A remote diagnosis service optimization method and device

Abstract

This application provides a method and apparatus for optimizing remote medical services, relating to the field of smart healthcare technology. The method includes: constructing a medical service simulation model, which comprises a coupled micro-behavioral simulation layer and a discrete event simulation layer; using the simulation model to simulate the medical service process under different remote medical scenarios and resource allocation strategies, obtaining simulation results; evaluating the effectiveness of the simulation results under different remote medical scenarios and resource allocation strategies from multiple dimensions, and determining a remote medical service optimization strategy based on the effectiveness evaluation results. By adopting the above-mentioned method and apparatus for optimizing remote medical services, the problem of low accuracy in remote medical service optimization schemes is solved.

Description

Technical Field

[0001] This application relates to the field of smart healthcare technology, and more specifically, to a method and apparatus for optimizing remote diagnosis and treatment services. Background Technology

[0002] With the rapid development of medical informatization, smart healthcare and telemedicine have become important directions for improving service quality and efficiency. However, due to issues such as uneven digital infrastructure development between urban and rural areas, imbalanced allocation of medical resources, differences in digital literacy among the population, and insufficient collaboration in the diagnosis and treatment process, telemedicine generally suffers from low efficiency, insufficient fairness, and a lack of quantitative basis for resource investment. How to scientifically evaluate and accurately optimize telemedicine services has become a key issue that urgently needs to be addressed. Currently, existing methods for evaluating and optimizing telemedicine typically involve collecting and analyzing telemedicine service data, and then formulating optimization plans based on the analysis results.

[0003] However, existing technologies cannot take into account both micro-level behaviors and macro-level processes, nor can they integrate digital infrastructure, resource allocation, population stratification, and treatment processes. This makes it difficult to predict the effectiveness of solutions, resulting in low accuracy of remote medical service optimization solutions. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a method and apparatus for optimizing remote medical services, so as to overcome at least one of the above-mentioned defects.

[0005] In a first aspect, embodiments of this application provide a method for optimizing remote medical services, including: A simulation model for diagnosis and treatment services is constructed, which includes a coupled micro-behavioral simulation layer and a discrete event simulation layer. Using a simulation model of medical services, we simulate the medical service process under different remote medical scenarios and different resource allocation strategies, and obtain simulation results. The simulation results of different remote diagnosis and treatment scenarios and different resource allocation strategies are evaluated from multiple dimensions, and optimization strategies for remote diagnosis and treatment services are determined based on the evaluation results.

[0006] In an optional implementation, the micro-behavioral simulation layer includes a patient agent and a doctor agent. Utilizing a medical service simulation model, it simulates the medical service process under different remote medical scenarios and resource allocation strategies to obtain simulation results. The steps include: in the micro-behavioral simulation layer, responding to the patient agent generating a patient medical request based on the patient's own attributes, selecting the target doctor agent corresponding to the patient's medical request; the target doctor agent, based on the patient's own attributes, determining the medical service duration and medical decision, and sending the medical information to the discrete event simulation layer, the medical information including the patient's own attributes, the medical service duration, and the medical decision; in the discrete event simulation layer, adding the medical task to the target doctor agent's process queue based on the patient's own attributes, generating a medical node sequence based on the medical decision, and determining the actual time consumption of each medical node; simulating each medical node in the medical node sequence to determine the medical result, and sending the medical result to the micro-behavioral simulation layer.

[0007] In an optional implementation, the patient's own attributes include the complexity of their condition, and the duration of the diagnosis and treatment service is determined in the following way: the duration of the diagnosis and treatment service is determined based on the complexity of the condition, the basic duration of diagnosis and treatment, and the efficiency of the doctor's work. The basic duration of diagnosis and treatment is determined based on the department.

[0008] In an optional implementation, after simulating each diagnosis node in the diagnosis node sequence to determine the diagnosis result, the method further includes: using the diagnosis result to update the agent attributes in the micro-behavior simulation layer.

[0009] In an optional implementation, a patient treatment request is generated based on the patient's own attributes in the following way: the treatment method is determined using a treatment method selection model, and the patient treatment request is generated based on the treatment method and the patient's own attributes.

[0010] In an optional implementation, the step of determining the treatment method using the treatment method selection model includes: inputting the patient's own attributes into the treatment method selection model to determine the probability of remote treatment; comparing the probability of remote treatment with a random value, and determining the treatment method based on the comparison result.

[0011] In an optional implementation, the actual time consumed by each diagnosis and treatment node includes fixed time and random fluctuation time. The random fluctuation time follows a uniform distribution, and the fixed time of the target diagnosis and treatment node is determined based on the diagnosis and treatment service duration and the digital infrastructure impact coefficient.

[0012] In an optional implementation, the digital infrastructure impact coefficient is determined based on 5G coverage, medical private network bandwidth, and network packet loss rate. The digital infrastructure impact coefficient has a non-linear relationship with 5G coverage, medical private network bandwidth, and network packet loss rate.

[0013] In optional implementations, different telemedicine scenarios include baseline treatment scenarios and investment-expanded treatment scenarios, and different resource allocation strategies include centralized resource allocation strategies and distributed resource allocation strategies.

[0014] Secondly, embodiments of this application also provide a remote diagnosis and treatment service optimization device, the device comprising: The model building module is used to build a simulation model of diagnosis and treatment services, which includes a micro-behavior simulation layer and a discrete event simulation layer that are coupled together. The simulation module is used to simulate the diagnosis and treatment service process under different remote diagnosis and treatment scenarios and different resource configuration strategies using the diagnosis and treatment service simulation model, and obtain simulation results. The results evaluation module is used to evaluate the effectiveness of simulation results under different remote diagnosis and treatment scenarios and different resource configuration strategies from multiple dimensions, and to determine the optimization strategy for remote diagnosis and treatment services based on the effectiveness evaluation results.

