Operation duration prediction and scheduling method based on machine learning

Abstract

The invention discloses an operation duration prediction and scheduling method based on machine learning, and belongs to the technical field of medical care informatics, and the method comprises the following steps: obtaining platform historical processing data of which the operation type is the same as a predicted operation type, and carrying out the analysis to obtain a historical processing difficulty index; a resource consumption index is obtained through analysis, and a duration prediction emergency judgment result is obtained through comprehensive analysis; analyzing to obtain a data transmission channel effect index, analyzing to obtain a final data segmentation judgment result based on the data segmentation preliminary judgment result and the data transmission channel effect index, and performing corresponding data processing to obtain a feature data packet; a corresponding data transmission protocol is selected based on the duration prediction emergency judgment result, so that the feature data packet is transmitted to an operation duration prediction module; and the operation duration prediction module receives the feature data packet, analyzes the feature data packet by using a machine learning algorithm to obtain an operation duration prediction value, and performs automatic scheduling prompt, so that the problem of slow operation duration prediction response in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of healthcare informatics, and particularly to a method for predicting and scheduling surgical duration based on machine learning. Background Art

[0002] With the continuous progress of medical technology and the improvement of informatization level, the accurate prediction of surgical duration has become the key to improving the quality of medical services. As the core link in the medical system, the accurate estimation of its duration has far-reaching significance for the efficient scheduling of hospital resources, the reasonable optimization of surgical arrangements, and the improvement of the patient treatment experience.

[0003] Existing surgical duration prediction systems achieve this by mutually learning and fusing features of different modalities, extracting key important features affecting surgical duration, and inputting them into existing surgical duration prediction models.

[0004] For example, a method for predicting surgical duration based on multi-modal feature fusion disclosed in the invention patent with publication number CN118711779A includes: using a feature initialization module to encode and initialize preoperative data to obtain feature information of different modalities; mutually learning and fusing features of different modalities through a cross-attention feature fusion module to extract key important features affecting surgical duration and obtain multi-modal features; through a multi-modal feature fusion and duration prediction module, fusing the multi-modal features, and using the fused features to predict the duration of the surgery.

[0005] For example, a surgical scheduling system, method, storage medium, and terminal based on deep learning disclosed in the invention patent with publication number CN111951946A includes: a surgical duration prediction module for constructing a surgical duration prediction model and using the surgical duration prediction model to predict the duration data of the current surgery; a surgical duration correction module for correcting the predicted duration data; a surgical resource reservation module for reserving resources related to the current surgery; a surgical scheduling optimization module for performing an optimized scheduling process on the predicted and / or corrected duration data according to a genetic algorithm.

[0006] However, in the process of implementing the technical solutions of the embodiments of the present invention in this application, it is found that the above technologies have at least the following technical problems: Existing medical platforms need to perform large-scale data processing and data analysis, so there will be some phenomena of data parallel processing, which easily leads to excessive consumption or bottlenecks of computing resources, thereby affecting the instant response and accuracy of the prediction model. Therefore, there is a problem of slow response in predicting surgical duration. Summary of the Invention

[0007] The present invention provides a method for predicting and scheduling surgical duration based on machine learning, which solves the problem in the prior art that when the medical system has insufficient resources caused by data concurrency, it is impossible to flexibly adjust resource allocation according to the urgency of duration prediction, resulting in slow response of surgical duration prediction, and realizes accurate and timely prediction of surgical duration.

[0008] The present invention provides a method for predicting and scheduling surgical duration based on machine learning, including the following steps: when the medical information platform receives a surgical duration prediction request, obtain the platform's historical processing data and analyze to obtain a historical processing difficulty index; obtain the resource parameters of the medical information platform within the first preset time period, analyze to obtain a resource consumption index, and comprehensively analyze based on the historical processing difficulty index and the resource consumption index to obtain a duration prediction urgency determination result, where the duration prediction urgency determination result includes duration prediction urgency and duration prediction non-urgency; the medical information platform automatically retrieves prediction feature data, and analyzes according to the duration prediction urgency determination result to obtain a preliminary data segmentation determination result; obtain the data transmission channel effect parameters, analyze to obtain a data transmission channel effect index, analyze based on the preliminary data segmentation determination result and the data transmission channel effect index to obtain a final data segmentation judgment result, and perform corresponding data processing to obtain feature data packets; select a corresponding data transmission protocol based on the duration prediction urgency determination result, and thus transmit the feature data packets to the surgical duration prediction module; the surgical duration prediction module receives the feature data packets, analyzes using a machine learning algorithm to obtain a surgical duration prediction value, and performs an automatic scheduling prompt.

[0009] Further, the steps of when the medical information platform receives a surgical duration prediction request, obtain the platform's historical processing data, and analyze to obtain a historical processing difficulty index are specifically as follows: when the medical information platform receives a surgical duration prediction request, obtain the predicted surgical type, and thus count the historical data of the medical information platform for processing surgeries of the same type as the predicted surgical type, and record it as the platform's historical processing data, where the platform's historical processing data includes historical average data loading duration, historical average CPU usage rate, historical average I / O waiting time, historical average number of input / output operations per second, and historical average GPU video memory occupancy rate; obtain a preset low-performance verification set in the medical database, where the low-performance verification set includes the maximum historical data loading duration, the maximum historical I / O waiting time, and the maximum historical number of input / output operations per second; analyze based on the platform's historical processing data and the low-performance verification set to obtain a historical processing difficulty index; the historical processing difficulty index can be obtained by comparing the historical average data loading duration, historical average I / O waiting time, and historical average number of input / output operations per second with the maximum historical data loading duration, the maximum historical I / O waiting time, and the maximum historical number of input / output operations per second respectively, and combining the historical average CPU usage rate and historical average GPU video memory occupancy rate, and then introducing influence weights and combining them respectively.

[0010] Further, the steps of obtaining the resource parameters of the medical information platform within the first preset period and analyzing to obtain the resource consumption index specifically include: obtaining the resource parameters of the medical information platform within the first preset period, where the resource parameters of the medical information platform include the number of surgical duration prediction requests, the average number of data queries per second, the average response duration, the prediction failure rate, the number of user accesses, and the number of concurrent requests; obtaining the ideal set of platform resources preset in the medical database and analyzing it with the resource parameters of the medical information platform to obtain the resource consumption index; the ideal set of platform resources preset in the medical database includes the ideal value of the number of surgical duration prediction requests, the ideal value of the average number of data queries per second, the ideal value of the average response duration, the ideal value of the prediction failure rate, the ideal value of the number of user accesses, and the ideal value of the number of concurrent requests; the resource consumption index is used to perform a fusion process by introducing corresponding influence weights on the comparison result of comparing the average number of data queries per second with the ideal value of the average number of data queries per second, and the comparison results of comparing the average response duration, the prediction failure rate, the number of user accesses, and the number of concurrent requests with the ideal value of the average response duration, the critical value of the prediction failure rate, the ideal value of the number of user accesses, and the ideal value of the number of concurrent requests respectively to obtain the resource consumption index.

