Medical service resource optimal configuration method based on user behaviors

Abstract

The invention relates to a user behavior-based medical service resource optimal configuration method, which comprises the following steps of: acquiring historical doctor seeing data and real-time doctor seeing data of a patient, predicting future traffic through a linear regression model, and acquiring a traffic predicted value; acquiring a peak time period identifier from the flow predicted value, optimizing a doctor scheduling scheme and a medical equipment use scheme by adopting a genetic algorithm, and adjusting a doctor personnel allocation proportion and a medical equipment allocation proportion to obtain an optimized scheduling table; and based on the optimized scheduling table, monitoring the actual flow of the patient in real time, and if the actual flow deviates from the predicted value, correcting the scheduling adjustment scheme. The hospital resource utilization rate is remarkably improved, the waiting time of the patient is shortened, and dynamic and accurate medical resource management is achieved.

Description

Technical Field

[0001] This invention relates to the field of medical service technology, and in particular to a method for optimizing the allocation of medical service resources based on user behavior. Background Technology

[0002] Optimizing the allocation of healthcare resources is a key area for improving the operational efficiency of medical institutions and patient satisfaction. Against the backdrop of an aging population and ever-increasing healthcare demands, medical institutions face the challenge of efficiently utilizing limited resources to meet the diverse needs of patients. Rational allocation of healthcare resources can not only shorten patient waiting times but also improve the quality of diagnosis and treatment and resource utilization, which is of great significance for promoting the sustainable development of the healthcare service system. However, many medical institutions still face difficulties in resource allocation and urgently need data-driven methods based on user behavior to achieve precise management. Existing resource allocation methods often rely on static experience-based judgments or simple historical data analysis, making it difficult to adapt to the dynamic changes in patient behavior and individual differences. For example, many hospitals schedule appointments according to fixed timetables, ignoring the diversity of patient visiting habits, resulting in overcrowding at certain times and idle resources at other times. Furthermore, existing methods often lack in-depth analysis of the deep characteristics of patient behavior, making it difficult to achieve dynamic matching of service supply and demand. This extensive management approach not only reduces resource utilization efficiency but may also negatively impact the patient experience due to excessively long waiting times. The core technical challenge lies in accurately capturing and analyzing the dynamic characteristics of patient behavior and transforming them into actionable resource allocation strategies. First, the dynamic nature of patient behavior manifests in complex variations in appointment times, departmental preferences, and treatment adherence. For example, peak appointment times differ significantly between weekdays and weekends, making a single scheduling strategy insufficient. Second, this dynamism requires the system to process large amounts of heterogeneous data rapidly and generate accurate resource allocation plans quickly. The complexity of patient behavior necessitates data analysis considering multi-dimensional characteristics such as age, disease type, and appointment frequency; however, the real-time integration and analysis of these characteristics is often limited by computational complexity and data processing latency. Therefore, how to predict patient flow in real time based on the dynamic characteristics of patient behavior and optimize physician scheduling and equipment usage has become a key issue in the allocation of healthcare resources. For example, in a general hospital, the emergency department is often overloaded on weekend evenings due to a surge in patients, while some equipment remains idle during the day. How to accurately predict flow and dynamically adjust resource allocation by analyzing the dynamic changes in patient behavior has become a pressing technical challenge. By solving these problems, healthcare institutions can allocate resources more efficiently, improving service quality and patient satisfaction. This requires not only overcoming the real-time bottleneck of data processing, but also establishing an intelligent configuration mechanism that can adapt to the dynamic changes in patient behavior, providing technical support for the sustainable development of the healthcare system. Summary of the Invention

[0003] The purpose of this invention is to propose a method for optimizing the allocation of medical service resources based on user behavior, so as to solve the problems existing in the prior art.

[0004] To achieve the above objectives, the present invention provides the following solution: User behavior-based methods for optimizing the allocation of healthcare service resources include: Obtain patients' historical and real-time medical data, predict future patient flow using a linear regression model, and obtain the predicted flow value. From the traffic forecast values, peak time period identifiers are obtained. Genetic algorithms are used to optimize the doctor scheduling scheme and medical equipment usage scheme, and the doctor staff allocation ratio and medical equipment allocation ratio are adjusted to obtain the optimized schedule. Based on the optimized schedule, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the schedule adjustment plan is revised.

