each medical task is constructed, so that a distributed doctor object corresponding to the medical task can be determined based on the task capacity portrait in the subsequent process.

Illustratively, the task capability image corresponding to the medical task of cardiac surgery may include: cardiac surgery capabilities, image analysis capabilities (analysis of cardiac imaging), etc.

Optionally, in order to enable candidates to match and analyze the assigned doctor objects corresponding to the medical task based on the task capability representation corresponding to the medical task, doctor basic information corresponding to each doctor object may also be analyzed, and doctor objects having the same type of medical task execution capability are divided into a group by using the medical task execution capability as a basic division index, so that various typical candidate medical roles are abstracted, and different candidate doctor roles correspond to different medical task execution capabilities.

Illustratively, the medical task performance capabilities of the candidate medical character 1 may include: cardiac surgery operational capabilities, image analysis capabilities (analysis of cardiac imaging); the medical task performance capabilities of the candidate medical character 2 may include: surgical anesthesia ability, etc.

Optionally, because a plurality of candidate doctor objects with the same medical task execution capability generally include, for example, for a ward round task, doctors in the department with the job titles of a doctor, an attending doctor, a subordinate principal doctor, a principal doctor, and the like can all execute the ward round task, and a plurality of doctor objects with the same job title exist in the same department, therefore, in order to avoid determining a doctor object directly based on a medical task, it is necessary to configure a medical task execution capability for each doctor object, which affects the matching efficiency of medical task capabilities, in one possible implementation, based on the medical task execution capability, a plurality of candidate doctor objects with the same medical task execution capability are abstracted into a single medical role, so as to form three-layer correspondence relationships of medical task-medical role-doctor object, and the like.

Because a target task capability image corresponding to the medical task is constructed, and candidate medical roles which characterize the execution capability of a class of medical tasks, the medical task and the candidate medical role may be associated based on the medical task performance capabilities to determine the candidate medical role for which the medical task may be performed, and, therefore, in one possible embodiment, during the intelligent shift scheduling process, if the intelligent shift scheduling system needs to determine the assigned doctor object corresponding to the target medical task, the intelligent scheduling system may analyze the target task capability representation corresponding to the target medical task, determine a target medical role matching the medical task performance capability from the abstracted candidate medical roles based on the medical task performance capability represented by the target task capability representation, and further determining an assigned doctor object corresponding to the target medical task based on the candidate doctor object corresponding to the target medical role.

Step 202, determining a target doctor object from at least one candidate doctor object corresponding to the target medical role based on a doctor allocation rule, wherein the doctor allocation rule at least comprises at least one of a doctor state rule and a historical allocation rule.

Since the target medical role may correspond to several candidate doctor objects, and a single target medical task is generally performed by a single doctor object, in one possible embodiment, the target doctor object is further determined from at least one candidate doctor object corresponding to the target medical role based on a doctor allocation rule.

The doctor allocation rule can be preset by a scheduling manager, and the corresponding intelligent scheduling system determines a target doctor object corresponding to the target medical task from a plurality of candidate doctor objects by analyzing the doctor allocation rule; optionally, the doctor allocation rule may also be learned by an intelligent analysis algorithm from historical allocation data corresponding to the department (or the hospital), and the corresponding intelligent scheduling system determines the target doctor object from the candidate doctor objects by using the intelligent analysis algorithm; optionally, the doctor allocation rule may also be determined by a dynamic planning algorithm, and the corresponding intelligent shift scheduling system determines the target doctor object from the candidate doctor objects by using the dynamic planning algorithm.

Optionally, when the doctor allocation rule is preset by the scheduling manager, the doctor allocation rule may be a doctor state rule, that is, according to the doctor state information corresponding to each current candidate doctor, it is determined which candidate doctor object is selected to execute the target medical task. For example, a doctor who specifies a certain class of job title may perform a certain class of medical tasks, a doctor with a certain operation skill may perform a certain class of medical tasks, a doctor subject with a certain working age may perform a certain class of medical tasks, a doctor subject in a vacation state does not need to perform a medical task, and the like.

Since the target medical role can only characterize the candidate doctor objects as having the capability of performing the target medical task, but since different hospital requirements and different department requirements are different, some doctor objects have the capability of performing the target medical task, but the department or hospital generally does not assign the medical task to the doctor object, for example, although the master doctor has the medical task performance capability of the medical task of ward round, the ward round task is generally assigned to the main doctor for execution; or although the doctor object a may also perform the ward round task, the doctor object a is in a vacation state and will not assign the medical task to the doctor object a, and therefore, in one possible embodiment, a target medical object suitable for performing the target medical task needs to be determined from a plurality of candidate doctor objects based on the doctor assignment rule.

Optionally, in the process of determining the target doctor object from the candidate doctor objects based on the doctor allocation rule, the target doctor object may be selected only according to a doctor state rule set by a shift manager, or an implicit historical allocation rule may be obtained by analyzing historical allocation data by using an intelligent algorithm; the doctor status rule and the historical allocation rule may be selected together, which is not limited in the embodiment of the present application.

Step 203, assigning the target medical task to the target physician subject.

In one possible embodiment, after the target medical task corresponding to the target medical task is determined, the target medical task may be assigned to the target medical task.

Schematically, the intelligent scheduling system establishes a scheduling table corresponding to the target medical task, and after a target doctor object corresponding to the target medical task is determined, the target doctor object can be input into a unit row corresponding to the target medical task; or the intelligent scheduling system establishes a scheduling form corresponding to a department doctor or a hospital doctor, and after a target doctor object corresponding to a target medical task is determined, the target medical task can be input into a unit row corresponding to the target doctor.

In summary, in the embodiment of the application, in a shift scheduling scene of a hospital, a target task capability portrait corresponding to a medical task to be assigned (target medical task) is analyzed, so that the target medical task which can execute the target medical task is found from candidate medical roles, and then an assignment object (target doctor object) of the target medical task is determined from a plurality of candidate doctor objects corresponding to the medical role based on a doctor assignment rule, so that the target medical task can be intelligently analyzed by computer equipment without manually assigning each medical task to be assigned, the target medical object corresponding to each medical task is further intelligently matched, and the shift scheduling efficiency of the medical task is improved.

In the intelligent scheduling process, some doctor state rules may relate to doctor state information of candidate doctor objects in the scheduling process based on the doctor state rules, and therefore, in a possible implementation manner, after a candidate medical object corresponding to a target medical task is determined, the doctor state information corresponding to the candidate medical object needs to be acquired, and whether the target medical task is allocated to the target doctor object is determined based on the doctor state information and the doctor state rules.

In one illustrative example, as shown in fig. 3, a flow chart of a medical task assignment method provided by another illustrative embodiment of the present application is shown. The embodiment of the present application is described by taking an example that the method is applied to a computer device, and the method includes:

step 301, acquiring the medical task execution capacity corresponding to each candidate doctor object.

Before the intelligent shift scheduling system performs intelligent shift scheduling, a task environment of the intelligent shift scheduling system needs to be initialized, and the task environment at least comprises: the medical task allocation method comprises the following steps of (1) relevant rules (artificially defined allocation rules) of medical task allocation, a medical task sequence (including medical tasks to be allocated and task attribute information corresponding to the medical tasks), doctor basic information or doctor state information, wherein the task attribute information can comprise task time, task quantity, task cost, task type and the like; the doctor basic information or doctor status information may include: personal information of the doctor (job title, service life), current state of the doctor (idle, working and vacation), medical task execution capacity corresponding to the doctor and the like.