[0015] The embodiments of this application bring the following beneficial effects: This application provides a method and apparatus for optimizing remote medical services. It employs bidirectional coupling modeling of a micro-behavioral simulation layer and a discrete event simulation layer, simultaneously considering both the heterogeneous behavior of patients at the micro-level and the constraints of the macro-level treatment process. This accurately simulates the remote medical process, significantly improving the accuracy and dynamism of remote medical service evaluation. Furthermore, it supports multi-strategy simulation under different remote medical scenarios, accurately predicting the effectiveness of remote medical service optimization schemes and providing reliable and accurate data support for optimizing remote medical services. Compared with existing remote medical service optimization methods, this solves the problem of low accuracy in remote medical service optimization schemes.

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart of the remote diagnosis and treatment service optimization method provided in the embodiments of this application is shown; Figure 2 A flowchart illustrating the simulation steps provided in the embodiments of this application is shown; Figure 3 A schematic diagram of the structure of the remote diagnosis and treatment service optimization device provided in an embodiment of this application is shown; Figure 4 A schematic diagram of the structure of the electronic device provided in the embodiments of this application is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0020] To facilitate understanding of this embodiment, the following description uses the remote diagnosis and treatment service optimization method provided in this application embodiment applied to a terminal device as an example to illustrate the above exemplary steps provided in this application embodiment.

[0021] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for optimizing remote medical services provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the method for optimizing remote diagnosis and treatment services includes: Step S101: Construct a simulation model for medical services.

[0022] The medical service simulation model is used to simulate the remote medical service process, while capturing micro-level individual behavior and macro-level medical process constraints, realizing two-way data feedback, and solving the evaluation bias problem of single modeling methods. The medical service simulation model includes a micro-behavior simulation layer, a discrete event simulation layer, and an interactive feedback layer, with the micro-behavior simulation layer and the discrete event simulation layer being coupled to each other.

[0023] In one embodiment, in order to achieve comprehensive and high-precision evaluation modeling, medical service data that can cover the supply, demand, basic support and medical process of remote diagnosis and treatment can be collected to ensure the representativeness and comprehensiveness of the data, so as to improve the accuracy of the medical service simulation model and the simulation process.

[0024] Medical service data includes multiple categories, such as medical resource distribution data, patient behavior data, digital infrastructure data, and medical process data.

[0025] The data on the distribution of medical resources can be obtained from health statistics yearbooks, annual reports of medical institutions, and special surveys. The data includes, but is not limited to, the number of practicing physicians, number of beds, configuration of telemedicine equipment, and drug availability at different levels (urban / rural / regional / institution), and is further subdivided into departments such as internal medicine, surgery, radiology, and general practice. Patient behavior data can be obtained by extracting anonymized medical records from the hospital information system and combining them with stratified online questionnaires and offline interviews. Patient behavior data includes, but is not limited to: patient demographic characteristics, disease type, annual frequency of visits, online registration rate, remote usage rate, digital literacy, initial satisfaction, and other data.

[0026] Digital infrastructure data was obtained from surveys conducted by the Ministry of Industry and Information Technology, telecommunications operators, and hospital information departments. This data includes, but is not limited to, 5G / 4G network coverage, medical private network bandwidth, system response time, and penetration rate of intelligent medical equipment, with different parameters for urban and rural areas.

[0027] The diagnosis and treatment process data is collected from hospital operation reports, outpatient logs, and data interfaces of logistics partners. The diagnosis and treatment process data includes, but is not limited to, core process time parameters such as average queuing time for online / offline diagnosis and treatment, communication time, examination appointment time, drug delivery time, and referral approval time.

[0028] In one embodiment, after acquiring medical service data, the medical service data can be preprocessed to standardize the heterogeneity, redundancy, and anomalies of the medical service data, thereby ensuring the validity of the model input data.

[0029] Preprocessing includes deduplication, numerical validation, outlier handling, and categorical variable coding. Deduplication removes duplicate records from medical resource data, such as deleting duplicate entries in patient behavior data based on patient ID. Numerical validation limits the reasonable range of core indicators, such as a doctor count greater than 0, bed count greater than or equal to 0, drug availability rate, 5G coverage rate, and intelligent medical device penetration rate within the range [0,1], and network bandwidth greater than 0. Outlier handling corrects or removes outliers that exceed the specified range, for example, limiting patient age to 18-85 years, digital literacy scores and initial satisfaction scores to the range [0,100], annual visit frequency greater than 0, and online registration rate and remote usage rate within the range [0,1]. Categorical variable coding digitizes non-numerical variables, such as setting urban areas to 1 and rural areas to 0 in region type; setting low-income levels to 0, middle-income levels to 1, and high-income levels to 2 in income level; and setting chronic diseases to 1 and acute diseases to 0 in disease type.

[0030] In one embodiment, the preprocessed medical service data can also be stratified and dynamically updated.

[0031] For example, a stratified sampling mechanism is used to stratify patient behavior data by region, age, income, and disease type to ensure sample representativeness. Simultaneously, a dynamic update mechanism is established to update the four core data categories at preset intervals, ensuring the medical service simulation model always reflects changes in actual treatment scenarios and improving the timeliness of the assessment.

[0032] In this embodiment, the micro-behavior simulation layer includes a patient agent, a doctor agent, interaction rules, and decision rules.

[0033] In one embodiment, both the patient agent and the doctor agent are built on the Mesa framework, and each agent is given heterogeneous attributes.