[0005] Optionally, obtaining traffic prediction values ​​includes: Multidimensional features, including age, disease type, and frequency of visits, are extracted from patients' historical medical data. Cluster analysis is then used to group similar behavioral patterns to obtain a set of behavioral categories. Based on the behavioral classification set, a time series analysis model is used to process dynamic changes, determine the peak trend of medical visits in each time period, and obtain the peak trend sequence. Based on the peak trend sequence, future traffic is predicted using a linear regression model. By integrating historical data and real-time medical data, a traffic prediction value is obtained.

[0006] Optionally, multidimensional features, including age, disease type, and frequency of visits, can be extracted from patients' historical medical data, and cluster analysis can be used to group similar behavioral patterns, including: The system extracts patient history data, analyzes the records, extracts age features, disease types, and frequency of visits, and generates a multidimensional feature dataset. By standardizing the multidimensional feature dataset and using the mean normalization method, a normalized feature set is obtained. Based on the normalized feature set, the K-means clustering algorithm is applied to group similar behavioral patterns and determine the behavioral classification set.

[0007] Optionally, based on the behavioral classification set, a time series analysis model is used to process dynamic changes and determine the peak visit trend for each time period, resulting in a peak trend sequence including: Using a pre-defined time series analysis model, dynamic change data is obtained from the behavioral classification set to generate a series of medical visit data for each time period; If the fluctuation range of the medical visit data sequence exceeds the preset threshold, the sequence is analyzed by an autoregressive moving average model to determine the peak points of medical visits in each time period. Based on the identified peak visit times, a trend sequence containing peak periods is generated using a time window segmentation method. Based on the trend sequence, the periodic trend is extracted using the time series decomposition method to generate the final peak trend sequence.

[0008] Optionally, based on the peak trend sequence, predicting future traffic using a linear regression model includes: If the peak trend sequence exceeds the preset threshold, the sequence is analyzed using the threshold judgment method to determine whether to trigger the prediction process. Based on the threshold judgment results, a data fusion method is used to integrate historical medical data and real-time medical data to obtain a fused dataset; The fused dataset is trained using a linear regression model to obtain the trained prediction model; The trained prediction model is used to analyze the peak trend sequence to obtain traffic flow prediction values.

[0009] Optionally, using genetic algorithms to optimize doctor scheduling and medical equipment usage plans includes: Peak period identifiers are extracted from traffic forecast values, and traffic characteristics of relevant time periods are extracted through data filtering to obtain a traffic feature set; Based on the traffic feature set, a genetic algorithm is used to iteratively optimize the doctor scheduling scheme to obtain a preliminary scheduling scheme. The medical equipment allocation scheme is optimized by using a genetic algorithm. Combined with the initial scheduling scheme, the equipment usage scheme is adjusted to obtain a resource allocation scheme. Based on the resource allocation plan, the personnel ratio and equipment usage plan are dynamically adjusted to obtain an optimized shift schedule.

[0010] Optionally, based on the optimized schedule, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the schedule adjustment plan is revised, including: To optimize the shift schedule, we assess patient visit rates and medical equipment utilization rates. If patient visit rates or medical equipment utilization rates are below average, we reallocate unscheduled doctors to peak hours and allocate more medical equipment to obtain a shift adjustment plan. Based on the scheduling adjustment plan, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the pre-parameters of the linear regression model are updated to obtain a corrected prediction model. By using the revised prediction model, the newly collected data is processed iteratively to continuously optimize doctor scheduling and medical equipment usage, resulting in dynamic resource allocation.

[0011] Optionally, a revised prediction model is used to iteratively process newly collected data and continuously optimize physician scheduling and medical equipment usage, including: The revised prediction model is used to obtain optimized traffic prediction results; Based on the optimized traffic prediction results, a linear programming algorithm was used to adjust the doctor's shift schedule and medical equipment scheduling scheme to obtain a preliminary resource allocation scheme. If the execution efficiency of the initial resource allocation plan is lower than the preset threshold, the traffic prediction results are updated by collecting feedback data through real-time monitoring, and a revised resource allocation plan is obtained. Based on the revised resource allocation scheme, the data collection frequency is dynamically adjusted to obtain continuously optimized resource scheduling results.