In a possible implementation manner, after the task environment of the intelligent scheduling system is initialized, a task capability portrait corresponding to each medical task can be constructed based on task attribute information of the medical task, and a plurality of candidate medical roles can be abstracted based on the medical task execution capability corresponding to each candidate doctor object to prepare for subsequent task allocation.

Step 302, dividing the candidate doctor objects into different doctor object sets based on the medical task execution capacity, wherein the candidate doctor objects in the same doctor object set have at least one same medical task execution capacity.

In order to abstract candidate medical roles having the same medical task execution capacity, in one possible implementation, the candidate medical roles having intersection in the medical task execution capacities are divided into the same medical object set by analyzing the medical task execution capacities corresponding to the candidate medical objects, so as to abstract the candidate medical roles, that is, the candidate medical objects in the same medical object set have at least one same medical task execution capacity, which is the intersection of the medical task execution capacities corresponding to the candidate medical objects.

Illustratively, if the candidate doctor object a has a medical task execution capability a + a medical task execution capability B, the candidate doctor object B has a medical task execution capability C + a medical task execution capability D, the candidate doctor object C has a medical task execution capability a + a medical task execution capability B + a medical task execution capability F, the candidate doctor object D has a medical task execution capability E, and the candidate doctor object E has a medical task execution capability D + a medical task execution capability C; candidate physician objects a-D may be divided into three sets of physician objects based on medical task performance capabilities: the doctor object set 1-3, the doctor object set 1 ═ candidate doctor object a, candidate doctor object C }, the doctor object set 2 ═ candidate doctor object B, candidate doctor object E }, and the doctor object set 3 ═ candidate doctor object D }.

Step 303, at least one candidate medical role is determined based on the respective set of physician objects.

After the candidate doctor objects are divided, the corresponding candidate medical roles can be abstracted based on the doctor object set, so that the candidate doctor objects with the medical task execution capacity intersection correspond to the same candidate medical role.

Optionally, the role name corresponding to the candidate medical role may be predefined by a scheduling manager, and correspondingly, the intelligent scheduling system may determine the candidate medical role for different doctor object sets based on the role name naming rule.

Illustratively, the candidate medical characters may be named in numerical order, such as candidate medical character 1, candidate medical character 2, candidate medical character 3, and so on.

Optionally, after determining the candidate medical role, the candidate medical role may be associated with the medical task execution capability, so that the target medical role may be subsequently screened from the candidate medical role based on the task capability representation corresponding to the medical task.

Optionally, in the intelligent scheduling process, if a newly added doctor object exists, the candidate medical role corresponding to the newly added doctor object is determined based on the basic doctor information corresponding to the newly added doctor object.

And 304, acquiring a target task capability portrait corresponding to the target medical task.

Since a plurality of candidate doctor objects are abstracted into medical roles with specific medical task execution capacity, and the target medical task requires that the allocated objects have the capacity of executing the target medical task in the task allocation process, in one possible implementation manner, in order to match the target medical role corresponding to the target medical task, a target task capacity portrait corresponding to the target medical task needs to be obtained or constructed first to determine one or more medical task execution capacities required for executing the target medical task, and then a screening process of subsequent candidate medical roles is performed based on the target task capacity portrait.

Step 305, in response to the target task capability depiction matching the medical task performance capabilities corresponding to the candidate medical role, determining the candidate medical role as the target medical role.

Since the target medical task may be executed only by a doctor object having a specific medical task execution capability, and the candidate medical role is used to indicate a doctor object having a certain type of medical task execution capability, in a possible implementation manner, the target capability representation corresponding to the target medical task may be compared with the medical task execution capabilities corresponding to the respective candidate medical roles, and if the medical task execution capabilities match, it indicates that the candidate doctor object corresponding to the candidate medical role has the medical task execution capability to execute the target medical task, and the target medical task may be selected to be allocated to the candidate doctor object corresponding to the target medical role.

It should be noted that, the target task capability image is matched with the medical task execution capability corresponding to the candidate medical role, which may refer to the medical task execution capability represented by the target task capability image, and is the same as the medical task execution capability corresponding to the candidate medical role, schematically, if the target task capability image indicates that the target medical task execution capability image needs to have medical task execution capability a + medical task execution capability B, and the candidate doctor object corresponding to the candidate medical role 1 also has medical task execution capability a + medical task execution capability B, the candidate medical role 1 is determined as the target medical role; optionally, the matching may also mean that the medical task performance capabilities corresponding to the candidate medical roles completely include a target task performance image, if the target task performance image indicates that the target medical task needs to be performed by the medical task performance capability a + the medical task performance capability B, and the candidate medical object corresponding to the candidate medical role 2 has the medical task performance capability a + the medical task performance capability B + the medical task performance capability C, the candidate medical role 2 is determined as the target medical role.

At step 306, at least one candidate doctor object is determined based on the target medical role.

Before the intelligent scheduling, a plurality of candidate medical roles are abstracted based on doctor basic information corresponding to each candidate doctor object, and meanwhile, a corresponding relation between each candidate doctor object and each candidate medical role is established, so that in the actual scheduling process, after a target medical role corresponding to a target medical task is determined, at least one candidate doctor object with the medical task execution capacity corresponding to the target medical task can be determined based on the corresponding relation between the target medical role and the candidate doctor object.

Step 307, obtaining candidate doctor state information corresponding to each candidate doctor object.

Different departments or different hospitals have some personalized requirements for whether a medical task can be allocated to a certain doctor object, and the personalized requirements are determined not only according to whether the doctor object has the medical task execution capacity, for example, although the doctor object has the cardiac operation capacity, the hospital stipulates that heart operations cannot be independently performed in less than two years of work in the hospital, so in one possible implementation mode, after candidate doctor objects having the medical task execution capacity corresponding to a target medical task are determined, candidate doctor state information corresponding to each candidate doctor object needs to be acquired, so that whether the candidate doctor state information meets the doctor state rule is further determined, and then the target doctor object is screened out.

Optionally, the candidate doctor state information may include title information corresponding to the candidate doctor object, a current state of the candidate doctor object, a working age corresponding to the candidate doctor object, an age corresponding to the candidate doctor object, and the like.

Optionally, the doctor state rule may further be related to a task type corresponding to the target medical task, for example, a corresponding doctor state rule is set for the surgical medical task and a corresponding doctor state rule is set for the research medical task in the doctor state rule, so that task attribute information corresponding to the target medical task may also be obtained in order to find the doctor state rules corresponding to different types of medical tasks, where the task attribute information at least includes task time, task type, and the like, so that candidates may jointly screen out a target doctor object based on the task attribute information, the doctor state information, and the doctor state rule.

And 308, in response to the candidate doctor state information meeting the doctor state rule, determining the candidate doctor object corresponding to the candidate doctor state information as the target doctor object.

In a possible implementation manner, a doctor state rule to be followed in a medical task allocation process is preset in the intelligent scheduling system, and correspondingly, in an actual scheduling process, the intelligent scheduling system needs to determine whether candidate doctor state information meets the doctor state rule, and further determines whether a target medical task is allocated to the candidate doctor object.

Optionally, the doctor status rule may include, corresponding to the doctor status information: the current state condition of the doctor, such as preferentially distributing medical tasks to candidate doctor objects in an idle state and not allowing the medical tasks to be distributed to candidate doctor objects in a vacation state; doctor job title conditions may also be included, such as, for example, a candidate doctor object for which a surgical-like medical task is not allowed to be assigned to the attending physician for the following job title; a physician age condition may also be included, such as a night ward round task not allowed to be assigned to a candidate physician subject over the age of 50; doctor working age conditions may also be included, such as surgical tasks not allowed to be assigned to candidate doctor subjects with working ages less than two years, etc.