[0034] When building patient and doctor agents based on the Mesa framework, the first step is to build the Mesa model main class and define the Mesa model parameters, which include the number of agents, medical resource nodes, and simulation time step.

[0035] Secondly, patient and doctor agent classes are constructed, both inheriting from Mesa's Agent class. These classes encapsulate agent attributes and decision-making methods. All agent attributes are presented as numerical indicators. If an attribute is fuzzy, it is converted into a quantifiable value. For example, if the attribute is a medical preference, the range is [0.7, 1] for remote medical preference, [0, 0.3] for offline medical preference, and [0.3, 0.7] for no preference. During agent initialization, heterogeneous agent attributes can be generated using Monte Carlo simulation to simulate the characteristic distribution of a real population. For instance, digital literacy scores follow a normal distribution, and the average digital literacy score of rural patients is 20 points lower than that of urban patients.

[0036] After constructing the agent class, a scheduler also needs to be constructed. The scheduler can schedule the behavior of agents according to the agent type to simulate the interaction between doctors and patients.

[0037] Finally, a data collector is built to collect real-time behavioral data (such as remote selection rate and treatment service duration) and status data (such as patient satisfaction and doctor workload) of the intelligent agent. Simultaneously, the simulation time step is set to 1 minute to match the time granularity of the treatment process, and the interaction results are output to the event queue of the discrete event simulation layer.

[0038] In one embodiment, the decision-making method of the patient agent includes a treatment selection method, which is implemented through a treatment method selection model.

[0039] For example, a treatment mode selection model can be constructed based on a logistic regression model. This model uses symptom severity, digital infrastructure score, digital literacy score, and medical preference as independent variables, and the probability of remote medical treatment as the dependent variable. The treatment mode selection model can be represented as follows: ; In the above formula, P represents the probability of remote diagnosis and treatment; The severity of symptoms; The overall score for digital infrastructure is [0, 100]. The score represents digital literacy, with a value range of [0, 100]. This represents medical treatment preference, with a value range of [0,1]. to These are the regression coefficients obtained from model training.

[0040] If the severity of symptoms is mild, then If the severity of symptoms is moderate, then If the severity of the symptoms indicates an acute condition, then And when the severity of symptoms is acute, mandatory If the value is 0, then the treatment method is directly determined to be offline treatment.

[0041] In one embodiment, the decision-making method of the doctor's intelligent agent includes a treatment decision-making method, which may refer to determining the duration of treatment services and making treatment decisions based on the patient's own attributes and doctor information. The doctor information includes the doctor's work efficiency.

[0042] For example, the consultation service duration is the ratio of the product of the basic consultation duration and the disease complexity coefficient to the doctor's work efficiency coefficient. The basic consultation duration is set according to the department, such as 5 minutes / person for general practice and 10 minutes / person for internal medicine.

[0043] In one embodiment, the micro-behavior simulation layer further includes interaction rules, which include matching rules, interaction termination conditions, and interaction result quantification.

[0044] The matching rule refers to the intelligent matching of a corresponding doctor based on the department, interaction priority, and doctor's workload after the patient's intelligent agent submits a patient treatment request. The interaction priority is determined based on the treatment method. When the patient's treatment request is for offline treatment, the interaction priority is determined based on the distance. The closer the distance, the higher the interaction priority. When the patient's treatment request is for online treatment, the interaction priority is determined based on the network quality. The better the network quality, the higher the interaction priority.

[0045] Interaction termination conditions can refer to the termination of providing medical services to patients if the diagnosis or treatment plan is completed, the patient abandons the treatment, or the waiting time exceeds the limit. The waiting time threshold for remote diagnosis and treatment is 15 minutes, and the waiting time threshold for offline diagnosis and treatment is 30 minutes.

[0046] Interaction result quantification can refer to converting interaction results into process trigger parameters. For example, if the remote consultation is successful, the "examination / medication" node of the discrete event simulation layer is triggered. If the consultation fails, the "referral" node is triggered and used as the event input of the discrete event simulation layer.

[0047] In this embodiment, the discrete event simulation layer is built based on the SimPy simulation tool. In the discrete event simulation layer, digital infrastructure parameters (such as 5G coverage) are embedded as core variables into the diagnosis and treatment service process to quantify their impact on the time consumption of each link, thereby accurately simulating the actual situation of key diagnosis and treatment links such as registration, waiting, remote consultation, examination, medication, drug delivery, and referral.

[0048] When building a discrete event simulation layer based on the SimPy simulation tool, the SimPy environment is first defined, and the total simulation duration is set, such as 30 days, and the simulation is performed 24 hours a day. At the same time, a resource class needs to be built to simulate limited resources such as doctors, remote devices, and medicine delivery vehicles, and the resource capacity is set, such as the doctor resource capacity of a certain department being the actual number of doctors.

[0049] Then, process functions (Process) for the medical service are written, encapsulating the behavioral logic, actual time calculation, and resource usage rules for each medical node. Simultaneously, the event queues (such as patient medical requests) input from the micro-behavioral simulation layer are converted into SimPy event trigger signals to drive the medical service process simulation. During the execution of the medical service process, real-time operational data for each medical node, such as actual time consumption, resource utilization, and number of waiting users, are collected, performance indicators are calculated, and fed back to the micro-behavioral simulation layer.

[0050] The performance indicators include, but are not limited to: service success rate, waiting timeout rate, and resource utilization rate. The service success rate is the ratio of the number of patients who successfully completed treatment to the total number of patients who applied for treatment; the waiting timeout rate is the ratio of the number of patients who gave up treatment due to exceeding the waiting time to the total number of patients waiting for treatment; and the resource utilization rate is the ratio of the actual time resources were used to the total available time resources.