[0011] Further, the steps of comprehensively analyzing based on the historical processing difficulty index and the resource consumption index to obtain the duration prediction emergency determination result specifically include: obtaining the preset historical processing difficulty lower limit value and the resource consumption index lower limit value in the medical information platform; comparing the historical processing difficulty index with the historical processing difficulty lower limit value, and comparing the resource consumption index with the resource consumption index lower limit value to obtain the duration prediction emergency determination result; if the historical processing difficulty index is above the historical processing difficulty lower limit value and the resource consumption index is above the resource consumption index lower limit value, then the duration prediction emergency determination result is that the duration prediction is urgent, otherwise it is that the duration prediction is not urgent; the duration prediction emergency determination result includes that the duration prediction is urgent and that the duration prediction is not urgent.

[0012] Further, the steps of analyzing according to the duration prediction emergency determination result to obtain the preliminary data segmentation determination result specifically include: obtaining the duration prediction emergency determination result, if the duration prediction emergency determination result is that the duration prediction is urgent, then the preliminary data segmentation determination result is to initially perform segmentation, if the duration prediction emergency determination result is that the duration prediction is not urgent, then the preliminary data segmentation determination result is to initially not perform segmentation.

[0013] Further, the steps of obtaining the effect parameters of the data transmission channel and analyzing to obtain the effect index of the data transmission channel specifically include: obtaining the effect parameters of the data transmission channel, where the effect parameters of the data transmission channel include the average network upload speed, average bandwidth, average CPU utilization rate, average channel occupancy rate, and channel error rate within a second preset time period; obtaining the ideal data set of channel effects preset in the medical information platform and analyzing it with the effect parameters of the data transmission channel to obtain the effect index of the data transmission channel; the ideal data set of channel effects includes the ideal value of network upload speed, ideal value of bandwidth, and limit value of channel error rate; the effect index of the data transmission channel is used to introduce an influence weight after comparing the average network upload speed with the ideal value of network upload speed, comparing the average bandwidth with the ideal value of bandwidth, and comparing the channel error rate with the limit value of channel error rate, and combining the average CPU utilization rate and average channel occupancy rate with the introduced influence weight to obtain the effect index of the data transmission channel.

[0014] Further, the steps of analyzing the initial determination result of data segmentation and the effect index of the data transmission channel to obtain the final determination result of data segmentation and performing corresponding data processing to obtain the feature data packet specifically include: obtaining the preset effect threshold of the data transmission channel in the medical information platform and comparing it with the effect index of the data transmission channel to obtain the quality evaluation result of the data transmission channel, where the quality evaluation result of the data transmission channel includes that the quality of the data transmission channel is qualified and the quality of the data transmission channel is unqualified; analyzing the initial determination result of data segmentation and the effect index of the data transmission channel to obtain the final determination result of data segmentation, where the final determination result of data segmentation includes data segmentation and no data segmentation; if the initial determination result of data segmentation is not to perform segmentation initially and the quality evaluation result of the data transmission channel is that the quality of the data transmission channel is qualified, then the final determination result of data segmentation is no data segmentation, otherwise segmentation; if the final determination result of data segmentation is data segmentation, then the data segmentation standard is analyzed and the data segmentation process is performed based on the data segmentation standard to obtain the feature data packet, where the data segmentation standard includes the first data segmentation standard, the second data segmentation standard, and the third data segmentation standard; among them, if the initial determination result of data segmentation is to perform segmentation initially and the quality evaluation result of the data transmission channel is that the quality of the data transmission channel is qualified, then the first data segmentation standard is analyzed and the data segmentation process is performed based on the first data segmentation standard to obtain the feature data packet; if the initial determination result of data segmentation is to perform segmentation initially and the quality evaluation result of the data transmission channel is that the quality of the data transmission channel is unqualified, then the second data segmentation standard is analyzed and the data segmentation process is performed based on the second data segmentation standard to obtain the feature data packet; if the initial determination result of data segmentation is not to perform segmentation initially and the quality evaluation result of the data transmission channel is that the quality of the data transmission channel is unqualified, then the third data segmentation standard is analyzed and the data segmentation process is performed based on the third data segmentation standard to obtain the feature data packet.

[0015] Further, the first data segmentation criterion is analyzed, and data segmentation processing is performed based on the first data segmentation criterion to obtain feature data packets. The specific steps include: analyzing to obtain a comprehensive emergency evaluation index based on the historical processing difficulty index and the resource consumption index; obtaining each preset comprehensive emergency evaluation index interval and the corresponding single - time data transmission standard in the medical information platform, and matching with the comprehensive emergency evaluation index. If the comprehensive emergency evaluation index is within a certain preset comprehensive emergency evaluation index interval, then obtain the corresponding single - time data transmission standard of this interval as the first data segmentation criterion; performing data segmentation processing based on the first data segmentation reference criterion to obtain feature data packets; the comprehensive emergency evaluation index is used to process the combination of the historical processing difficulty index and the historical processing difficulty impact weight and the combination result of the resource consumption index and the resource consumption impact weight to obtain the comprehensive emergency evaluation index.

[0016] Further, the second data segmentation criterion is analyzed, and data segmentation processing is performed based on the second data segmentation criterion to obtain feature data packets. The specific steps include: analyzing based on the comprehensive emergency evaluation index and the data transmission channel effect index to obtain a processing evaluation efficiency index; obtaining the preset processing evaluation efficiency index threshold in the medical information platform, and performing a difference processing with the processing evaluation efficiency index to obtain a processing evaluation efficiency difference, and performing comprehensive processing based on the processing evaluation efficiency difference to obtain a processing evaluation efficiency deviation; obtaining each preset processing evaluation efficiency deviation interval and the corresponding single - time data transmission standard in the medical information platform. If the processing evaluation efficiency deviation is within a certain preset processing evaluation efficiency deviation interval, then obtain the corresponding single - time data transmission standard of this interval as the second data segmentation criterion; performing data segmentation processing based on the analyzed second data segmentation criterion and based on the second data segmentation criterion to obtain feature data packets; the processing evaluation efficiency index is used to characterize the comprehensive complexity of data processing by the comprehensive emergency evaluation index and the data transmission channel effect index.

[0017] Further, based on the duration prediction emergency determination result, the corresponding data transmission protocol is selected, and thus the feature data packets are transmitted to the surgical duration prediction module. The specific steps include: analyzing based on the duration prediction emergency determination result to obtain the corresponding data transmission protocol, and the data transmission protocol includes the UDP protocol and the TCP protocol; if the duration prediction emergency determination result is that the duration prediction is urgent, then use the UDP protocol to upload the prediction feature data to the surgical duration prediction module. If the duration prediction emergency determination result is that the duration prediction is not urgent, then use the TCP protocol to upload the prediction feature data to the surgical duration prediction module.

[0018] One or more technical solutions provided in the present invention have at least the following technical effects or advantages: 1. The method for predicting and scheduling surgical duration based on machine learning provided by the present invention obtains the platform historical processing data of the same surgical type as the predicted surgical type and the resource parameters of the medical information platform, and comprehensively analyzes to obtain the emergency determination result of duration prediction, so as to dynamically evaluate the emergency degree of duration prediction, and adjust the data processing flow according to the emergency degree of duration prediction, realizing accurate and timely prediction of surgical duration, and effectively solving the problem in the prior art that the resource allocation cannot be flexibly adjusted according to the emergency degree of duration prediction, resulting in slow response of surgical duration prediction.