Illustratively, if the target medical task is a cardiac procedure, the physician state information of the corresponding candidate physician object can be shown in table one.

Watch 1

Based on the doctor status rules: preferentially allocating medical tasks to the candidate doctor objects in the idle state, not allowing the medical tasks to be allocated to the candidate doctor objects in the vacation state, and not allowing the medical tasks in the operation class to be allocated to the candidate doctor objects of the following job titles of doctors; the surgery type task is not allowed to be allocated to candidate doctor objects with working years less than two years, and the candidate doctor object 3 can be determined as the target doctor object corresponding to the target medical task.

Step 309, assigning the target medical task to the target physician subject.

Step 203 may be referred to in the implementation manner of this step, and this embodiment is not described herein again.

Step 310, in response to the completion of the assignment of the target medical task, updating a task assignment status corresponding to the target medical task and updating target doctor status information corresponding to the target doctor object.

Because the intelligent scheduling task is a cyclic task, namely after the task allocation of a single target medical task is completed, the task allocation work of the next target medical task needs to be continued until the initialized medical task sequence allocation is completed, whereas the result of the assignment of the last target medical task may influence the assignment of the next target medical task, for example, if the task of the target medical task is completed, the task of the target medical doctor is completed, the target medical task may be performed by the target medical subject in time, and the next target medical task may not be assigned to the target medical subject, and therefore, to avoid the above, or to avoid the repeated assignment of medical tasks, after each task allocation is finished, the task allocation state corresponding to the target medical task needs to be updated based on the allocation result corresponding to the target medical task, and meanwhile, the target doctor state information corresponding to the target doctor object needs to be updated.

Wherein, updating the task allocation state corresponding to the target medical task means: and updating the task allocation state of the target medical task from unallocated to allocated, or from in allocation to complete allocation or allocated.

Illustratively, if the target medical task is already allocated, the target medical task may be removed from the medical task sequence, directly deleted, or added to the allocated task sequence; if the target medical task is allocated to the target doctor object, the allocated task number corresponding to the target doctor object can be increased by one, meanwhile, whether the target doctor object has spare time is determined based on the task time corresponding to the target medical task, and if the target doctor object does not have spare time, the target doctor object is set to be in a medical task unallocated state; otherwise, the target medical object is set to the assignable medical task state.

In the embodiment, the object relationship between the candidate doctor object and the medical role is established by analyzing the doctor state information corresponding to each candidate doctor object, so that the corresponding target medical role can be matched based on the task capability image of the target medical task in the actual scheduling process, and further the distribution object corresponding to the target medical task is determined; in addition, the scheduling manager is allowed to customize the personalized doctor state rule, so that the target medical object screened out based on the doctor state rule meets the task allocation requirement of the department or the hospital; in addition, the task attribute information corresponding to the target medical task and the doctor state information of the target doctor object can be updated in time based on the distribution result of the target medical task, the repeated execution of the distribution process of the target medical task is avoided, and the violation of the doctor state rule caused by the untimely update of the state information in the task distribution process is avoided.

Because some task allocation rules cannot be directly written into the intelligent scheduling system in a customized form, in order to still enable the task allocation results to meet the current department conditions, in a possible implementation manner, historical allocation data can be analyzed by adopting an artificial intelligence algorithm, so that a target doctor object can be selected.

On the basis of FIG. 3, as shown in FIG. 4, steps 306-308 can be replaced by steps 401-403.

At step 401, at least one candidate doctor object is determined based on the target medical role.

Step 305 may be referred to in the implementation manner of step 401, and this embodiment is not described herein again.

Step 402, acquiring candidate doctor state information corresponding to each candidate doctor object and target task attribute information corresponding to the target medical task.

Since whether the target medical task can be allocated to the candidate medical task may be influenced by the state information of the candidate medical task corresponding to the candidate medical task or may be influenced by the attribute information of the target medical task corresponding to the target medical task, in order to facilitate the task allocation model to determine the target task object corresponding to the target medical task, in a possible implementation manner, the state information of the candidate medical task corresponding to each candidate medical task and the attribute information of the target task corresponding to the target medical task need to be acquired.

Illustratively, the target task attribute information includes at least a task time, a task type, a task cost, and the like of the target medical task.

And 403, inputting the candidate doctor state information, the target task attribute information and the historical distribution data into a task distribution model to obtain a target doctor object output by the task distribution model, wherein the task distribution model is used for analyzing a historical distribution rule represented by the historical distribution data and determining the target doctor object based on the historical distribution rule.

During the application process of the task allocation model, historical allocation data needs to be analyzed, implicit historical allocation rules are mined, allocation probabilities corresponding to candidate doctor objects are determined based on the historical allocation rules, and then target doctor objects are determined.

In a possible implementation mode, based on the distribution principle of the task distribution model, candidate doctor state information, target task attribute information and historical distribution data are input into the task distribution model together, the probability of each candidate doctor object executing the target medical task is predicted according to the historical distribution data of the corresponding task distribution model, and then the target doctor object is selected based on the probability.

In an illustrative example, step 403 may include step 403A and step 403B.

Step 403A, inputting the candidate doctor state information, the target task attribute information and the historical distribution data into the task distribution model to obtain candidate distribution probabilities corresponding to the candidate doctor objects output by the task distribution model.

The task allocation model may analyze an implicit historical allocation rule of the department or hospital based on the historical allocation data, and therefore, in one possible implementation, the target task attribute information and the historical allocation data are input into the task allocation model, the task allocation model searches a historical allocation result corresponding to a target medical task similar to or identical to the target medical task from the historical allocation data based on the target task attribute information, and then analyzes a candidate allocation probability for allocating the target medical task to each candidate doctor object based on doctor information of the doctor object indicated by the historical allocation result, and then screens out the target doctor object based on the candidate allocation probability.

In the process of outputting the candidate assignment probability by the task assignment model, basic information of each candidate doctor object, such as age, title, name, and age, in the candidate doctor state information needs to be used.

Step 403B, determining a target doctor object from the candidate doctor objects based on the candidate distribution probability and the candidate doctor state information.

Optionally, after determining the candidate distribution probability corresponding to each candidate doctor object, the candidate doctor object with the highest candidate distribution probability may be directly determined as the target doctor object.

Optionally, if the candidate doctor object with the highest candidate distribution probability is directly selected as the target doctor object, if the task of the target doctor object is completed, the target doctor object may be overloaded to work; or the target doctor object is on vacation, which may cause the task to be executed without people, in order to improve the accuracy of task allocation, in a possible implementation manner, after obtaining the candidate allocation probability corresponding to each candidate doctor object, it is further necessary to determine the current working state corresponding to each candidate doctor object from the candidate doctor state information, and further determine the target doctor object from the candidate doctor objects based on the candidate allocation probability and the candidate doctor state information, that is, it is necessary to jointly screen the target doctor objects based on the candidate allocation probability and the candidate doctor state information.

Optionally, in the process of screening the target doctor object based on the candidate allocation probability and the candidate doctor state information, the candidate allocation probability is preferentially the highest, and the candidate doctor object indicated as an idle state by the candidate doctor state information is determined as the target doctor object.