[0051] In one embodiment, the discrete event simulation layer is used to break down the diagnosis and treatment service into diagnosis and treatment nodes and construct a time function for each diagnosis and treatment node, and use the time function to calculate the actual time consumption of each diagnosis and treatment node.

[0052] For example, the treatment process can be broken down into multiple nodes based on whether it's online or offline. A time function is then constructed for each node to determine the actual time consumption. This time function includes both fixed and randomly fluctuating time consumption. The fixed time consumption of the target treatment node is determined based on the digital infrastructure impact coefficient; for example, a remote consultation node is considered the target treatment node.

[0053] The random fluctuation time follows a uniform distribution U(0, α), where α represents the fluctuation coefficient. The fluctuation coefficient is set according to the characteristics of the process node. For example, when the diagnosis and treatment node is the registration node, α is 0.5, and when the diagnosis and treatment node is the remote consultation node, α is 1.

[0054] Taking online medical consultation (telemedicine) as an example, the telemedicine process is as follows: registration, remote waiting, remote consultation, examination appointment, and medication delivery. Therefore, the time function for the registration node is: The time function for remote waiting nodes is: T represents the time taken for doctor matching; the time function of the remote consultation node is: K represents the digital infrastructure impact coefficient; the time function for checking the reservation node is: The time function for drug delivery nodes is: If the patient is located in a rural or remote area, additional delivery time will be required. The digital infrastructure impact coefficient is determined based on 5G coverage, medical private network bandwidth, and network packet loss rate. The digital infrastructure impact coefficient K has a non-linear relationship with 5G coverage, medical private network bandwidth (hereinafter referred to as bandwidth), and network packet loss rate. The formula for calculating K is: ; In the above formula, when K is less than 0.5, K is 0.5, which represents the optimal infrastructure; when K is greater than 2, K is 2, which represents the worst infrastructure. K directly affects the time consumed by nodes such as remote consultation, data transmission, and online communication; the larger K is, the longer the time consumed.

[0055] In one embodiment, after constructing a simulation model of medical services, the model parameters can be calibrated.

[0056] Specifically, a combination of genetic algorithms and gradient descent can be used to calibrate model parameters in order to minimize the error between the model output and the actual data.

[0057] For example: First, the root mean square error (RMSE) is used as the objective function.

[0058] In the global search of the genetic algorithm, the core model parameters (such as regression coefficients, K-value weights, and process time coefficients) are used as genes, and the population size, number of generations, crossover probability, and mutation probability are set to search the optimal parameter range globally.

[0059] Meanwhile, in the local optimization of the gradient descent method, the optimal parameter range obtained by the global search of the genetic algorithm is used as the initial value, and the stochastic gradient descent method is used for local fine calibration.

[0060] After each parameter adjustment, the medical service simulation model is run, and the data output by the simulation model is compared with the actual data to determine whether the root mean square error is less than the set threshold. If the root mean square error is greater than or equal to the set threshold, the global search and local optimization are repeated until the error between the macroscopic statistical indicators output by the medical service simulation model and the actual data is less than the set threshold.

[0061] In one embodiment, after constructing the simulation model of the diagnosis and treatment service, it is necessary to verify the effectiveness of the model.

[0062] Specifically, a triple verification mechanism of expert review, data verification, and extreme condition testing can be used to ensure the logical rationality and scenario adaptability of the model.

[0063] For example, when conducting expert reviews, an expert scoring system (out of 100) is established, which scores the model from four dimensions: logical rationality (30 points), simulation scenario fit (25 points), scientific parameter setting (25 points), and coupling mechanism effectiveness (20 points). If the score is greater than or equal to 80 points, the review is deemed to have passed.

[0064] When validating data, time series cross-validation can be used. The historical data is divided into training and validation sets in a 7:3 ratio. The Pearson correlation coefficient and mean absolute percentage error (MAPE) between the model output and the actual data are determined. If the Pearson correlation coefficient is ≥0.85 and the mean absolute percentage error (MAPE) is <10%, the validation is considered successful.

[0065] When conducting extreme condition testing, six types of extreme test cases are constructed, and the expected logical results are clearly defined. If the model output of all extreme test cases is consistent with the expected results, the test is considered to have passed.

[0066] For example, the extreme test case is a complete network outage, with parameters set to 5G coverage = 0, network bandwidth = 0, and K value = 2. The expected logical result is that the success rate of remote diagnosis and treatment services is 0, all patients are switched to offline diagnosis and treatment, and the offline waiting time increases by more than 50%.

[0067] The extreme test case is a sudden decrease in the number of doctors. The parameter is set to a 50% reduction in the number of doctors in a certain department. The expected logical result is that the waiting time for diagnosis and treatment in that department will increase by more than 80%, the service success rate will decrease by more than 30%, and some patients will be referred to other departments.

[0068] The extreme test case is a surge in remote applications. The parameter is set to increase the number of remote medical applications by 200% compared to the baseline. The expected logical result is that the remote waiting time increases by more than 100%, the waiting timeout rate increases by more than 40%, and the doctor's workload is ≥90%.

[0069] In one embodiment, the micro-behavior simulation layer and the discrete event simulation layer adopt an "event-driven and time-synchronized dual coupling mechanism" to ensure the consistency of the simulation of micro-behavior and macro-process.

[0070] Time synchronization refers to the simulation time step being the same for the micro-behavior simulation layer and the discrete event simulation layer. For example, if both are set to 1 minute, real-time synchronization via the simulation clock can avoid feedback deviations caused by time differences.