[0012] The beneficial effects of this invention are as follows: This invention first extracts features such as age, disease type, and frequency of visits from patients' historical medical data. It then identifies similar behavioral patterns through cluster analysis, forming behavioral classification sets, and subsequently uses time series analysis to capture peak visit trends. When a peak trend exceeds a threshold, this embodiment uses a linear regression model to predict future traffic, fusing historical and real-time data to generate traffic prediction values. Furthermore, it optimizes doctor scheduling and equipment allocation using a genetic algorithm to generate an efficient scheduling table. If the visit rate or equipment utilization rate is below average, this invention dynamically adjusts unscheduled doctors and equipment to peak hours, generating an optimized configuration scheme. Simultaneously, this invention continuously optimizes resource allocation by iteratively updating the prediction model through real-time monitoring of actual traffic. Ultimately, this invention significantly improves hospital resource utilization, reduces patient waiting time, and achieves dynamic and precise medical resource management. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of the process for optimizing the allocation of medical service resources based on user behavior, according to an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1 As shown, this embodiment proposes a method for optimizing the allocation of medical service resources based on user behavior, including: Obtain patients' historical and real-time medical data, predict future patient flow using a linear regression model, and obtain the predicted flow value. From the traffic forecast values, peak time period identifiers are obtained. Genetic algorithms are used to optimize the doctor scheduling scheme and medical equipment usage scheme, and the doctor staff allocation ratio and medical equipment allocation ratio are adjusted to obtain the optimized schedule. Based on the optimized schedule, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the schedule adjustment plan is revised.

[0018] Furthermore, obtaining traffic forecast values ​​includes: Multidimensional features, including age, disease type, and frequency of visits, are extracted from patients' historical medical data. Cluster analysis is then used to group similar behavioral patterns to obtain a set of behavioral categories. Based on the behavioral classification set, a time series analysis model is used to process dynamic changes, determine the peak trend of medical visits in each time period, and obtain the peak trend sequence. Based on the peak trend sequence, future traffic is predicted using a linear regression model. By integrating historical data and real-time medical data, a traffic prediction value is obtained.

[0019] Furthermore, multidimensional features, including age, disease type, and frequency of visits, were extracted from patients' historical medical data. Cluster analysis was then used to group similar behavioral patterns, including: The system extracts patient history data, analyzes the records, extracts age features, disease types, and frequency of visits, and generates a multidimensional feature dataset. By standardizing the multidimensional feature dataset and using the mean normalization method, a normalized feature set is obtained. Based on the normalized feature set, the K-means clustering algorithm is applied to group similar behavioral patterns and determine the behavioral classification set.

[0020] Furthermore, based on the behavioral classification set, a time series analysis model is used to process dynamic changes and determine the peak visit trend for each time period, resulting in a peak trend sequence including: Using a pre-defined time series analysis model, dynamic change data is obtained from the behavioral classification set to generate a series of medical visit data for each time period; If the fluctuation range of the medical visit data sequence exceeds the preset threshold, the sequence is analyzed by an autoregressive moving average model to determine the peak points of medical visits in each time period. Based on the identified peak visit times, a trend sequence containing peak periods is generated using a time window segmentation method. Based on the trend sequence, the periodic trend is extracted using the time series decomposition method to generate the final peak trend sequence.

[0021] Furthermore, based on the peak trend sequence, future traffic is predicted using a linear regression model, including: If the peak trend sequence exceeds the preset threshold, the sequence is analyzed using the threshold judgment method to determine whether to trigger the prediction process. Based on the threshold judgment results, a data fusion method is used to integrate historical medical data and real-time medical data to obtain a fused dataset; The fused dataset is trained using a linear regression model to obtain the trained prediction model; The trained prediction model is used to analyze the peak trend sequence to obtain traffic flow prediction values.

[0022] Specifically, in this embodiment, patient visit data is obtained from patient history, the record content is parsed, and age features, disease type, and visit frequency are extracted to generate a multidimensional feature dataset. Specifically, in the hospital information system, historical patient visit data is extracted from the electronic medical record database using SQL queries. For example, all records from the past five years are selected, and complete fields such as visit date, diagnosis code, and patient ID are filtered out, resulting in an original dataset containing 1000 patients, with an average of 15 visit records per person. Next, the data is loaded using Python's Pandas library, and age, disease type, and visit frequency are extracted through multidimensional feature engineering. First, the age feature is calculated as the current date minus the age / birth date, with a mean age of 45.6 years and a standard deviation of 12.3. Disease types are clustered into five categories using ICD-10 encoding, such as cardiovascular (0.25%), respiratory (0.18%), etc. Visit frequency is calculated as an average of 3.2 visits per year with a standard deviation of 1.8, forming a standardized feature matrix X with a shape of 1000 rows and 15 columns. The K-Means clustering algorithm was then imported from the Scikit-learn library, with a preset K value of 5. The optimal K=4 was determined by calculating the SSE curve using the elbow method. The iterative process started with random initialization of the centroid and was updated 100 times before convergence. Euclidean distance was used as a similarity measure. Finally, patients were divided into four groups with similar behavioral patterns: Group 1: High-frequency cardiovascular patients with an average frequency of 4.5 times, older age 62.3 years, accounting for 28%; Group 2: Low-frequency respiratory patients with an average frequency of 1.2 times, older age 35.7 years, accounting for 22%; Group 3: Chronic multi-disease type with an average frequency of 3.8 times, older age 51.4 years, accounting for 30%; Group 4: Occasional other types with an average frequency of 0.9 times, older age 28.5 years, accounting for 20%.