Illustratively, if the target medical task corresponds to the candidate doctor object 1 to the candidate doctor object 4, the candidate assignment probabilities corresponding to the candidate doctor objects output by the task assignment model are respectively: p₁When the value is 0.4, the assignment probability of the candidate corresponding to the candidate doctor object 1 is 0.4, and P is₂When the value is 0.5, the candidate assignment probability corresponding to the candidate doctor object 2 is 0.5, and P is₃When the value is 0.05, the candidate assignment probability corresponding to the candidate doctor object 3 is 0.05, and P is₄When the candidate assignment probability corresponding to the candidate doctor object 4 is 0.05, it is indicated that the candidate doctor state information corresponding to each candidate doctor object indicates: the candidate doctor object 1 is in an idle state, the candidate doctor object 2 is in work, the number of tasks is saturated, the candidate doctor object 3 is in an idle state, and the candidate doctor object 4 is in vacation, wherein the candidate assignment probability corresponding to the candidate doctor object 2 is the highest, but the candidate doctor state information corresponding to the candidate doctor object indicates that the number of tasks is saturated, and then the target medical task is assigned to the candidate doctor object 1.

Optionally, the output of the task allocation model may be a candidate allocation probability corresponding to each candidate doctor object, and the candidate doctor object with the highest candidate allocation probability may be determined as the target doctor object.

Illustratively, if the target medical task is a cardiac surgery task and corresponds to the candidate doctor objects 1-6, the task allocation model analyzes historical allocation data to know that the cardiac surgery task of the department is allocated to the candidate doctor object 3 with a high probability, and correspondingly, the target medical task can be allocated to the candidate doctor object 3.

Optionally, since the task allocation model only needs to predict the probability of each candidate doctor object executing the target medical task, in order to reduce the data processing amount of the task allocation model, only the historical allocation data corresponding to each candidate doctor object may be input into the task allocation model.

Optionally, a target doctor object may be jointly screened from candidate doctor objects based on a task allocation model (that is, a historical allocation rule) and a medical state rule, in a possible implementation manner, after a candidate allocation probability corresponding to each candidate doctor object output by the task allocation model is obtained, a candidate doctor object with the highest candidate allocation probability is determined, and then it is determined whether candidate doctor state information corresponding to the candidate doctor object satisfies the doctor state rule, if so, the candidate doctor object is determined as the target doctor object, and if not, it is further determined whether a candidate doctor object with the second allocation probability satisfies the doctor state rule, and so on until a target doctor object satisfying both the historical allocation data and the doctor state rule is determined.

In the embodiment, the implicit historical allocation rule in the historical allocation data is learned through the task allocation model, and the target doctor object is determined from the candidate doctor objects, so that the determined target doctor object meets the actual requirements of departments or hospitals, the accuracy of the intelligently determined target doctor object is improved, and the accuracy of the scheduling result is further improved.

In order to enable the task allocation model to have the capability of analyzing historical allocation data, in a possible implementation manner, before actual scheduling is performed, reinforcement learning needs to be performed on the task allocation model, so that the task allocation model can determine a target doctor object meeting the requirement of a department scheduling index from a plurality of candidate doctor objects in a model application process.

Referring to fig. 5, a flowchart of a method for training a task assignment model according to an exemplary embodiment of the present application is shown, the method including:

step 501, obtaining ith sample task attribute information, ith sample doctor state information and ith sample historical distribution data, wherein i is a positive integer.

The training purpose of the task allocation model is as follows: the task allocation model needs to determine an allocation object of the sample medical task from the sample doctor objects based on analysis of historical allocation data of the sample, so that an allocation result of the sample medical task meets a preset training index.

Optionally, the training index of the task allocation model is generally measured by an overall allocation result of the sample medical task sequence, for example, allocation balance, the number of allocated doctors, overtime time of a doctor object, and the like, and therefore, in order to integrally evaluate an allocation effect of the task allocation model, in a possible implementation, the sample medical task sequence is configured, the sample medical task sequence is composed of m sample medical tasks, the sample medical task sequence is scheduled by the task allocation model to determine allocated doctors corresponding to each sample medical task in the sample medical task sequence, and the task allocation model is trained based on the overall scheduling result corresponding to the sample medical task sequence. Wherein m is a positive integer, i is less than or equal to m, the value of m can be 20, and the corresponding sample medical task sequence can contain 20 sample medical tasks.

In the scheduling process of the sample medical task sequence, distributed doctor objects corresponding to sample medical tasks in the sample medical task sequence need to be determined, for example, an ith distributed doctor object corresponding to an ith sample medical task is determined, the computer equipment obtains attribute information of the ith sample task corresponding to the ith sample task sequence, at least one ith sample doctor object with the medical task execution capacity corresponding to the ith sample medical task, ith sample doctor state information of the ith sample doctor object and ith sample historical distribution data, so that the sample data is used for training a task distribution model.

Alternatively, the ith sample historical assignment data may be historical assignment data associated with each ith sample physician subject.

Alternatively, the determination manner of the ith sample doctor object may be specified by a human being, or may be determined according to the determination manner of the candidate doctor object in the above embodiment.

Optionally, the sample medical task sequence may be trained repeatedly, that is, after the sample medical task completes the ith round of training of the task allocation model, the task allocation model obtained after the ith round of training may be used to perform task allocation on the sample medical task sequence again.

Step 502, inputting the ith sample task attribute information, the ith sample doctor state information and the ith sample historical distribution data into a task distribution model to obtain an ith distribution doctor object corresponding to the ith sample medical task output by the task distribution model.

In one possible implementation mode, the ith sample task attribute information, the ith sample doctor state information and the ith sample historical allocation data are input into a task allocation model, the ith sample historical allocation data are analyzed by the task allocation model, an implicit historical allocation rule is learned from the ith sample historical allocation data, and then the ith allocated doctor object corresponding to the ith sample medical task is determined based on the implicit historical allocation rule.

The task allocation model outputs the allocation probability corresponding to each ith sample doctor object, and the ith sample doctor object with the highest allocation probability is determined as the ith allocation doctor object.

And 503, responding to the completion of the distribution of the sample medical task sequence, and performing reinforcement learning on the task distribution model based on the distribution doctor object corresponding to the sample medical task sequence.

In a scheduling scenario, in order to analyze task allocation performance of a task allocation model, training indexes are generally involved, for example, whether the number of sample medical tasks allocated to each sample doctor object is average, whether the number of sample doctor objects occupied by the sample medical tasks in the sample medical task sequence is small, whether the overtime of each sample doctor object is small, and the like, and these training indexes are determined based on an allocation result of the sample medical task sequence after the task allocation model allocates the sample medical task sequence, so that in a possible implementation manner, after the task allocation model determines the ith allocated doctor object corresponding to the ith sample medical task, the i +1 th sample medical task in the sample medical task sequence may be continuously scheduled to determine the ith +1 th allocated doctor object corresponding to the ith +1 th sample medical task, and analogizing in sequence until the task allocation of all the sample medical tasks in the sample medical task sequence is completed to obtain each allocated doctor object corresponding to the sample medical task sequence, and then performing reinforcement learning on the task allocation model based on the allocation result.

Optionally, in the scheduling process of the sample medical task sequence, along with the allocation of the sample medical task, the sample medical state information of each sample medical object may change, and the allocation state of the sample medical task may also change, so in a possible implementation, the ith +1 sample task attribute information, the ith +1 sample medical state information, and the ith +1 sample historical allocation data may be determined based on the ith allocation medical object corresponding to the ith sample medical task.