[0019] 2. The present invention can accurately judge the state and efficiency of the data transmission channel through comprehensive analysis based on historical processing difficulty indicators and resource consumption indicators, and evaluation of data transmission channel effect parameters, and then realizes the selection of a more efficient data transmission protocol to accelerate the data transmission speed in the case of emergency duration prediction, ensuring the timeliness and accuracy of surgical duration prediction data, and improving the response ability and service quality of the medical information platform in emergency situations.

[0020] 3. By determining the reference standard for data segmentation based on data transmission channel effect indicators, the data size in the data transmission process can be optimized, and then while maintaining data integrity, the delay and bandwidth occupation of data transmission are reduced, and the data processing efficiency and prediction accuracy of the surgical duration prediction model are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flowchart of the method for predicting and scheduling surgical duration based on machine learning provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The embodiment of the present application provides a method for predicting and scheduling surgical duration based on machine learning, which solves the problem in the prior art that when there is insufficient resources caused by data concurrency in the medical system, the resource allocation cannot be flexibly adjusted according to the emergency degree of duration prediction, resulting in slow response of surgical duration prediction, and achieves accurate and timely prediction of surgical duration.

[0023] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0024] Such as Figure 1As shown in the figure, it is a flowchart of a surgical duration prediction and scheduling method based on machine learning provided by an embodiment of the present application. The method includes the following steps: When the medical information platform receives a surgical duration prediction request, it obtains the historical processing data of the platform, analyzes and obtains the historical processing difficulty index; obtains the resource parameter of the medical information platform within the first preset period, analyzes and obtains the resource consumption index, and comprehensively analyzes based on the historical processing difficulty index and the resource consumption index to obtain a duration prediction emergency determination result, where the duration prediction emergency determination result includes duration prediction emergency and duration prediction non-emergency; the medical information platform automatically retrieves the prediction feature data, and analyzes according to the duration prediction emergency determination result to obtain a preliminary data segmentation determination result; obtains the data transmission channel effect parameter, analyzes and obtains the data transmission channel effect index, analyzes based on the preliminary data segmentation determination result and the data transmission channel effect index to obtain the final data segmentation judgment result, and performs corresponding data processing to obtain the feature data packet; selects the corresponding data transmission protocol based on the duration prediction emergency determination result, and thus transmits the feature data packet to the surgical duration prediction module; the surgical duration prediction module receives the feature data packet, analyzes using a machine learning algorithm to obtain the surgical duration prediction value, and performs an automatic scheduling prompt.

[0025] In this embodiment, by dynamically analyzing the device processing capacity and the resource consumption of the information platform, the data processing flow can be flexibly adjusted to ensure that emergency surgeries are given priority, and it is ensured that when the system resource allocation is insufficient, the system resources can be dynamically allocated according to the urgency of the surgery. The system efficiency is improved. According to the duration prediction urgency and the effect of the data transmission channel, the data transmission protocol is intelligently selected and data segmentation is performed, effectively reducing the data transmission delay and bandwidth occupancy. Through precise data processing and transmission strategies, as well as advanced machine learning algorithms, the accuracy and reliability of surgical duration prediction are improved.

[0026] Further, when the medical information platform receives a surgical duration prediction request, it obtains the platform's historical processing data and analyzes to obtain the historical processing difficulty index. The specific steps are as follows: When the medical information platform receives a surgical duration prediction request, it obtains the predicted surgical type, and thereby counts the historical data of the medical information platform for processing surgeries of the same type as the predicted surgical type, which is recorded as the platform's historical processing data. The platform's historical processing data includes the historical average data loading duration, historical average CPU usage rate, historical average I / O waiting time, historical average number of input / output operations per second, and historical average GPU video memory occupancy rate; it obtains the preset low-performance verification set in the medical database. The low-performance verification set includes the maximum historical data loading duration, the maximum historical I / O waiting time, and the maximum historical number of input / output operations per second; it analyzes to obtain the historical processing difficulty index based on the platform's historical processing data and the low-performance verification set; the historical processing difficulty index can be obtained by comparing the historical average data loading duration, historical average I / O waiting time, and historical average number of input / output operations per second with the maximum historical data loading duration, the maximum historical I / O waiting time, and the maximum historical number of input / output operations per second respectively, and combining with the historical average CPU usage rate and historical average GPU video memory occupancy rate, and then introducing influence weights to combine with them respectively.

[0027] In this embodiment, it should be noted that the historical average data loading duration of the computer refers to the historical average loading duration when the computer opens the surgical duration prediction module, and the CPU usage rate refers to the percentage of the CPU occupied when processing tasks. The historical average CPU usage rate shows the historical average CPU load of the computer. A high CPU usage rate indicates that the computer system is running a large number of computationally intensive tasks, or there are certain processes occupying too much CPU resources. The historical average I / O waiting time, historical average number of input / output operations per second, and historical average GPU video memory occupancy rate are all historical average data when the computer device opens the surgical duration prediction module.

[0028] The platform's historical processing data includes the historical average data loading duration, historical average CPU usage rate, historical average I / O waiting time, historical average number of input / output operations per second, and historical average GPU video memory occupancy rate, which can be obtained by collecting these performance indicators through monitoring tools (such as Nagios, Zabbix, or Windows Performance Monitor) installed on the server.

[0029] Through the comprehensive monitoring solution, multiple key performance indicators of computer devices can be obtained simultaneously, providing a comprehensive and accurate data basis for the calculation of historical processing difficulty indicators, which helps to more accurately evaluate the difficulty level of the device in analyzing historical processing data. Based on the comprehensive historical processing data and the low-performance verification set, the medical information platform can more intelligently analyze the device performance bottleneck, optimize resource allocation, and the accurate historical processing difficulty indicators contribute to improving the accuracy of surgical duration prediction.

[0030] By analyzing the historical processing difficulty indicators, the historical processing performance and difficulty level of the device can be understood, which helps the medical information platform to reasonably allocate computing resources when receiving prediction requests, ensuring the timeliness and accuracy of prediction tasks. By monitoring the historical processing data of the platform, potential performance problems can be discovered and solved in a timely manner, which helps to improve the overall stability of the medical information platform and reduce the failure or delay of prediction tasks caused by device failures or performance degradation. The historical processing difficulty indicators provide an important decision-making basis for the medical information platform. When the platform faces multiple prediction tasks, it can intelligently select the most suitable device for prediction according to the performance bottleneck and difficulty level of the device to optimize the overall performance and resource utilization rate.

[0031] Through the historical processing difficulty indicators, the performance bottlenecks of the platform in processing specific surgical types, such as data loading speed, CPU usage rate, I / O waiting time, etc., can be identified, helping the platform to optimize resource allocation and ensure the stable operation of the medical information platform.

[0032] By analyzing the historical processing data of the platform (historical average data loading duration of the computer, historical average CPU usage rate, historical average I / O waiting time, historical average number of input / output operations per second, and historical average GPU video memory occupancy rate), the mutual influence relationships between these parameters are considered. For example: when the I / O waiting time is long, the data loading duration will increase, and there is a mutual dependence relationship between the CPU usage rate and IOPS. A high value of one will lead to a high value of the other. The CPU usage rate and GPU video memory occupancy rate affect each other in data processing tasks. When the GPU video memory occupancy rate is high, the CPU needs to handle more data scheduling tasks, resulting in an increase in the CPU usage rate; conversely, when the CPU usage rate is high, it will affect the scheduling efficiency of GPU tasks. The I / O waiting time is closely related to IOPS. When IOPS is high, the I / O device becomes a bottleneck, resulting in an increase in the I / O waiting time. When the data loading duration is long, the CPU needs to wait for the data to be loaded, resulting in a decrease in the CPU usage rate.