[0071] Event-driven simulation refers to the immediate triggering of reverse feedback when a diagnosis or treatment node in the discrete event simulation layer is completed or interrupted. The behavior of the agent in the micro-behavior simulation layer is adjusted in real time according to the feedback result, and the adjusted behavior is then used as a new event input to the discrete event simulation layer, forming a closed-loop coupled simulation.

[0072] Step S102: Using a medical service simulation model, simulate the medical service process under different remote medical scenarios and different resource configuration strategies to obtain simulation results.

[0073] Before using the medical service simulation model for simulation, the simulation model needs to be initialized.

[0074] For example, on the one hand, the values ​​of regression coefficients in the model for selecting treatment methods should be clearly defined, as well as the initial weights of the digital infrastructure impact coefficient K, the fixed time consumption of treatment nodes, and the random fluctuation time consumption coefficients. The initial parameters can be set based on industry survey data and expert experience.

[0075] On the other hand, initialize medical resource nodes. For example, based on the actual situation of the county / institution, initialize the resource capacity (such as the number of doctors and equipment), service radius (such as 5km in urban areas and 20km in rural areas) and network parameters (such as 5G coverage and medical private network bandwidth) of the medical resource nodes. Thirdly, initialize the number of agents. For example, set the number of agents according to the ratio of population to medical resources. For instance, if a county has a resident population of 500,000, generate 500 patient agents at a ratio of 1:1000 and 100 doctor agents at a ratio of 1:5000 to ensure that the simulation scale is close to reality.

[0076] The following reference Figure 2 The simulation process will be introduced below. Specifically, a bidirectional coupling feedback between the microscopic behavior simulation layer and the discrete event simulation layer is implemented in the interactive feedback layer, enabling the model to simulate the dynamic changes in real-world diagnostic and treatment scenarios.

[0077] Figure 2 A flowchart illustrating the simulation steps provided in the embodiments of this application is shown, as follows: Figure 2 As shown, the simulation steps include: Step S1021: In the micro-behavior simulation layer, in response to the patient agent generating a patient treatment request based on the patient's own attributes, the target doctor agent corresponding to the patient treatment request is selected.

[0078] The micro-behavioral simulation layer includes patient agents and doctor agents.

[0079] Patient attributes include, but are not limited to: disease type, disease complexity, symptom severity, digital infrastructure parameters, digital literacy score, and healthcare preferences. Digital infrastructure parameters include 5G coverage, medical private network bandwidth, network packet loss rate, system response time, and penetration rate of intelligent medical devices (artificial intelligence penetration rate).

[0080] In the micro-behavioral simulation layer, the patient agent autonomously uses a treatment method selection model to determine the treatment method and generates a patient treatment request based on the treatment method and the patient's own attributes. The treatment method selection model is constructed based on a pre-defined regression model.

[0081] For example, the 5G coverage rate, medical private network bandwidth, network packet loss rate, system response time, and intelligent medical device penetration rate are normalized. At the same time, the time difference between the preset value and the normalized system response time is calculated. The sum of the weights of the time difference, the normalized 5G coverage rate, medical private network bandwidth, network packet loss rate, and intelligent medical device penetration rate is determined as the comprehensive score of digital infrastructure.

[0082] Then, the severity of symptoms, comprehensive digital infrastructure score, digital literacy score, and medical treatment preferences are input into the treatment method selection model to determine the probability of remote treatment. This probability is compared with a random value, and the treatment method is determined based on the comparison result. The preset value can be a set numerical value, such as 1. The random value is a number between 0 and 1. If the probability of remote treatment is greater than the random value, online treatment (remote treatment) is selected; otherwise, offline treatment is selected. The formula for the treatment method selection model has been previously introduced and will not be repeated here.

[0083] The patient intelligent agent generates a patient treatment request based on the patient's own attributes according to the treatment method. For example, it generates an online treatment request for patient A, which records the patient's own attributes.

[0084] In the micro-behavioral simulation layer, the target doctor agent is selected from the departments that match the disease type, with the highest interaction priority and a workload that meets the requirements. The doctor's workload refers to the level of activity of the doctor agent within a unit of time, being occupied and in a state of providing medical services; it is a quantitative indicator used for resource scheduling, priority allocation, and agent attribute updates.

[0085] In step S1022, the target doctor agent determines the duration of the diagnosis and treatment service and the diagnosis and treatment decision based on the patient's own attributes, and sends the diagnosis and treatment information to the discrete event simulation layer.

[0086] Medical information includes patient attributes, doctor information, duration of medical services, and treatment decisions. Doctor information includes doctor work efficiency.

[0087] After receiving a patient's treatment request, the target doctor agent obtains the patient's own attributes. Based on the patient's own attributes, the doctor's work efficiency, and the basic treatment time, the treatment service time is determined. For example, the ratio of the product of the basic treatment time and the disease complexity coefficient to the doctor's work efficiency coefficient is determined as the treatment service time.

[0088] The disease complexity coefficient is determined based on the complexity of the disease; the higher the complexity, the larger the coefficient; conversely, the lower the complexity, the smaller the coefficient. The doctor's work efficiency coefficient is determined based on the doctor's work efficiency; the higher the doctor's work efficiency, the larger the coefficient; conversely, the lower the doctor's work efficiency, the smaller the coefficient.

[0089] At the same time, the doctor's AI agent will also make treatment decisions based on the complexity of the condition, such as whether examinations are needed, whether to refer the patient to another hospital, and whether to prescribe medication.

[0090] After determining the duration of medical services and the treatment decision, the micro-behavioral simulation layer packages the patient's own attributes, doctor information, duration of medical services, and treatment decision into medical information, and sends the medical information to the discrete event simulation layer in the form of medical events.