For the mode of determining the attribute information of the (i + 1) th sample task, if the distribution of the ith sample medical task is completed, the (i + 1) th sample medical task can be determined based on the ith sample medical task, and then the (i + 1) th sample medical task is determined; for the method of determining the i +1 th sample doctor state information, if the i th sample medical task allocation is completed, the number of allocation tasks of the sample doctor object corresponding to the i th allocation doctor object is increased by one, and the allocable task time corresponding to the sample doctor object is reduced, if the i th allocation doctor object also has the medical task execution capacity of the i +1 th sample medical task, the i +1 th sample doctor state information corresponding to the i +1 th sample medical task is influenced by the allocation result of the i th sample medical task; for the way of determining the ith +1 th sample historical allocation data, after the allocation task of the ith sample medical task is completed, the allocation result data of the ith sample medical task may be added to the ith +1 th sample historical allocation data.

Optionally, because actual scheduling scenarios of different departments or different academies are different, or scheduling effects achieved by the required task allocation models are different, in a possible implementation manner, different training indexes may be designed, and the task allocation models are subjected to reinforcement learning with the purpose of the different training indexes. In one illustrative example, step 503 may include step 503A or step 503B.

Step 503A, taking the task allocation balance of the sample medical task sequence as a training index, and performing reinforcement learning on the task allocation model based on the allocated doctor object corresponding to the sample medical task sequence.

Optionally, the training index of the task allocation model may be balance of task allocation of the sample medical task sequence, that is, if the task allocation model can allocate each sample medical task in the sample medical task sequence to each sample doctor object more evenly, it may not happen that some sample doctor objects have more sample medical tasks, and some sample doctor objects have fewer sample medical tasks, it indicates that the task allocation model is trained completely.

In an exemplary example, with the task allocation balance as a training index, the process of performing reinforcement learning on the task allocation model may include the following steps one to two.

Determining sample task balance scores based on the assigned doctor objects corresponding to the sample medical task sequence, wherein the task balance scores are determined by the task assignment number corresponding to each assigned doctor object.

In a possible implementation manner, after the sample medical task sequence is distributed, distributing doctor objects in the sample medical task sequence are obtained, the task distribution quantity corresponding to each distributing doctor object is determined, and then the sample task balance score is determined based on the task distribution quantity corresponding to each distributing doctor object, so that the sample task balance score is compared with the target task balance score subsequently, and the reward of the task distribution model is determined.

Optionally, for the mode of determining the sample task balance score, the task allocation numbers corresponding to the respective allocated doctor objects may be compared to obtain the maximum task allocation number and the minimum task allocation number, and the sample task balance score may be determined based on a difference between the maximum task allocation number and the minimum task allocation number. The larger the difference value is, the larger the sample task balance score is, and the larger the sample task balance score is, the more unbalanced the task allocation quantity is.

Illustratively, if the maximum task allocation number is 8 and the minimum task allocation number is 1, the sample task balance score may be 7.

And secondly, performing reinforcement learning on the task allocation model based on the sample task balance score and the target task balance score.

The target task balance score may be set by a developer, and illustratively, the target task balance score may be 0 or 1.

In one possible implementation, the assignment model is reinforcement learned by comparing the difference between the sample task balance score and the target task balance score.

Wherein the reward of the assignment model may be determined by a difference between the sample task balance score and the target task balance score.

Step 503B, taking the sample doctor objects of which the sample medical task sequence occupies the target number as training indexes, and performing reinforcement learning on the task allocation model based on the allocated doctor objects corresponding to the sample medical task sequence.

Optionally, the training index of the task allocation model may also be sample doctor objects whose sample medical task sequence occupies a target number, that is, the number of sample doctor objects occupied by the sample medical task sequence is preset, the task allocation model performs task allocation on the sample medical task sequence, and if the allocation result is consistent with the preset occupied target number, and there is no excessive occupation of doctor resources, it indicates that the task allocation model is trained completely.

In an exemplary example, with the target number of sample medical task sequences as a training index, the process of performing reinforcement learning on the task allocation model may include the following three steps and four steps.

And thirdly, determining the number of sample doctors corresponding to the allocated doctor objects based on the allocated doctor objects corresponding to the sample medical task sequence.

In a possible implementation manner, when the sample medical task sequence is completely allocated, the allocated doctor object corresponding to the sample medical task sequence is obtained, the number of sample doctors corresponding to the allocated doctor object is determined, and whether doctor resources are excessively occupied is judged based on the number of sample doctors.

And fourthly, performing reinforcement learning on the task allocation model based on the number of the sample doctors and the number of the target doctors.

The number of target doctors can be preset by a developer, the number of target doctors can be the number of doctors occupied by the sample medical task sequence in the manual scheduling process, and illustratively, if the sample medical task sequence includes 20 medical task sequences, the number of corresponding target doctors can be 5.

In one possible embodiment, the task assignment model is reinforcement learned by comparing the difference between the sample and target physician numbers. Wherein the reward of the task allocation model can be determined by the difference between the number of sample doctors and the number of target doctors; the larger the difference between the number of sample physicians and the number of target physicians, the smaller the reward for the task assignment model.

Optionally, the task allocation model may be subjected to reinforcement learning by using the minimum overtime as a training index, and in a possible implementation manner, a scheduling result of the sample medical task sequence is obtained, working hours corresponding to each allocation doctor are counted, the overtime is screened out, and the task allocation model is subjected to reinforcement learning based on the overtime. Wherein, the longer the overtime duration is, the smaller the task reward of the task allocation model is.

In the embodiment, the task allocation model has the capability of analyzing historical allocation data by performing reinforcement learning on the task allocation model, so that the given historical allocation data can be accurately analyzed in the model application process to determine the target doctor object corresponding to the target medical task.

In a possible application scenario, the scheduling index requirements of the same department or hospital in different time periods may differ, for example, a working day generally requires the least overtime, and a holiday may require less doctor resource occupation, so that in an actual scheduling process, an appropriate task allocation model needs to be selected based on different scheduling indexes (allocation indexes).

Referring to fig. 6, a flowchart of a medical task assignment method provided by another exemplary embodiment of the present application is shown. The embodiment of the present application is described by taking an example that the method is applied to a computer device, and the method includes:

step 601, determining a target medical role corresponding to the target medical task from the candidate medical roles based on the target task capability portrait corresponding to the target medical task.

The implementation of step 601 may refer to the above embodiments, which are not described herein.

Step 602, obtaining a target distribution index corresponding to a target medical task sequence, where the target medical task sequence includes a target medical task.

In a possible implementation manner, when initializing a target medical task sequence, a scheduling manager selects a target allocation index corresponding to the target medical task sequence, and acquires the target allocation index corresponding to the target medical task sequence corresponding to an intelligent scheduling system.

Optionally, a task allocation model obtained by training (allocating) with different training indexes is preset in the computer device, and the different training indexes are displayed in the intelligent shift front-end interface, so that shift managers can select the shift models based on the current scene.

Optionally, the target allocation index may also be determined by the intelligent scheduling system based on a scheduling time period corresponding to the target medical task sequence, in a possible implementation manner, when the scheduling manager inputs the target medical task sequence and the related task attribute information into the intelligent scheduling system, the intelligent scheduling management system may obtain the task time period corresponding to the sample task sequence, and further determine the target allocation index based on the task time period.

Illustratively, if the task time period is a normal working day, it may be determined that the target allocation index may be the least overtime, and if the task time period is a holiday, the target allocation index may be the least occupied doctor resources.