[0033] The historical processing difficulty indicators are obtained through the following specific method: Wherein, H represents the historical processing difficulty index, T represents the historical average data loading duration of the computer, and T a represents the maximum value of the historical data loading duration, e represents the natural constant, V represents the historical average CPU usage rate, D represents the historical average I / O waiting time, and D a represents the maximum value of the historical I / O waiting time, and C a represents the maximum value of the historical number of input / output operations per second, C represents the historical average number of input / output operations per second, X represents the historical average GPU video memory occupancy rate, α1 represents the influence weight of the data loading duration, α2 represents the influence weight of the CPU usage rate, α3 represents the influence weight of the I / O waiting time, α4 represents the influence weight of the number of input / output operations, and α5 represents the influence weight of the GPU video memory occupancy rate.

[0034] The influence weights of the data loading duration, CPU usage rate, I / O waiting time, number of input / output operations, and GPU video memory occupancy rate can be obtained from the database. For example, the influence weight of the data loading duration can be obtained by acquiring the historical data loading duration stored in the database and the corresponding influence weight of the data loading duration, thereby constructing a data loading duration mapping set, in which there is a one-to-one or many-to-one correspondence relationship. By inputting the data loading duration data to be used into the data loading duration mapping set, the influence weight of the data loading duration can be obtained. The acquisition methods of other influence weights, such as the influence weight of the CPU usage rate, I / O waiting time, number of input / output operations, and GPU video memory occupancy rate, are the same as that of the influence weight of the data loading duration and can be obtained by matching in the corresponding mapping sets.

[0035] Furthermore, obtain the resource parameters of the medical information platform within the first preset time period and analyze to obtain the resource consumption index. The specific steps include: obtaining the resource parameters of the medical information platform within the first preset time period, where the resource parameters of the medical information platform include the number of surgical duration prediction requests, average number of data queries per second, average response duration, prediction failure rate, number of user accesses, and number of concurrent requests; obtaining the preset ideal set of platform resources in the medical database and analyzing it with the resource parameters of the medical information platform to obtain the resource consumption index; the preset ideal set of platform resources in the medical database includes the ideal value of the number of surgical duration prediction requests, ideal value of the average number of data queries per second, ideal value of the average response duration, critical value of the prediction failure rate, ideal value of the number of user accesses, and ideal value of the number of concurrent requests; the resource consumption index is used to characterize the consumption degree of the number of surgical duration prediction requests, average number of data queries per second, average response duration, prediction failure rate, number of user accesses, and number of concurrent requests of the medical information platform on the resources of the medical information platform within the first preset time period.

[0036] In this embodiment, the surgical duration prediction request volume can be obtained by counting the number of surgical duration prediction requests received in the background management system of the medical information platform within the first preset time period. The average number of data queries per second can be obtained by calculating the average value of the number of query requests processed by the monitoring platform database within a unit time. The average response duration can be obtained by calculating the average value of the response times of the platform for processing each request. The prediction failure rate can be obtained by calculating the proportion of the number of failed requests in the surgical duration prediction requests within the first preset time period to the total request volume. The user access volume can be obtained by counting the number of users accessing the platform within the first preset time period through the background management system. The concurrent request volume can be obtained by counting the number of requests processed simultaneously by the background management system within a unit time.

[0037] By obtaining and analyzing the resource parameters of the medical information platform within the first preset time period and comparing them with the preset ideal set of platform resources, it can be concluded that the resource consumption index can intuitively reflect the resource usage of the medical information platform at different time periods, helping managers accurately understand the resource consumption trend of the platform, so as to make more reasonable resource planning and management. Through in-depth analysis of the resource consumption index, it is possible to discover waste or inefficiency in resource usage, such as resource idleness caused by the prediction request volume being far lower than the ideal value, or resource tension caused by the concurrent request volume being too high.

[0038] Analyzing the resource consumption index can help evaluate the usage efficiency of platform resources, identify situations where resource usage is too high or too low, and thus make reasonable adjustments. By analyzing the resource consumption index, the platform can better perform load balancing and avoid system performance degradation caused by overuse of certain resources. By monitoring the resource consumption index, the platform can promptly detect resource shortages, avoid prediction failures caused by resource shortages, and improve the prediction success rate. By optimizing resource consumption, the platform can shorten the response time, reduce the user waiting time, and enhance the user experience.

[0039] It should be noted that the prediction failure rate refers to the proportion of the number of timeouts or errors caused by resource shortages to the total number of predictions, and the concurrent request volume refers to the number of access requests processed by the medical information platform at the same time.

[0040] The resource consumption index is obtained by introducing the corresponding influence weights for fusion processing of the comparison result of the average number of data queries per second with the ideal value of the average number of data queries per second, and the comparison results of the average response duration, prediction failure rate, user access volume, and concurrent request volume with the ideal value of the average response duration, critical value of the prediction failure rate, ideal value of the user access volume, and ideal value of the concurrent request volume, respectively.

[0041] The specific method for obtaining the resource consumption index is as follows: It should be noted that XH represents the resource consumption index, YC represents the volume of surgical duration prediction requests, YC1 represents the ideal value of the volume of surgical duration prediction requests, CX represents the average number of data queries per second, CX1 represents the ideal value of the average number of data queries per second, SC represents the average response duration, SC1 represents the ideal value of the average response duration, SA represents the prediction failure rate, SA1 represents the ideal value of the prediction failure rate, SM represents the number of user accesses, SM1 represents the ideal value of the number of user accesses, BF represents the number of concurrent requests, BF1 represents the ideal value of the number of concurrent requests, β1 represents the influence weight of the volume of surgical duration prediction requests, β2 represents the influence weight of the number of data queries, β3 represents the influence weight of the response duration, β4 represents the influence weight of the prediction failure rate, β5 represents the influence weight of the number of user accesses, and β6 represents the influence weight of the number of concurrent requests.

[0042] The influence weights of the volume of surgical duration prediction requests, the number of data queries, the response duration, the prediction failure rate, the number of user accesses, and the number of concurrent requests can be obtained from the database. For example, the influence weight of the volume of surgical duration prediction requests can be obtained by acquiring the historical volume of surgical duration prediction requests stored in the database and the corresponding influence weight of the volume of surgical duration prediction requests, thereby constructing a mapping set of the volume of surgical duration prediction requests. There is a one-to-one or many-to-one correspondence in this mapping set. By inputting the data of the volume of surgical duration prediction requests to be used into the mapping set of the volume of surgical duration prediction requests, the influence weight of the volume of surgical duration prediction requests can be obtained. The acquisition methods of other influence weights, such as the influence weight of the number of data queries, the influence weight of the response duration, the influence weight of the prediction failure rate, the influence weight of the number of user accesses, and the influence weight of the number of concurrent requests, are the same as that of the influence weight of the volume of surgical duration prediction requests and can all be obtained by matching in the corresponding mapping sets.