[0091] Step S1023: In the discrete event simulation layer, the diagnosis and treatment task is added to the process queue of the target doctor agent based on the patient's own attributes, a sequence of diagnosis and treatment nodes is generated based on the diagnosis and treatment decision, and the actual time consumption of each diagnosis and treatment node is determined.

[0092] In the discrete event simulation layer, after receiving a diagnosis and treatment event, the severity of symptoms is extracted from the patient's own attributes. Based on the severity of symptoms, it is determined whether it is an acute, moderate, or mild case. The diagnosis and treatment event is added to the process queue of the target doctor agent in descending order of priority, so as to occupy the limited resources such as the target doctor agent and remote medical equipment, and the discrete event process simulation is started.

[0093] Then, based on the treatment decision, a corresponding treatment node sequence is dynamically generated. For example, if the treatment decision is "examination required", the treatment node sequence is the standard remote treatment process node sequence of online registration, remote waiting, remote consultation, examination appointment, medication pickup, and medication delivery; if the treatment decision is "no examination required", the treatment node sequence is online registration, remote waiting, remote consultation, medication pickup, and medication delivery; if the treatment decision is "referral", a referral application is generated, and the current process ends.

[0094] Meanwhile, the discrete event simulation layer calculates the actual time consumption of each diagnosis and treatment node based on a preset time function and a digital infrastructure influence coefficient K. The time function includes fixed time consumption and random fluctuation time consumption. The fixed time consumption of the target diagnosis and treatment node can be determined based on the diagnosis and treatment service duration in the diagnosis and treatment information. For example, the fixed time consumption of a remote consultation node is determined by multiplying the diagnosis and treatment service duration by the digital infrastructure influence coefficient K.

[0095] Step S1024: Simulate each diagnosis and treatment node in the diagnosis and treatment node sequence to determine the diagnosis and treatment result, and send the diagnosis and treatment result to the micro-behavioral simulation layer.

[0096] For example, during simulation, the actual time consumed by the current treatment node and the corresponding medical resources are used to time the treatment. Corresponding treatment events are triggered based on preset conditions met during the treatment process, such as triggering a timeout if the waiting time exceeds a threshold, or terminating treatment if there are insufficient doctor resources. When the current treatment node is completed, the corresponding medical resources are released, and the simulation continues for the next treatment node, and so on, until all treatment nodes are completed.

[0097] If all treatment nodes are completed, a successful treatment result is generated; if a waiting timeout, insufficient resources, or process failure occurs, a treatment termination / failure result is generated; if the referral conditions are met (such as the complexity of the condition exceeding the department's capacity or the doctor's resources being continuously fully utilized), a referral request event is generated.

[0098] After the simulation is completed, the core performance indicators such as treatment service duration, remote usage, satisfaction score, remote usage, and resource utilization are statistically analyzed in real time based on the treatment results, and the treatment results are sent to the micro-behavioral simulation layer.

[0099] In one embodiment, after determining the diagnosis and treatment results, the agent attributes in the micro-behavior simulation layer can also be updated using the diagnosis and treatment results.

[0100] For example, for the patient's intelligent agent, if the diagnosis results indicate that a waiting timeout has occurred, the digital literacy score will be reduced (e.g., digital literacy score decreases by 5), the remote diagnosis preference will decrease (e.g., remote diagnosis preference decreases by 0.2), and the probability of the next remote diagnosis will decrease (the probability of the next remote diagnosis is multiplied by 0.7). If the diagnosis results indicate that the diagnosis process is smooth (e.g., the total time is less than the industry average, and the total time is the sum of the actual time of all diagnosis nodes), the remote diagnosis preference will increase (e.g., remote diagnosis preference increases by 0.3), and the satisfaction will improve (e.g., satisfaction increases by 10). If the diagnosis results indicate that the drug delivery has exceeded the time limit, the satisfaction will decrease (e.g., satisfaction decreases by 15).

[0101] For the doctor's intelligent agent, if the diagnosis and treatment process takes too long (e.g., the diagnosis and treatment service time is more than twice the basic diagnosis and treatment time), the doctor's work efficiency coefficient will be reduced (e.g., the doctor's work efficiency coefficient will be reduced by 0.1), and job satisfaction will be reduced (e.g., job satisfaction will be reduced by 8); if the resource utilization rate is too high (e.g., greater than 90%), the doctor's workload will be increased (e.g., the doctor's workload will be increased by 0.2), and the diagnosis and treatment service time will be increased (e.g., the diagnosis and treatment service time will be increased by 2 minutes per person); if the remote equipment fails, job satisfaction will be reduced.

[0102] During the simulation process, if the success rate of remote diagnosis and treatment in a certain area is less than 60% (the lowest acceptable threshold in the industry), the overall probability of remote diagnosis and treatment for all patients in that area will be multiplied by 0.8, triggering a "medical resource optimization" signal to update the agent's attributes.

[0103] During the simulation process, if the discrete event simulation layer reports a diagnosis failure (service success rate of 0), a waiting timeout, or a resource utilization rate greater than or equal to 85%, it will trigger the micro-behavior simulation layer to generate a new "referral request" event, so as to re-enter the processing flow of the discrete event simulation layer based on the "referral request" event.

[0104] In one embodiment, different remote diagnosis and treatment scenarios include, but are not limited to: baseline diagnosis and treatment scenarios and investment-expanded diagnosis and treatment scenarios; different resource allocation strategies include, but are not limited to: centralized resource allocation strategy and distributed resource allocation strategy.