[0023] Specifically, in this embodiment, based on the behavioral classification set, after differential stabilization, the fitting parameters p=2 (autoregression order), d=1 (difference order), and q=1 (moving average order) are used. The model AIC value is -1456.3, and the residual white noise test p value is 0.78, confirming its effectiveness. This allows for the prediction of the trend of the number of patients in each time period. For example, the peak number of patients is expected to be 150 people during the morning peak of 8-10 am on weekdays, 120 people during the afternoon peak of 12-2 pm, and 90 people during the evening peak of 18-20 pm on weekends. Next, based on these prediction results, the peak trend of medical visits for each period is determined. For example, the threshold is set to 1.5 times the average, i.e., 180 people / day. The 24-hour series is analyzed by the sliding window algorithm, with an hourly window width of 1 hour and a step size of 0.5 hours. The proportion of time periods where the peak exceeds the threshold is calculated to obtain the peak trend series, such as Monday morning peak trend index 1.8, Tuesday noon peak 1.6, and weekend evening peak 1.2, forming the sequence [1.8, 1.6, 1.4, 1.2, 1.0, 0.8, 1.2].

[0024] Specifically, in this embodiment, when predicting future traffic, the system first detects whether the peak trend sequence exceeds a preset threshold. Taking traffic data per minute (unit: Mbps) as an example, the threshold is set to 1000 Mbps. Traffic data from the past hour is collected to generate a time series, such as [950, 980, 1010, 1020, 1050]. By calculating the mean of the series, 1022 Mbps, it is determined that it exceeds the threshold of 1000 Mbps, triggering the prediction mechanism. Next, a linear regression model is called, trained using historical data (sampling every minute over the past 24 hours, totaling 1440 data points). Assume a segment in the historical data is [900, 920, 950, ..., 1010], corresponding to timestamps [1, 2, 3, ..., 1440]. The linear regression model fitting formula is y=ax+b, where y is the traffic volume and x is the time point. The slope a=0.5 and the intercept b=900 are calculated using the least squares method. After model training, the predicted values ​​are adjusted by combining real-time input data (such as the last 5 minutes [1010, 1020, 1050, 1030, 1040]) through weighted fusion (historical data weight 0.7, real-time data weight 0.3). The predicted traffic for the next minute is y = 0.5 × 1441 + 900 × 0.7 + 1040 × 0.3 = 1931.5 Mbps.

[0025] Furthermore, the use of genetic algorithms to optimize doctor scheduling and medical equipment usage plans includes: Peak period identifiers are extracted from traffic forecast values, and traffic characteristics of relevant time periods are extracted through data filtering to obtain a traffic feature set; Based on the traffic feature set, a genetic algorithm is used to iteratively optimize the doctor scheduling scheme to obtain a preliminary scheduling scheme. The medical equipment allocation scheme is optimized by using a genetic algorithm. Combined with the initial scheduling scheme, the equipment usage scheme is adjusted to obtain a resource allocation scheme. Based on the resource allocation plan, the personnel ratio and equipment usage plan are dynamically adjusted to obtain an optimized shift schedule.