Step 603, determining a target task allocation model based on the target allocation indexes, wherein different task allocation models correspond to different training indexes.

The computer equipment stores task distribution models trained by different training indexes, and the training indexes correspond to the distribution indexes.

In one possible implementation, the target task allocation model is determined based on the target allocation indicator, so that the target doctor object is determined to meet the current scene requirements based on the target task allocation model.

It should be noted that, for the same target medical task sequence, generally corresponding to the same target allocation index, the target task allocation model only needs to be determined when the first medical task of the target medical task sequence is allocated, and then the target task allocation model can be directly used for task allocation without repeated determination.

Step 604, determining a target medical role from at least one candidate medical role corresponding to the target medical role based on the target task assignment model.

In a possible implementation manner, after the target task allocation model is determined, candidate doctor state information corresponding to the candidate doctor object, task attribute information corresponding to the target medical task, and historical allocation data may be input into the target task allocation model, and the target task allocation model analyzes the historical allocation data to obtain an implicit historical allocation rule, so as to determine the target doctor object from the candidate doctor objects based on the historical allocation rule.

FIG. 7 is a diagram illustrating a process for training and applying a task assignment model according to an exemplary embodiment of the present application. In the training process of the task allocation model, firstly, carrying out system initialization on the intelligent scheduling system, importing and analyzing the rules which are well formulated, and importing historical data required by training to complete task rule environment setting; training an artificial intelligence algorithm (task allocation model) based on historical data, training to generate a feasible strategy, indicating the trained task allocation model by the feasible strategy, and adding the trained task allocation model into a strategy pool so as to select a proper strategy in the scheduling process; optionally, a historical policy may also be imported into the policy pool; after the task allocation model is trained, a scheduling rehearsal and testing process can be carried out, different task allocation models are selected from the strategy pool for scheduling, a scheduling result is generated, the scheduling result is visualized and analyzed, and an appropriate strategy (an appropriate task allocation model) meeting design indexes is selected.

Step 605, assign the target medical task to the target physician subject.

The implementation manner of step 605 may refer to the above embodiments, which are not described herein.

In this embodiment, by providing task allocation models with different training indexes (different allocation indexes), in an actual scheduling process, an appropriate task allocation model can be selected based on scheduling requirements of different scenes, so that accuracy of task allocation is improved, and a task allocation result meets an actual scheduling condition.

Referring to fig. 8, a system architecture diagram of an intelligent shift scheduling system according to an exemplary embodiment of the present application is shown, the intelligent shift scheduling system 800 includes: a task environment initialization module 801, a data storage module 802, a front-end interaction module 803, a core decision module 804, a policy analysis module 805, and a result visualization module 806.

The task environment initialization module 801 is mainly responsible for receiving input and analysis of each initialization parameter of the shift scheduling task environment, and the initialization parameters mainly include: task attribute information of each medical task (for example, task time, task type, task cost, and medical task execution capacity required by the task), doctor state information (basic information such as the working age, and title of the doctor, and the current state of the doctor) corresponding to the hospital (department) in which the doctor is located, various rules and restrictions (for example, doctor allocation rules) that are formulated in advance, and the like.

The data storage module 802 is mainly responsible for storing internal and external data to be stored in the system operation process, including but not limited to historical data imported in advance, various trained algorithm models (for example, task allocation models trained with different training indexes), real-time decision result snapshots, final decision results, and the like, and corresponding data I/O work.

The front-end interaction module 803 is responsible for providing corresponding interfaces and functions of system management for the scheduling staff of the hospital, including but not limited to updating of task environment information, calling of past results, analyzing of scheduling results corresponding to different decision algorithms, manual adjustment of the scheduling results, and the like.

The core decision module 804 is mainly responsible for analyzing the scheduling task and generating a core decision result. That is, the core decision module 804 is used to generate target physician objects, i.e., shift results, for performing various medical tasks.

The policy analysis module 805 is mainly responsible for subsequent analysis of various generated decision results. That is, in the actual application process, whether the generated scheduling result meets the pre-specified distribution index is analyzed, so as to further correct the scheduling strategy.

The result visualization module 806 is mainly responsible for visualization display of the task environment, the shift arrangement result, and the related analysis result. For example, the scheduling result corresponding to the medical task sequence is displayed in the front-end interface.

Based on the architecture diagram of the intelligent shift scheduling system shown in fig. 8, as shown in fig. 9, a process diagram of medical task assignment shown in an exemplary embodiment of the present application is shown. The core decision module 804 is a core module in the intelligent shift scheduling management system, and is mainly responsible for task environment calling, training various artificial intelligence algorithms, and outputting shift scheduling results, and mainly includes: decision core 905, task pool 901, doctor pool 902, role rule module 904, rule pool 903, historical data pool 907, and policy pool 906.

The task pool 901 dynamically stores each medical task that needs to be scheduled in the scheduling process, and the corresponding related task attribute information, including: task information of a plurality of aspects such as task time (e.g., task cycle period, task start time, task end time, etc.), number of tasks, task cost, and task type (task of an associated type, task of an exclusive type, task of a special rule type, etc.). Optionally, the medical tasks in the task pool 901 may update the task status as the shift progresses.

The doctor pool 902 dynamically stores relevant information of each doctor object in the shift scheduling system, including: doctor personal information (name, age, sex, job title, and age of job) and doctor corresponding ability, a doctor corresponding role list, and a doctor status (whether the doctor is on vacation, on-duty, or in idle). Optionally, the doctor information in the doctor pool 902 is updated as the shift progresses.

The rule pool 903 dynamically stores various rules to be followed in the shift scheduling system, including task related rules, role related rules, doctor related rules, and the like.

The role rule module 904, acting as an intermediary between the two main entities, the doctor pool and the task pool, parses and stores the role-task correspondence and the corresponding rules. In the actual scheduling process, the role rule module 904 can also undertake the information interaction tasks of the task pool and the doctor pool, for example, the role rule module 904 can acquire task attribute information in the task pool 901, and construct a task capability portrait to establish the corresponding relationship between the medical task and the medical role; the role rules module 904 also obtains physician (status) information in the physician pool 902 to abstract out various physician roles. Optionally, when the role rule module 904 constructs the correspondence between the medical task and the medical role, a role rule with a finer granularity is further set, for example, a task that each medical role can complete, an assumed workload, a rotation/rest policy, and some other special rules.

The historical data pool 907 is responsible for storing and retrieving sample historical allocation data required by the training strategy, as well as historical allocation data required during the shift scheduling process.

The strategy pool 906 is responsible for storing strategies trained for different training indexes, that is, task allocation models trained for different training indexes, and also stores dynamic planning algorithms not requiring training (the dynamic planning algorithms may include methods such as a greedy method, a branch backtracking method, and a minimum subtask decomposition method), so that the decision core is used in the scheduling process. Optionally, the trained strategy belongs to an artificial intelligence algorithm. Optionally, the policy pool 906 invokes historical allocation data from the historical data pool 907 to train the artificial intelligence algorithm.

In the scheduling system decision process, the intelligent scheduling system determines a target medical role corresponding to a target medical task based on a task-role relationship, the decision core 905 acquires doctor state information corresponding to a candidate doctor object from the doctor pool 902, acquires task attribute information corresponding to the target medical task from the task pool 901, selects an algorithm used for decision from the strategy pool 906, for example, selects an artificial intelligence algorithm, analyzes historical allocation data, the doctor state information and the task attribute information by the artificial intelligence algorithm, determines the target doctor object corresponding to the target medical task, and issues an allocation instruction to the doctor pool 902, that is, allocates the target medical task to the target doctor object; optionally, after the target medical task is assigned, the task state in the task pool 901 is updated.