[0043] The resource consumption index is obtained by analyzing the resource parameters of the medical information platform (the volume of surgical duration prediction requests, the average number of data queries per second, the average response duration, the prediction failure rate, the number of user accesses, and the number of concurrent requests), taking into account the mutual influence relationships among these parameters. For example, an increase in the volume of surgical duration prediction requests will lead to an increase in the number of concurrent requests, thereby increasing the platform load. An increase in the number of data queries per second will lead to an increase in the database load, and then extend the average response duration. An overly long average response duration will cause some requests to time out, increasing the prediction failure rate. An increase in the number of user accesses will lead to an increase in the number of concurrent requests, thereby increasing the platform load. An increase in the number of concurrent requests will directly result in an increase in the consumption of resources such as CPU, memory, and network bandwidth.

[0044] By analyzing these parameters, the performance bottlenecks of the platform can be identified. For example, if the average response time is too long and the prediction failure rate is high, it indicates low database query efficiency; if the number of concurrent requests is too high and the resource consumption is large, it means insufficient system resources.

[0045] Furthermore, based on the comprehensive analysis of the historical processing difficulty index and the resource consumption index, the duration prediction emergency determination result is obtained. The specific steps include: obtaining the preset historical processing difficulty lower limit value and the resource consumption index lower limit value in the medical information platform; comparing the historical processing difficulty index with the historical processing difficulty lower limit value, and comparing the resource consumption index with the resource consumption index lower limit value to obtain the duration prediction emergency determination result; if the historical processing difficulty index is above the historical processing difficulty lower limit value and the resource consumption index is above the resource consumption index lower limit value, the duration prediction emergency determination result is that the duration prediction is urgent, otherwise it is that the duration prediction is not urgent; the duration prediction emergency determination result includes that the duration prediction is urgent and that the duration prediction is not urgent.

[0046] In this embodiment, by comprehensively considering the historical processing difficulty index and the resource consumption index, the urgency of the duration prediction can be evaluated more comprehensively, avoiding the one-sidedness that may be brought by a single index judgment, and being able to capture the dynamic relationship between the surgical complexity and the hospital resource utilization, so as to more accurately assign priorities to surgeries. By classifying surgeries into urgent and non-urgent categories, it is convenient to allocate limited medical resources more effectively. Urgent surgeries can quickly obtain the resources for platform surgical duration prediction, improving the treatment efficiency and quality, while non-urgent surgeries can be arranged when resources permit, avoiding resource waste, and detailed analysis can be carried out on the non-urgent duration prediction to make the surgical duration prediction more accurate. Through data analysis and comparison, the surgical arrangement process can be continuously monitored and improved, improving the overall medical service quality.

[0047] Furthermore, according to the analysis of the duration prediction emergency determination result, the preliminary data segmentation determination result is obtained. The specific steps include: obtaining the duration prediction emergency determination result. If the duration prediction emergency determination result is that the duration prediction is urgent, the preliminary data segmentation determination result is to initially perform segmentation; if the duration prediction emergency determination result is that the duration prediction is not urgent, the preliminary data segmentation determination result is to initially not perform segmentation.

[0048] In this embodiment, by deciding whether to perform data segmentation based on the urgency of the duration prediction, the method can quickly respond to the data needs of urgent duration predictions, avoid unnecessary processing steps, and thus improve the efficiency of overall data processing. For urgent duration prediction needs, timely data segmentation can ensure that key medical information is prioritized and stored, which helps the medical platform quickly obtain the required information and make accurate diagnosis and treatment decisions. For non-urgent duration predictions, avoiding unnecessary data segmentation can reduce data redundancy and storage costs, and optimize the management and utilization of medical data. Through the automated data segmentation judgment process, manual intervention is reduced and the accuracy and efficiency of data processing are improved.

[0049] Furthermore, data transmission channel effect parameters are obtained, and data transmission channel effect indicators are obtained by analysis. The specific steps include: obtaining data transmission channel effect parameters, which include the average network upload speed, average bandwidth, average CPU utilization, average channel occupancy and channel bit error rate in a second preset time period; obtaining an ideal channel effect data set preset in the medical information platform, and analyzing it with the data transmission channel effect parameters to obtain a data transmission channel effect indicator; the ideal channel effect data set includes an ideal network upload speed value, an ideal bandwidth value and a channel bit error rate limit value; the data transmission channel effect indicator is used to characterize the degree of influence of the average network upload speed, average bandwidth, average CPU utilization, average channel occupancy and channel bit error rate on the data transmission of the data transmission channel.

[0050] In this embodiment, by acquiring and analyzing the data transmission channel effect parameters (such as average network upload speed, average bandwidth, etc.), the status of the data transmission channel can be monitored in real time, and potential performance bottlenecks can be discovered and resolved in a timely manner, thereby improving data transmission efficiency and stability. By monitoring the channel bit error rate, errors in data transmission can be discovered in a timely manner, and measures can be taken to reduce the bit error rate to ensure the accuracy and integrity of medical data. Monitoring the average channel occupancy rate helps to optimize the priority and efficiency of emergency data transmission, ensuring that critical medical information can be transmitted quickly and accurately at critical moments. Based on data such as average CPU utilization, server resource allocation can be reasonably adjusted to avoid resource waste, improve resource utilization, and reduce medical resource waste and cost expenditures caused by poor or interrupted data transmission.

[0051] Feedback on data transmission channel performance indicators helps users of the medical information platform understand the data transmission status and improve users' trust and satisfaction with the platform. By optimizing the data transmission channel, the overall performance and user experience of the medical information platform can be improved, and users' dependence on and loyalty to the platform can be enhanced.

[0052] The data transmission channel effect index is obtained by analyzing the data transmission channel effect parameters (average network upload speed, average bandwidth, average CPU utilization rate, channel average occupancy rate, and channel bit error rate) within the second preset period, taking into account the mutual influence relationships among these parameters. For example, a faster upload speed means that data can reach the destination more quickly. The wider the bandwidth, the larger the amount of data that can be transmitted at the same time, thus supporting a higher upload speed. A channel with a large average bandwidth can accommodate more data for simultaneous transmission, and the full utilization of the bandwidth also depends on the processing power of the CPU. When a large amount of data needs to be uploaded quickly, the CPU needs to process this data efficiently, which is an important indicator for measuring whether the CPU is overloaded when handling data upload tasks. If the channel is frequently occupied, it will affect its response speed and reliability, and thus affect the performance of the overall data transmission channel. A too high bit error rate indicates that errors or losses occur during data transmission, which will increase the need for retransmitting data, thereby reducing the overall transmission efficiency. At the same time, a too high bit error rate will also have a negative impact on the CPU utilization rate.

[0053] Through the data transmission channel effect index, bottlenecks in data transmission (such as network speed, bandwidth, CPU utilization rate, etc.) can be identified, thereby optimizing data transmission efficiency and reducing data transmission time. By monitoring parameters such as the channel bit error rate, problems in data transmission can be detected in a timely manner, avoiding data loss or damage, and ensuring the stability of data transmission. By analyzing the channel effect index, the platform can reasonably allocate network resources, avoiding the performance degradation of other channels caused by excessive occupancy of some channels. In case of emergency, the data transmission channel effect index can help select the optimal transmission protocol (such as UDP or TCP) to ensure the timely transmission of data.