[0026] Specifically, in this embodiment, peak period identifiers are extracted from the predicted traffic values. Assuming the system predicts patient traffic data for the next hour as [120, 150, 180, 220, 250, 200] (unit: people / 15 minutes), analysis reveals that peak traffic values ​​of 220 and 250 occur in time periods t=4 and t=5 (i.e., 45-60 minutes). The system sets a peak threshold of 200; when t=4 and t=5 exceed the threshold, they are automatically marked as peak periods. Next, the system uses a genetic algorithm to optimize doctor scheduling and medical equipment allocation. The genetic algorithm initializes the population size to 50, with individuals representing scheduling schemes encoded as doctor and equipment allocation matrices. For example, [3, 2, 1, 0] represents 3, 2, 1, and 0 doctors allocated to four time periods, respectively; equipment allocation is similar. The fitness function comprehensively considers doctor working hours (maximum 8 hours per person), equipment utilization rate (maximum utilization rate of 80% per CT scanner), and patient waiting time (target less than 15 minutes). Through 100 iterations with a crossover probability of 0.8 and a mutation probability of 0.1, the algorithm arrives at the optimal solution: allocate 4 doctors and 2 CT scanners at t=4, allocate 3 doctors and 2 CT scanners at t=5, and allocate 1 doctor and 1 CT scanner for each of the other time periods. System analysis of historical data (672 data points sampled every 15 minutes over the past 7 days) revealed that the average patient wait time during peak periods was 20 minutes, which is expected to be reduced to 12 minutes after optimization. The final scheduling table is generated, stored in the database, and accessed by the hospital management system. This system automatically adjusts the schedule and pushes it to doctors' terminals, ensuring efficient utilization of medical resources.

[0027] Furthermore, based on the optimized schedule, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the schedule adjustment plan is revised as follows: To optimize the shift schedule, we assess patient visit rates and medical equipment utilization rates. If patient visit rates or medical equipment utilization rates are below average, we reallocate unscheduled doctors to peak hours and allocate more medical equipment to obtain a shift adjustment plan. Based on the scheduling adjustment plan, the actual patient flow is monitored in real time. If the actual flow deviates from the predicted value, the pre-parameters of the linear regression model are updated to obtain a corrected prediction model. By using the revised prediction model, the newly collected data is processed iteratively to continuously optimize doctor scheduling and medical equipment usage, resulting in dynamic resource allocation.

[0028] Furthermore, the revised prediction model is used to iteratively process newly collected data and continuously optimize doctor scheduling and medical equipment usage, including: The revised prediction model is used to obtain optimized traffic prediction results; Based on the optimized traffic prediction results, a linear programming algorithm was used to adjust the doctor's shift schedule and medical equipment scheduling scheme to obtain a preliminary resource allocation scheme. If the execution efficiency of the initial resource allocation plan is lower than the preset threshold, the traffic prediction results are updated by collecting feedback data through real-time monitoring, and a revised resource allocation plan is obtained. Based on the revised resource allocation scheme, the data collection frequency is dynamically adjusted to obtain continuously optimized resource scheduling results.

[0029] This embodiment addresses the problems of low efficiency in hospital resource allocation, inaccurate prediction of peak visit times, and insufficient utilization of equipment and personnel. It achieves precise resource allocation through multi-dimensional data analysis and dynamic optimization. First, this invention extracts features such as age, disease type, and visit frequency from patients' historical visit data. Cluster analysis identifies similar behavioral patterns, forming behavioral classification sets. Time series analysis is then used to capture peak visit trends. When the peak trend exceeds a threshold, this embodiment uses a linear regression model to predict future traffic, integrating historical and real-time data to generate traffic prediction values. A genetic algorithm optimizes doctor scheduling and equipment allocation to generate an efficient scheduling table. If the visit rate or equipment utilization rate is below average, this embodiment dynamically adjusts unscheduled doctors and equipment to peak periods, generating an optimized configuration plan. Simultaneously, this embodiment continuously optimizes resource allocation by iteratively updating the prediction model through real-time monitoring of actual traffic. Ultimately, this embodiment significantly improves hospital resource utilization, reduces patient waiting time, and achieves dynamic and precise medical resource management.