[0105] Among them, the benchmark treatment scenario refers to the scenario corresponding to the current level of digital infrastructure and patient behavior characteristics, which is used to simulate the current treatment services; the investment-expanded treatment scenario can refer to the scenario after targeted digital infrastructure optimization in rural or remote areas through investment and development based on the benchmark treatment scenario. The investment-expanded treatment scenario includes the benchmark treatment scenario and the treatment scenario that has not been deployed (planned investment), such as increasing the 5G coverage rate in rural areas by 50%, doubling the network bandwidth, and increasing the patient's digital literacy score by 10 points, to simulate the effect of treatment services after infrastructure investment.

[0106] When conducting simulations under different remote diagnosis and treatment scenarios and different resource allocation strategies, the diagnosis and treatment service process is simulated multiple times in the following scenarios: benchmark diagnosis and treatment scenario with centralized resource allocation strategy, benchmark diagnosis and treatment scenario with distributed resource allocation strategy, investment-expanded diagnosis and treatment scenario with distributed resource allocation strategy, and investment-expanded diagnosis and treatment scenario with distributed resource allocation strategy, to obtain multiple simulation results.

[0107] Step S103: Evaluate the effectiveness of simulation results under different remote diagnosis and treatment scenarios and different resource configuration strategies from multiple dimensions, and determine the optimization strategy for remote diagnosis and treatment services based on the effectiveness evaluation results.

[0108] In this step, through multi-diagnosis and treatment scenario simulation and multi-strategy simulation, the impact of different digital infrastructure investment and resource allocation strategies on remote diagnosis and treatment services can be quantified, providing a quantitative basis for optimization decisions.

[0109] Simulation results refer to the results obtained after simulating different remote diagnosis and treatment scenarios and resource allocation strategies using a diagnosis and treatment service simulation model. Simulation results include, but are not limited to, all quantitative indicators and operational data such as diagnosis and treatment results, diagnosis and treatment completion / termination / referral status, time consumption of each diagnosis and treatment node, physician workload, resource utilization rate, service success rate, waiting timeout rate, and patient satisfaction.

[0110] In one embodiment, the simulation results of different telemedicine scenarios and different resource allocation strategies can be evaluated from four dimensions: treatment efficiency, treatment fairness, satisfaction, and strategy effectiveness.

[0111] Among them, the evaluation of diagnosis and treatment efficiency is the core operational indicator of telemedicine, which quantifies the service efficiency and accessibility. The indicators of diagnosis and treatment efficiency include, but are not limited to: average remote diagnosis time, average remote treatment cycle, rural remote usage rate, average number of people receiving remote services per day, and time spent at each stage of the process.

[0112] The equity of healthcare assessment measures the equitable distribution of healthcare services by comparing the differences in remote access rates among different groups, with a focus on addressing the urban-rural digital divide. Equity of healthcare indicators include, but are not limited to: differences in remote access rates between urban and rural areas, differences in remote access rates among different income groups, and differences in remote access rates among different age groups.

[0113] Satisfaction assessment is used to quantify the experience of medical services from the perspectives of both patients and doctors. Satisfaction indicators include, but are not limited to: average patient satisfaction (determined based on process smoothness, diagnostic efficiency, and service convenience) and average doctor satisfaction (determined based on workload, communication efficiency, and ease of equipment operation).

[0114] The strategy effectiveness dimension assessment is used to compare different resource allocation strategies and quantitatively evaluate the advantages and disadvantages of different resource allocation strategies. Strategy effectiveness indicators include, but are not limited to: average remote diagnosis time under different resource allocation strategies, rural remote usage rate, patient / doctor satisfaction, resource utilization rate, and cross-regional transmission loss rate.

[0115] After determining the effectiveness assessment results, optimization strategies for remote medical services are generated based on the effectiveness assessment results. These optimization strategies include, but are not limited to: digital infrastructure optimization strategies, medical resource allocation strategies, population stratification service strategies, and medical process reengineering strategies.

[0116] Among them, the digital infrastructure optimization strategy is used to clarify the priority of digital infrastructure in rural / remote areas, such as prioritizing the improvement of 5G coverage and expanding the bandwidth of medical private networks; the medical resource allocation strategy is used to recommend centralized / distributed resource allocation schemes adapted to actual scenarios and propose remote equipment configuration standards for primary healthcare institutions; the population stratification service strategy is used to provide operational guidance and simplify remote diagnosis and treatment processes for low-literacy groups based on the quantitative results of patients' digital literacy; and the diagnosis and treatment process reengineering strategy is used to propose solutions such as online and offline integration optimization and drug delivery scheduling optimization for time-consuming process nodes.

[0117] The remote diagnosis and treatment service optimization method provided in this application has the following technical effects: First, a bidirectional coupled architecture of microscopic behavior simulation layer and discrete event simulation layer is adopted to simultaneously characterize the heterogeneous behavior of microscopic patients and the constraints of macroscopic diagnosis and treatment process, solving the problems of inaccurate evaluation and insufficient dynamism of single simulation model, and realizing high-precision simulation of remote diagnosis and treatment in all scenarios.

[0118] Secondly, it covers multiple dimensions of evaluation, including efficiency, fairness, satisfaction, and strategy effectiveness. Combined with genetic algorithms and gradient descent for precise calibration, it reduces the error between model output and real data, effectively supporting scientific decision-making.

[0119] Third, a two-way data closed loop is established between the micro-behavior simulation layer and the discrete event simulation layer. The simulation results are used to correct the behavior and attributes of the intelligent agent in real time, and the impact of policy, resource and infrastructure changes on services can be dynamically deduced, supporting forward-looking configuration.

[0120] Fourth, the impact of quantifiable digital infrastructure and resource allocation on the disparity in remote usage rates between urban and rural areas can be used to provide targeted, actionable strategies for bridging the digital divide and improving fairness.