[0054] In summary, these parameters are interrelated and interact with each other, jointly determining the performance of the data transmission channel. A data channel with good performance should have a fast upload speed, sufficient bandwidth, reasonable CPU utilization rate, efficient channel response ability, and low bit error rate. By comprehensively analyzing these parameters, the quality of the data transmission channel and the amount of data it can carry can be evaluated, thus providing strong support for optimizing data transmission.

[0055] The data transmission channel effect index is used to introduce an influence weight after comparing the average network upload speed with the ideal value of the network upload speed, comparing the average bandwidth with the ideal value of the bandwidth, and comparing the channel bit error rate with the limit value of the channel bit error rate, and combining the average CPU utilization rate with the influence weight introduced and the channel average occupancy rate to obtain the data transmission channel effect index.

[0056] The specific method for obtaining the data transmission channel effect index is as follows: Wherein, GX represents the data transmission channel effect index, DS represents the average network upload speed, DS1 represents the ideal value of the network upload speed, UK represents the average bandwidth, UK1 represents the ideal value of the bandwidth, US represents the average CPU utilization rate, e represents the natural constant, OE represents the average channel occupancy rate, WM represents the channel bit error rate, WM1 represents the limit value of the channel bit error rate, μ1 represents the influence weight of the network upload speed, μ2 represents the influence weight of the bandwidth, μ3 represents the influence weight of the CPU utilization rate, μ4 represents the influence weight of the average channel occupancy rate, and μ5 represents the influence weight of the channel bit error rate.

[0057] The influence weights of the network upload speed, bandwidth, CPU utilization rate, average channel occupancy rate, and channel bit error rate can be obtained from the database. For example, the influence weight of the network upload speed can be obtained by acquiring the historical network upload speeds stored in the database and the corresponding influence weights of the network upload speeds. Thus, a network upload speed mapping set is constructed, in which there is a one-to-one or many-to-one correspondence. By inputting the network upload speed data to be used into the network upload speed mapping set, the influence weight of the network upload speed can be obtained. The acquisition methods of other influence weights, such as the influence weights of the bandwidth, CPU utilization rate, average channel occupancy rate, and channel bit error rate, are the same as that of the influence weight of the network upload speed and can all be obtained by matching in the corresponding mapping sets.

[0058] It should be noted that the comprehensive emergency evaluation index is obtained based on the analysis of the historical processing difficulty index and the resource consumption index. The specific method is as follows: ; wherein, PZ represents the comprehensive emergency evaluation index, H represents the historical processing difficulty index, XH represents the resource consumption index, γ1 represents the influence weight of the historical processing difficulty, and γ2 represents the influence weight of the resource consumption.

[0059] The influence weights of the historical processing difficulty and the resource consumption can be obtained from the database. For example, the influence weight of the historical processing difficulty can be obtained by acquiring the historical processing difficulty indexes within the historical preset period stored in the database and the corresponding influence weights of the historical processing difficulty indexes within the historical preset period. Thus, a historical processing difficulty mapping set is constructed, in which there is a one-to-one or many-to-one correspondence. By inputting the historical processing difficulty data to be used into the historical processing difficulty mapping set, the influence weight of the historical processing difficulty can be obtained. The acquisition method of the influence weight of the resource consumption is the same as that of the influence weight of the historical processing difficulty and can all be obtained by matching in the corresponding mapping sets.

[0060] The comprehensive emergency assessment indicators can help the platform quickly identify emergency situations (such as excessive resource consumption, great difficulties in historical processing, etc.), so as to take corresponding emergency measures. Through the comprehensive emergency assessment indicators, the platform can assign different priorities to different surgical duration prediction requests to ensure that emergency requests are processed in a timely manner. In case of an emergency, the platform can dynamically adjust resource allocation according to the comprehensive emergency assessment indicators to ensure that critical tasks receive sufficient resource support. Through the comprehensive emergency assessment indicators, the platform can optimize the system response strategy, reduce the response time in case of an emergency, and improve the overall efficiency.

[0061] Furthermore, based on the preliminary determination result of data segmentation and the analysis of the data transmission channel effect indicators, the final data segmentation judgment result is obtained, and corresponding data processing is performed to obtain the feature data packet. The specific steps include: obtaining the preset data transmission channel effect threshold in the medical information platform and comparing it with the data transmission channel effect indicators to obtain the data transmission channel quality evaluation result, where the data transmission channel quality evaluation result includes qualified data transmission channel quality and unqualified data transmission channel quality; obtaining the final data segmentation judgment result based on the preliminary determination result of data segmentation and the analysis of the data transmission channel effect indicators, where the final data segmentation judgment result includes data segmentation and data non-segmentation; if the preliminary determination result of data segmentation is preliminary non-execution of segmentation and the data transmission channel quality evaluation result is qualified data transmission channel quality, then the final data segmentation judgment result is data non-segmentation, otherwise segmentation; if the final data segmentation judgment result is data segmentation, then the data segmentation standard is analyzed, and data segmentation processing is performed based on the data segmentation standard to obtain the feature data packet, where the data segmentation standard includes the first data segmentation standard, the second data segmentation standard, and the third data segmentation standard; among them, if the preliminary determination result of data segmentation is preliminary execution of segmentation and the data transmission channel quality evaluation result is qualified data transmission channel quality, then the first data segmentation standard is analyzed, and data segmentation processing is performed based on the first data segmentation standard to obtain the feature data packet; if the preliminary determination result of data segmentation is preliminary execution of segmentation and the data transmission channel quality evaluation result is unqualified data transmission channel quality, then the second data segmentation standard is analyzed, and data segmentation processing is performed based on the second data segmentation standard to obtain the feature data packet; if the preliminary determination result of data segmentation is preliminary non-execution of segmentation and the data transmission channel quality evaluation result is unqualified data transmission channel quality, then the third data segmentation standard is analyzed, and data segmentation processing is performed based on the third data segmentation standard to obtain the feature data packet.

[0062] In this embodiment, by obtaining data transmission channel effect indicators (such as network upload speed, bandwidth, CPU utilization, etc.), the data segmentation strategy can be dynamically adjusted according to the current network and resource status. For example, when the network upload speed is low, the data can be segmented into smaller fragments to reduce the amount of data transmitted in a single transmission, thereby improving the transmission efficiency and ensuring the timely transmission of critical data, avoiding delays caused by network congestion.

[0063] The reference criteria for data segmentation include data size and data time window, and the segmentation strategy can be flexibly adjusted according to specific requirements. For example, for large-scale data, a finer-grained segmentation method can be adopted to reduce the complexity of single processing; while for structured data, optimized segmentation can be performed based on data types to improve processing efficiency.

[0064] It should be noted that if the first data segmentation criterion is analyzed and the feature data packet is obtained through data segmentation processing based on the first data segmentation criterion, the specific method is as follows: Obtain the first data segmentation criterion, which includes the data size and data time window for data segmentation, and perform data cutting on the predicted feature data based on the first data segmentation criterion. For example, if the first data segmentation criterion is: data size 100MB, data time window per 5-minute data block, then the original data is segmented according to the preset single transmission standard to ensure that the size of each data packet is close to but does not exceed the maximum carrying capacity of the channel (100MB), and the data time window does not exceed 5 minutes.