In the following, embodiments of the apparatus of the present application are referred to, and for details not described in detail in the embodiments of the apparatus, the above-described embodiments of the method can be referred to.

Fig. 10 is a block diagram of a medical task assigning apparatus according to an exemplary embodiment of the present application. The device includes:

a first determining module 1001, configured to determine, from candidate medical roles, a target medical role corresponding to a target medical task based on a target task capability portrait corresponding to the target medical task, where the target task capability portrait is used to characterize medical task performance capabilities required for performing the target medical task, and different candidate medical roles correspond to different medical task performance capabilities;

a second determining module 1002, configured to determine a target doctor object from at least one candidate doctor object corresponding to the target medical role based on a doctor allocation rule, where the doctor allocation rule at least includes at least one of a doctor state rule and a historical allocation rule;

an assigning module 1003 for assigning the target medical task to the target physician subject.

Optionally, the doctor allocation rule is the doctor status rule;

the second determining module 1002 includes:

a first determination unit for determining at least one of the candidate doctor objects based on the target medical role;

the first acquisition unit is used for acquiring candidate doctor state information corresponding to each candidate doctor object;

and a second determining unit, configured to determine, in response to that the candidate doctor state information satisfies the doctor state rule, a candidate doctor object corresponding to the candidate doctor state information as the target doctor object.

Optionally, the doctor allocation rule is the historical allocation rule;

the second determining module 1002 includes:

a third determination unit for determining at least one of the candidate doctor objects based on the target medical role;

the second acquisition unit is used for acquiring candidate doctor state information corresponding to each candidate doctor object and target task attribute information corresponding to the target medical task;

and a fourth determining unit, configured to input the candidate doctor state information, the target task attribute information, and historical allocation data into a task allocation model, so as to obtain the target doctor object output by the task allocation model, where the task allocation model is configured to analyze the historical allocation rule represented by the historical allocation data, and determine the target doctor object based on the historical allocation rule.

Optionally, the fourth determining unit is further configured to:

inputting the candidate doctor state information, the target task attribute information and the historical distribution data into the task distribution model to obtain candidate distribution probability corresponding to each candidate doctor object output by the task distribution model;

determining the target physician object from the candidate physician objects based on the candidate assignment probabilities and the candidate physician state information.

Optionally, the apparatus further comprises:

the first acquisition module is used for acquiring the medical task execution capacity corresponding to each candidate doctor object;

a dividing module, configured to divide the candidate doctor objects into different doctor object sets based on the medical task execution capacity, where the candidate doctor objects in the same doctor object set have at least one same medical task execution capacity;

a third determination module for determining at least one of the candidate medical roles based on the respective set of physician objects.

Optionally, the first determining module 1001 includes:

the third acquisition unit is used for acquiring the target task capability portrait corresponding to the target medical task;

a fifth determining unit, configured to determine the candidate medical role as the target medical role in response to the target task capability representation matching the medical task performance capability corresponding to the candidate medical role.

Optionally, the apparatus further comprises:

and the updating module is used for responding to the completion of the target medical task allocation, updating the task allocation state corresponding to the target medical task and updating the target doctor state information corresponding to the target doctor object.

Optionally, the historical allocation rule is determined by a task allocation model;

the device further comprises:

a second obtaining module, configured to obtain ith sample task attribute information, ith sample doctor state information, and ith sample historical allocation data, where the ith sample task attribute information is task attribute information corresponding to an ith sample medical task, the ith sample medical task belongs to a sample medical task sequence, the ith sample doctor state information is doctor state information corresponding to an ith sample doctor object, the ith sample doctor object is a doctor object having a medical task execution capability corresponding to the ith sample medical task, and i is a positive integer;

a fourth determining module, configured to input the ith sample task attribute information, the ith sample doctor state information, and the ith sample historical allocation data into the task allocation model, so as to obtain an ith allocated doctor object corresponding to the ith sample medical task output by the task allocation model;

and the reinforcement learning module is used for responding to the completion of the sample medical task sequence distribution and carrying out reinforcement learning on the task distribution model based on the distribution doctor object corresponding to the sample medical task sequence.

Optionally, the reinforcement learning module includes at least one of:

the first reinforcement learning unit is used for performing reinforcement learning on the task allocation model based on the allocated doctor object corresponding to the sample medical task sequence by taking the task allocation balance of the sample medical task sequence as a training index;

and the second reinforcement learning unit is used for taking the sample doctor objects of which the sample medical task sequence occupies the target number as training indexes and carrying out reinforcement learning on the task allocation model based on the allocated doctor objects corresponding to the sample medical task sequence.

Optionally, the first reinforcement learning unit is further configured to:

determining a sample task balance score based on the assigned doctor objects corresponding to the sample medical task sequence, wherein the task balance score is determined by the task assignment number corresponding to each assigned doctor object;

and performing reinforcement learning on the task allocation model based on the sample task balance score and the target task balance score.

Optionally, the second reinforcement learning unit is further configured to:

determining the number of sample doctors corresponding to the distribution doctor object based on the distribution doctor object corresponding to the sample medical task sequence;

and performing reinforcement learning on the task allocation model based on the number of the sample doctors and the number of the target doctors.

Optionally, the apparatus further comprises:

and the fifth determining module is used for determining the (i + 1) th sample task attribute information, the (i + 1) th sample doctor state information and the (i + 1) th sample historical distribution data based on the (i) th distribution doctor object corresponding to the i-th sample medical task.

Optionally, the second determining module 1002 includes:

a third obtaining unit, configured to obtain a target allocation indicator corresponding to a target medical task sequence, where the target medical task sequence includes the target medical task;

a sixth determining unit, configured to determine a target task allocation model based on the target allocation index, where different task allocation models correspond to different training indexes;

a seventh determining unit, configured to determine the target doctor object from the at least one candidate doctor object corresponding to the target medical role based on the target task assignment model.

In summary, in the embodiment of the application, in a shift scheduling scene of a hospital, a target task capability portrait corresponding to a medical task to be allocated (target medical task) is analyzed, so that the target medical task which can execute the target medical task is found from candidate medical roles, and then an allocation object (target doctor object) of the target medical task is determined from a plurality of candidate doctor objects corresponding to the medical role based on a doctor allocation rule, so that each medical task to be allocated does not need to be manually allocated, the target medical task can be intelligently analyzed by computer equipment, the target medical objects corresponding to each medical task are further intelligently matched, and the allocation efficiency of the medical task is improved.

Referring to fig. 11, a schematic structural diagram of a computer device according to an embodiment of the present application is shown. The computer apparatus 1100 includes a Central Processing Unit (CPU) 1101, a system Memory 1104 including a Random Access Memory (RAM) 1102 and a Read-Only Memory (ROM) 1103, and a system bus 1105 connecting the system Memory 1104 and the Central Processing unit 1101. The computer device 1100 also includes a basic Input/Output system (I/O) 1106, which facilitates transfer of information between devices within the computer, and a mass storage device 1107 for storing an operating system 1113, application programs 1114, and other program modules 1115.

The basic input/output system 1106 includes a display 1108 for displaying information and an input device 1109 such as a mouse, keyboard, etc. for user input of information. Wherein the display 1108 and input device 1109 are connected to the central processing unit 1101 through an input/output controller 1110 connected to a system bus 1105. The basic input/output system 1106 may also include an input/output controller 1110 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, input/output controller 1110 also provides output to a display screen, a printer, or other type of output device.