[0030] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for optimizing allocation of medical service resources based on user behavior, characterized in that, The application relates to a medical staff scheduling method and device. The method comprises the following steps: obtaining historical and real-time visit data of patients, predicting future traffic by a linear regression model, and obtaining a traffic prediction value; from the traffic prediction value, obtaining a peak period identifier, optimizing a doctor scheduling scheme and a medical equipment use scheme by a genetic algorithm, adjusting the doctor personnel allocation ratio and the medical equipment allocation ratio, and obtaining an optimized scheduling table; 2. The method for optimizing allocation of medical service resources based on user behavior according to claim 1, characterized in that, based on the optimized scheduling table, real-time monitoring of actual patient traffic is performed, and if the actual traffic deviates from the prediction value, the scheduling adjustment scheme is corrected. The traffic prediction value comprises: extracting multi-dimensional features including age, disease type and visit frequency from the historical visit data of patients, grouping similar behavior patterns by cluster analysis, and obtaining a behavior classification set; according to the behavior classification set, a time series analysis model is used to process dynamic changes, the peak trend of each period is determined, and a peak trend sequence is obtained; 3. The method for optimizing allocation of medical service resources based on user behavior according to claim 2, characterized in that, based on the peak trend sequence, a linear regression model is used to predict future traffic, and the traffic prediction value is obtained by fusing historical data and real-time visit data. extracting multi-dimensional features including age, disease type and visit frequency from the historical visit data of patients, grouping similar behavior patterns by cluster analysis comprises: obtaining visit data from patient history, analyzing the record content, extracting age features, disease types and visit frequencies, and generating a multi-dimensional feature data set; standardizing the multi-dimensional feature data set by using a mean normalization method to obtain a normalized feature set; 4. The method for optimizing allocation of medical service resources based on user behavior according to claim 2, wherein, according to the normalized feature set, a K-means clustering algorithm is applied to group similar behavior patterns and determine a behavior classification set. According to the behavior classification set, a time series analysis model is used to process dynamic changes, the peak trend of each period is determined, and a peak trend sequence is obtained, which comprises: using a preset time series analysis model to obtain dynamic change data from the behavior classification set, and generating visit data sequences of each period; if the fluctuation amplitude of the visit data sequence exceeds a preset threshold, the sequence is analyzed by using an autoregressive moving average model to determine the peak point of each period; according to the determined peak point, a time window division method is used to generate a trend sequence containing the peak period; 5. The method for optimizing allocation of medical service resources based on user behavior according to claim 2, wherein, according to the trend sequence, a time series decomposition method is used to extract a periodic trend, and a final peak trend sequence is generated. Based on the peak trend sequence, a linear regression model is used to predict future traffic, which comprises: if the peak trend sequence exceeds a preset threshold, the sequence is analyzed by using a threshold judgment method to determine whether the prediction process is triggered; according to the threshold judgment result, a data fusion method is used to integrate the historical visit data and the real-time visit data to obtain a fusion data set; the fusion data set is trained by using a linear regression model to obtain a trained prediction model; 6. The method for optimizing configuration of medical service resources based on user behavior according to claim 1, characterized in that, the trained prediction model is used to analyze the peak trend sequence to obtain a traffic prediction value. The genetic algorithm is used to optimize the doctor scheduling scheme and the medical equipment use scheme, which comprises: extracting the peak period identifier from the traffic prediction value, extracting the traffic features of the related time period by data screening, and obtaining a traffic feature set; according to the traffic feature set, a genetic algorithm is used to iteratively optimize the doctor scheduling scheme to obtain a preliminary scheduling scheme; The genetic algorithm is used to optimize the medical equipment allocation scheme, and the equipment use scheme is adjusted in combination with the preliminary scheduling scheme to obtain a resource allocation scheme; According to the resource allocation scheme, the personnel ratio and equipment use scheme are dynamically adjusted to obtain an optimized scheduling table.

7. The method for optimizing configuration of medical service resources based on user behavior according to claim 1, characterized in that, Based on the optimized scheduling table, the actual patient flow is monitored in real time, and if the actual flow deviates from the predicted value, the scheduling adjustment scheme is corrected, including: For the optimized scheduling table, the patient visit rate and medical equipment usage rate are determined, and if the patient visit rate is lower than the average level or the medical equipment usage rate is lower than the average level, the un-scheduled doctors are re-allocated to the peak period, and more medical equipment is allocated to obtain a scheduling adjustment scheme; According to the scheduling adjustment scheme, the actual patient flow is monitored in real time, and if the actual flow deviates from the predicted value, the linear regression model pre-parameters are updated to obtain a corrected prediction model Using the corrected prediction model, the newly collected data is iteratively processed to continuously optimize the doctor scheduling and medical equipment use, and a dynamic resource allocation result is obtained. 8.The method of claim 7, wherein, Using the corrected prediction model, the newly collected data is iteratively processed to continuously optimize the doctor scheduling and medical equipment use, including: Using the corrected prediction model, the optimized flow prediction result is obtained; Using the linear programming algorithm, the doctor scheduling plan and medical equipment scheduling scheme are adjusted based on the optimized flow prediction result to obtain a preliminary resource allocation scheme; If the execution efficiency of the preliminary resource allocation scheme is lower than the preset threshold, feedback data is collected through real-time monitoring, the flow prediction result is updated, and a corrected resource allocation scheme is obtained; According to the corrected resource allocation scheme, the data collection frequency is dynamically adjusted to obtain a continuously optimized resource scheduling result.

Images