[0121] Fifth, provide reusable smart healthcare simulation and evaluation tools to reduce resource waste, improve service efficiency and public satisfaction, and assist in urban planning and dynamic regulation.

[0122] Based on the same inventive concept, this application also provides a remote diagnosis and treatment service optimization device corresponding to the remote diagnosis and treatment service optimization method. Since the principle of the device in this application is similar to the remote diagnosis and treatment service optimization method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0123] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of a remote diagnosis and treatment service optimization device provided in an embodiment of this application. Figure 3 As shown, the remote medical service optimization device 200 includes: The model building module 201 is used to build a simulation model of diagnosis and treatment services, which includes a micro-behavior simulation layer and a discrete event simulation layer. The simulation module 202 is used to simulate the diagnosis and treatment service process under different remote diagnosis and treatment scenarios and different resource configuration strategies using the diagnosis and treatment service simulation model, and obtain simulation results. The results evaluation module 203 is used to evaluate the effectiveness of simulation results under different remote diagnosis and treatment scenarios and different resource configuration strategies from multiple dimensions, and to determine the optimization strategy for remote diagnosis and treatment services based on the effectiveness evaluation results.

[0124] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 300 includes a processor 310, a memory 320, and a bus 330.

[0125] The memory 320 stores machine-readable instructions executable by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate via the bus 330. When the machine-readable instructions are executed by the processor 310, they can perform the operations described above. Figure 1 The steps of the remote diagnosis and treatment service optimization method in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0126] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the remote diagnosis and treatment service optimization method in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0127] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0128] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0129] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0130] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0131] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0132] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for optimizing remote medical services, characterized in that, include: A simulation model for diagnosis and treatment services is constructed, which includes a micro-behavioral simulation layer and a discrete event simulation layer that are coupled together. Using the aforementioned medical service simulation model, the medical service process under different remote medical scenarios and different resource configuration strategies is simulated to obtain simulation results. The simulation results of different remote diagnosis and treatment scenarios and different resource allocation strategies are evaluated from multiple dimensions, and optimization strategies for remote diagnosis and treatment services are determined based on the evaluation results.

2. The method according to claim 1, characterized in that, The micro-behavioral simulation layer includes a patient agent and a doctor agent. The step of using the medical service simulation model to simulate the medical service process under different remote medical scenarios and different resource allocation strategies, and obtaining the simulation results, includes: In the micro-behavior simulation layer, in response to the patient agent generating a patient treatment request based on the patient's own attributes, the target doctor agent corresponding to the patient treatment request is selected. The target doctor agent determines the duration of the treatment service and the treatment decision based on the patient's own attributes, and sends the treatment information to the discrete event simulation layer. The treatment information includes the patient's own attributes, the duration of the treatment service, and the treatment decision. In the discrete event simulation layer, the diagnosis and treatment tasks are added to the process queue of the target doctor agent based on the patient's own attributes, a sequence of diagnosis and treatment nodes is generated based on the diagnosis and treatment decisions, and the actual time consumption of each diagnosis and treatment node is determined. The diagnosis and treatment results are determined by simulation of each diagnosis and treatment node in the diagnosis and treatment node sequence, and the diagnosis and treatment results are sent to the micro-behavior simulation layer.

3. The method according to claim 2, characterized in that, The patient's own attributes, including the complexity of their condition, are used to determine the duration of medical services in the following ways: The duration of medical services is determined based on the complexity of the condition, the basic treatment time, and the doctor's work efficiency. The basic treatment time is determined based on the department.

4. The method according to claim 2, characterized in that, After simulating and determining the treatment result for each treatment node in the treatment node sequence, the process further includes: Using the diagnostic results, update the agent attributes in the micro-behavior simulation layer.

5. The method according to claim 2, characterized in that, Patient treatment requests can be generated based on the patient's own attributes in the following ways: The treatment method selection model is used to determine the treatment method; Based on the treatment method and the patient's own attributes, a patient treatment request is generated.

6. The method according to claim 5, characterized in that, The steps for determining the treatment method using the treatment method selection model include: The patient's own attributes are input into the treatment method selection model to determine the probability of remote treatment; The remote diagnosis and treatment probability is compared with a random value, and the diagnosis and treatment method is determined based on the comparison result.

7. The method according to claim 2, characterized in that, The actual time consumed at each diagnosis and treatment node includes fixed time and random fluctuation time. The random fluctuation time follows a uniform distribution, and the fixed time of the target diagnosis and treatment node is determined based on the diagnosis and treatment service duration and the digital infrastructure impact coefficient.

8. The method according to claim 7, characterized in that, The digital infrastructure impact coefficient is determined based on 5G coverage, medical private network bandwidth, and network packet loss rate. The digital infrastructure impact coefficient has a non-linear relationship with 5G coverage, medical private network bandwidth, and network packet loss rate.

9. The method according to claim 1, characterized in that, The different remote diagnosis and treatment scenarios include the baseline diagnosis and treatment scenario and the investment-expanded diagnosis and treatment scenario; the different resource allocation strategies include the centralized resource allocation strategy and the distributed resource allocation strategy.

10. A device for optimizing remote medical services, characterized in that, include: The model building module is used to build a simulation model of diagnosis and treatment services, which includes a micro-behavior simulation layer and a discrete event simulation layer that are coupled together. The simulation module is used to simulate the diagnosis and treatment service process under different remote diagnosis and treatment scenarios and different resource configuration strategies using the diagnosis and treatment service simulation model, and obtain simulation results. The results evaluation module is used to evaluate the effectiveness of simulation results under different remote diagnosis and treatment scenarios and different resource configuration strategies from multiple dimensions, and to determine the optimization strategy for remote diagnosis and treatment services based on the effectiveness evaluation results.

Images