[0065] If the second data segmentation criterion is analyzed and the feature data packet is obtained through data segmentation processing based on the second data segmentation criterion, the specific method is as follows: Obtain the second data segmentation criterion, which includes the data size and data time window for data segmentation, and perform data cutting on the predicted feature data based on the second data segmentation criterion. For example, if the second data segmentation criterion is: data size 50MB, data time window per 2-minute data block, then the original data is segmented according to the preset single transmission standard to ensure that the size of each data packet is close to but does not exceed the maximum carrying capacity of the channel (50MB), and the data time window does not exceed 2 minutes.

[0066] If the third data segmentation criterion is analyzed and the feature data packet is obtained through data segmentation processing based on the third data segmentation criterion, the specific method is to obtain the third data segmentation criterion, which includes the data size and data time window for data segmentation, and perform data cutting on the predicted feature data based on the third data segmentation criterion. For example, if the second data segmentation criterion is: data size 70MB, data time window per 3-minute data block, then the original data is segmented according to the preset single transmission standard to ensure that the size of each data packet is close to but does not exceed the maximum carrying capacity of the channel (70MB), and the data time window does not exceed 3 minutes.

[0067] By analyzing resource parameters such as CPU utilization and bandwidth, the data segmentation strategy can be dynamically adjusted to avoid resource overload or idleness. For example, when the CPU utilization is high, the complexity of data segmentation can be reduced to lower the computational burden; while when the bandwidth is sufficient, the amount of data transmitted at one time can be increased to improve resource utilization. This dynamic adjustment mechanism helps to optimize system performance and ensure the efficient use of resources.

[0068] When the channel occupancy rate is high, the UDP protocol can be used for data transmission, sacrificing some reliability for higher transmission speed, which can ensure the timely transmission of prediction data and improve the emergency response ability. Through data segmentation, large-scale data can be decomposed into smaller segments, facilitating efficient processing by machine learning models. For example, in the surgical duration prediction model, data segmentation can reduce the amount of data processed at one time, lower the computational complexity of the model, and at the same time ensure the integrity and consistency of the data, which helps to improve the prediction accuracy and generalization ability of the model.

[0069] It should also be noted that to obtain the third criterion for data segmentation and perform corresponding data processing to obtain the feature data packet, the specific method is as follows: Obtain the preset data transmission channel effect index interval in the medical information platform and the single data transmission standard corresponding to each data transmission channel effect index interval. If the data transmission channel effect index is within a certain preset data transmission channel effect index interval, obtain the single data transmission standard corresponding to this interval as the third criterion for data segmentation.

[0070] Furthermore, analyze to obtain the first criterion for data segmentation, and based on the first criterion for data segmentation, perform data segmentation processing to obtain the feature data packet. The specific steps include: Analyze to obtain the comprehensive emergency evaluation index based on the historical processing difficulty index and the resource consumption index; Obtain the preset comprehensive emergency evaluation index intervals in the medical information platform and the single data transmission standards corresponding to each comprehensive emergency evaluation index interval, and match them with the comprehensive emergency evaluation index. If the comprehensive emergency evaluation index is within a certain preset comprehensive emergency evaluation index interval, obtain the single data transmission standard corresponding to this interval as the first criterion for data segmentation; Perform data segmentation processing based on the first reference criterion for data segmentation to obtain the feature data packet; The comprehensive emergency evaluation index is used to process the result of combining the historical processing difficulty index with the historical processing difficulty influence weight and the result of combining the resource consumption index with the resource consumption influence weight to obtain the comprehensive emergency evaluation index.

[0071] In this embodiment, it should be noted that based on the processing evaluation efficiency difference, perform comprehensive processing to obtain the processing evaluation efficiency deviation degree. The specific method is as follows: Divide the processing evaluation efficiency difference by the processing evaluation efficiency index threshold to obtain the processing evaluation efficiency deviation degree.

[0072] Based on the analysis of the comprehensive emergency assessment index and the data transmission channel effect index, the processing evaluation efficiency index is obtained. The specific method is as follows: ; where BX represents the processing evaluation efficiency index, GX represents the data transmission channel effect index, PZ represents the comprehensive emergency assessment index, B1 represents the weight of the processing evaluation efficiency impact, and B2 represents the weight of the comprehensive emergency assessment impact.

[0073] The weight of the processing evaluation efficiency impact and the weight of the comprehensive emergency assessment impact can be obtained from the database. For example, the weight of the processing evaluation efficiency impact can be obtained by acquiring the historical processing evaluation efficiency stored in the database and the corresponding weight of the processing evaluation efficiency impact. Thus, a processing evaluation efficiency mapping set is constructed, in which there is a one-to-one or many-to-one correspondence relationship. By inputting the processing evaluation efficiency data to be used into the processing evaluation efficiency mapping set, the weight of the processing evaluation efficiency impact can be obtained. The acquisition method of the weight of the comprehensive emergency assessment impact is the same as that of the weight of the processing evaluation efficiency impact.

[0074] Through data segmentation, the system can decompose large-scale data into smaller segments, reducing the amount of data transmitted at one time, which helps to optimize the utilization of network bandwidth, significantly improve the data transmission efficiency, and ensure that the surgical duration prediction module can obtain the required data in a timely manner. After data segmentation, each data segment contains metadata (such as shard identifier, data range, storage location, and hash function), which can flexibly manage and retrieve data segments. For example, when recombining data, relevant data segments can be quickly located and obtained through the shard identifier and storage location, improving the flexibility of data processing.

[0075] By storing data segments according to the set storage locations, storage resources can be utilized more efficiently. For example, data segments can be distributed across different storage locations according to the storage location, avoiding excessive storage pressure on a single device and improving the efficiency of data access at the same time.

[0076] After data segmentation, different data segments can be processed in parallel. For example, in the surgical duration prediction module, multiple data segments can be analyzed simultaneously, thus accelerating the data processing speed and improving the prediction speed. By using the hash function to verify data segments, the integrity and consistency of data during transmission and storage can be ensured. For example, when recombining data segments, the hash function can be used to verify whether the data has been tampered with or damaged, thus guaranteeing the reliability of the data.

[0077] Further, a second data segmentation criterion is analyzed, and data segmentation processing is performed based on the second data segmentation criterion to obtain feature data packets. The specific steps include: analyzing based on the comprehensive emergency assessment index and the data transmission channel effect index to obtain a processing evaluation efficiency index; obtaining a preset processing evaluation efficiency index threshold in the medical information platform, and performing a difference processing with the processing evaluation efficiency index to obtain a processing evaluation efficiency difference, and performing comprehensive processing based on the processing evaluation efficiency difference to obtain a processing evaluation efficiency deviation degree; obtaining each preset processing evaluation efficiency deviation degree interval and the single - time data transmission standard corresponding to the processing evaluation efficiency deviation degree interval in the medical information platform. If the processing evaluation efficiency deviation degree is within a certain preset processing evaluation efficiency deviation degree interval, then obtain the single - time data transmission standard corresponding to this interval as the second data segmentation criterion; performing data segmentation processing based on the analyzed second data segmentation criterion to obtain feature data packets; the processing evaluation efficiency index is used to characterize the comprehensive complexity of data processing by the comprehensive emergency assessment index and the data transmission channel effect index.