The mass storage device 1107 is connected to the central processing unit 1101 through a mass storage controller (not shown) that is connected to the system bus 1105. The mass storage device 1107 and its associated computer-readable media provide non-volatile storage for the computer device 1100. That is, the mass storage device 1107 may include a computer-readable medium (not shown) such as a hard disk or a CD-ROM (Compact disk Read-Only Memory) drive.

Without loss of generality, the computer-readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash Memory or other solid state Memory technology, CD-ROM, DVD (Digital Video Disc) or other optical, magnetic, or other magnetic storage devices. Of course, those skilled in the art will appreciate that the computer storage media is not limited to the foregoing. The system memory 1104 and mass storage device 1107 described above may be collectively referred to as memory.

According to various embodiments of the present application, the computer device 1100 may also operate as a remote computer connected to a network via a network, such as the Internet. That is, the computer device 1100 may connect to the network 1112 through the network interface unit 1111 that is coupled to the system bus 1105, or may connect to other types of networks or remote computer systems (not shown) using the network interface unit 1111.

The memory also includes one or more programs stored in the memory and configured to be executed by the one or more central processing units 1101.

The present application further provides a computer-readable storage medium having at least one instruction, at least one program, a set of codes, or a set of instructions stored therein, which is loaded and executed by a processor to implement the medical task assigning method provided by any of the above-mentioned exemplary embodiments.

Embodiments of the present application provide a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the terminal reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the terminal executes the medical task assignment method provided in the above-mentioned alternative implementation.

It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.

The above description is only exemplary of the present application and should not be taken as limiting, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A medical task assignment method, the method comprising:

assigning the target medical task to the target physician subject.

2. The method of claim 1, wherein the physician-assignment rule is the physician-status rule;

determining a target doctor object from at least one candidate doctor object corresponding to the target medical role based on the doctor allocation rule, wherein the determining includes:

determining at least one of the candidate doctor objects based on the target medical role;

acquiring candidate doctor state information corresponding to each candidate doctor object;

and in response to the candidate doctor state information meeting the doctor state rule, determining a candidate doctor object corresponding to the candidate doctor state information as the target doctor object.

3. The method of claim 1, wherein the physician allocation rule is the historical allocation rule;

acquiring candidate doctor state information corresponding to each candidate doctor object and target task attribute information corresponding to the target medical task;

inputting the candidate doctor state information, the target task attribute information and historical distribution data into a task distribution model to obtain the target doctor object output by the task distribution model, wherein the task distribution model is used for analyzing the historical distribution rule represented by the historical distribution data and determining the target doctor object based on the historical distribution rule.

4. The method of claim 3, wherein said entering said candidate physician state information, said target task attribute information, and historical assignment data into a task assignment model to obtain said target physician object output by said task assignment model comprises:

5. The method of any of claims 1 to 4, wherein before determining the target medical role corresponding to the target medical task from the candidate medical roles based on the target task capability representation corresponding to the target medical task, the method further comprises:

acquiring the medical task execution capacity corresponding to each candidate doctor object;

dividing the candidate doctor objects into different doctor object sets based on the medical task execution capacity, wherein the candidate doctor objects in the same doctor object set have at least one same medical task execution capacity;

determining at least one of the candidate medical roles based on the respective set of physician objects.

6. The method of claim 5, wherein determining the target medical role corresponding to the target medical task based on the target task capability representation corresponding to the target medical task comprises:

acquiring the target task capability portrait corresponding to the target medical task;

in response to the target task capability representation matching a medical task performance capability corresponding to the candidate medical role, determining the candidate medical role as the target medical role.

7. The method of any of claims 1 to 4, wherein after said assigning said target medical task to said target physician subject, said method further comprises:

and responding to the completion of the target medical task allocation, updating a task allocation state corresponding to the target medical task, and updating target doctor state information corresponding to the target doctor object.

8. The method of any of claims 1 to 4, wherein the historical allocation rules are determined by a task allocation model;

before determining the target medical role corresponding to the target medical task based on the target task capability portrait corresponding to the target medical task, the method further includes:

acquiring ith sample task attribute information, ith sample doctor state information and ith sample historical allocation data, wherein the ith sample task attribute information is task attribute information corresponding to an ith sample medical task, the ith sample medical task belongs to a sample medical task sequence, the ith sample doctor state information is doctor state information corresponding to an ith sample doctor object, the ith sample doctor object is a doctor object with the medical task execution capacity corresponding to the ith sample medical task, and i is a positive integer;

inputting the ith sample task attribute information, the ith sample doctor state information and the ith sample historical distribution data into the task distribution model to obtain an ith distribution doctor object corresponding to the ith sample medical task output by the task distribution model;

and responding to the completion of the sample medical task sequence distribution, and performing reinforcement learning on the task distribution model based on a distribution doctor object corresponding to the sample medical task sequence.

9. The method of claim 8, wherein the task assignment model is reinforcement learned based on assigned physician objects corresponding to the sample medical task sequence, and comprises at least one of:

taking the task allocation balance of the sample medical task sequence as a training index, and performing reinforcement learning on the task allocation model based on an allocated doctor object corresponding to the sample medical task sequence;

and taking the sample doctor objects occupying the target number by the sample medical task sequence as training indexes, and performing reinforcement learning on the task allocation model based on the allocated doctor objects corresponding to the sample medical task sequence.

10. The method of claim 9, wherein the step of performing reinforcement learning on the task allocation model based on the allocated doctor object corresponding to the sample medical task sequence by using task allocation balance of the sample medical task sequence as a training index comprises:

11. The method of claim 9, wherein the step of performing reinforcement learning on the task allocation model based on the allocated doctor objects corresponding to the sample medical task sequence by using the sample doctor objects occupying the target number of the sample medical task sequence as training indexes comprises:

12. The method of claim 8, wherein after the i-th sample task attribute information, the i-th sample doctor status information, and the i-th sample historical assignment data are input into the task assignment model to obtain an i-th assigned doctor object corresponding to the i-th sample medical task output by the task assignment model, the method further comprises:

and determining the (i + 1) th sample task attribute information, the (i + 1) th sample doctor state information and the (i + 1) th sample historical distribution data based on the (i) th distribution doctor object corresponding to the (i) th sample medical task.

13. The method of any of claims 1 to 4, wherein determining a target medical role from the at least one candidate medical role corresponding to the target medical role based on the medical assignment rule comprises:

acquiring a target distribution index corresponding to a target medical task sequence, wherein the target medical task sequence comprises the target medical task;

determining a target task allocation model based on the target allocation indexes, wherein different task allocation models correspond to different training indexes;

determining the target doctor object from at least one candidate doctor object corresponding to the target medical role based on the target task allocation model.

14. A medical task assigning apparatus, characterized in that the apparatus comprises:

15. A computer device, characterized in that the computer device comprises a processor and a memory, wherein at least one program is stored in the memory, and the at least one program is loaded and executed by the processor to implement the medical task assigning method according to any one of claims 1 to 13.

16. A computer-readable storage medium, in which at least one program is stored, which is loaded and executed by a processor to implement a medical task assigning method according to any one of claims 1 to 13.

17. A computer program product, characterized in that it comprises computer instructions stored in a computer readable storage medium, from which a processor of a computer device reads said computer instructions, which processor executes to implement the medical task assigning method according to any one of claims 1 to 13.