[0078] In this embodiment, when the duration prediction emergency determination result is "duration prediction is urgent", the UDP protocol is used to transmit data. The connectionless feature of the UDP protocol makes it have lower latency and higher transmission speed, and can quickly transmit data to the surgical duration prediction module, ensuring timely processing of key data and enhancing the emergency response ability. When the duration prediction emergency determination result is "duration prediction is not urgent", the TCP protocol is used to transmit data. The reliability and connection - oriented feature of the TCP protocol can ensure the integrity and order of data, avoiding data loss or errors.

[0079] Dynamically selecting the transmission protocol according to the duration prediction emergency determination result can optimize the utilization of resources. For example, in an emergency, the UDP protocol is preferentially used to reduce resource occupancy and ensure the rapid transmission of key data; in a non - emergency situation, the TCP protocol is used to make full use of system resources to ensure the reliability of data transmission. In an emergency, the UDP protocol can be used to transmit data to reduce the amount of transmitted data and further improve the transmission efficiency; in a non - emergency situation, the TCP protocol can be used to transmit prediction feature data to ensure the integrity and accuracy of data.

[0080] Further, based on the emergency determination result of duration prediction, the corresponding data transmission protocol is selected, and the feature data packet is transmitted to the surgical duration prediction module accordingly. The specific steps include: analyzing based on the emergency determination result of duration prediction to obtain the corresponding data transmission protocol, where the data transmission protocol includes the UDP protocol and the TCP protocol; if the emergency determination result of duration prediction is that the duration prediction is urgent, the predicted feature data is uploaded to the surgical duration prediction module using the UDP protocol, and if the emergency determination result of duration prediction is that the duration prediction is not urgent, the predicted feature data is uploaded to the surgical duration prediction module using the TCP protocol.

[0081] In this embodiment, after the predicted feature data is uploaded to the surgical duration prediction module, the surgical duration prediction module receives the feature data packet and uses a machine learning algorithm to analyze and obtain the surgical duration prediction value, and performs an automatic scheduling prompt. The day surgery arrangement can be freely set according to the surgical duration prediction value, ensuring the reasonable progress of the day surgery arrangement. By predicting the surgical duration, it provides a basis for the subsequent arrangement of the surgical duration and the scheduling of medical staff, avoiding the problem of inaccurate surgical duration prediction caused by the traditional method of predicting the surgical duration based on the personal experience of doctors. This scientific and refined surgical duration prediction facilitates a more reasonable arrangement of day surgeries, reduces the waste of operating room resources caused by inaccurate duration prediction, and also avoids the problem that medical staff cannot get effective rest due to inaccurate duration prediction.

[0082] In summary, in this embodiment, by obtaining the platform historical processing data of the same surgical type as the predicted surgical type and the resource parameters of the medical information platform, and comprehensively analyzing to obtain the emergency determination result of duration prediction, it is possible to dynamically evaluate the emergency level of duration prediction, and intelligently adjust the data processing process according to the emergency level of duration prediction, achieving accurate and timely prediction of the surgical duration, and effectively solving the problem in the prior art that it is impossible to flexibly adjust resource allocation according to the emergency level of duration prediction, resulting in a slow response in surgical duration prediction.

[0083] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0084] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A method for predicting and scheduling surgery duration based on machine learning, characterized in that: The following steps are involved: When the medical information platform receives a request for surgery duration prediction, it obtains the platform's historical processing data and analyzes it to obtain historical processing difficulty indicators; Obtaining resource parameters of the medical information platform within a first preset time period, analyzing to obtain resource consumption indicators, and obtaining duration prediction emergency determination results based on comprehensive analysis of historical processing difficulty indicators and resource consumption indicators, wherein the duration prediction emergency determination results include duration prediction emergency and duration prediction non-urgent; The medical information platform automatically retrieves the predicted feature data and analyzes the emergency judgment results based on the duration prediction to obtain the preliminary judgment results of data segmentation; Obtain data transmission channel effect parameters, analyze and obtain data transmission channel effect indicators, obtain final data segmentation judgment results based on the preliminary judgment results of data segmentation and the data transmission channel effect indicators, and perform corresponding data processing to obtain feature data packets; Selecting a corresponding data transmission protocol based on the duration prediction emergency determination result, thereby transmitting the feature data packet to the surgery duration prediction module; The surgery duration prediction module receives feature data packets, uses machine learning algorithms to analyze and obtain the predicted surgery duration value, and provides automatic scheduling prompts.

2. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 1, characterized in that: When the medical information platform receives the surgery duration prediction request, it obtains the platform's historical processing data and analyzes the historical processing difficulty index. The specific steps include: When the medical information platform receives a surgery duration prediction request, it obtains the predicted surgery type, and thereby collects historical data of the medical information platform processing the same type of surgery as the predicted surgery type, and records it as the platform's historical processing data, wherein the platform's historical processing data includes the historical average data loading time, the historical average CPU usage rate, the historical average I / O waiting time, the historical average number of input and output operations per second, and the historical average GPU memory occupancy rate; Obtaining a preset low-performance verification set in a medical database, wherein the low-performance verification set includes a historical maximum data loading duration, a historical maximum I / O waiting time, and a historical maximum number of input and output operations per second; The historical processing difficulty index is obtained based on the platform's historical processing data and low-performance validation set analysis; The historical processing difficulty index can be obtained by comparing the historical average data loading time, the historical average I / O waiting time, and the historical average input and output operations per second with the historical maximum data loading time, the historical maximum I / O waiting time, and the historical maximum input and output operations per second, and combining the historical average CPU usage and the historical average GPU memory occupancy, and then introducing influence weights to combine them.

3. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 1, characterized in that: The steps of obtaining the resource parameters of the medical information platform within the first preset time period and analyzing to obtain the resource consumption index include: Obtaining resource parameters of the medical information platform within a first preset time period, wherein the resource parameters of the medical information platform include the number of surgery duration prediction requests, the average number of data queries per second, the average response time, the prediction failure rate, the number of user accesses, and the number of concurrent requests; Obtain the ideal set of platform resources preset in the medical database and analyze it with the medical information platform resource parameters to obtain resource consumption indicators; The ideal set of platform resources preset in the medical database includes an ideal value of the number of surgery duration prediction requests, an ideal value of the average number of data queries per second, an ideal value of the average response time, a critical value of the prediction failure rate, an ideal value of the number of user accesses, and an ideal value of the number of concurrent requests; The resource consumption index is used to compare the average number of data queries per second with the ideal value of the average number of data queries per second, and to compare the average response time, the predicted failure rate, the number of user accesses and the number of concurrent requests with the ideal value of the average response time, the critical value of the predicted failure rate, the ideal value of the number of user accesses and the ideal value of the number of concurrent requests, respectively, and introduce the corresponding impact weights for fusion processing to obtain the resource consumption index.

4. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 1, characterized in that: The step of obtaining the emergency judgment result of duration prediction based on comprehensive analysis of historical processing difficulty indicators and resource consumption indicators includes: Obtain the historical processing difficulty lower limit and resource consumption indicator lower limit preset in the medical information platform; Compare the historical processing difficulty index with the historical processing difficulty lower limit, and compare the resource consumption index with the resource consumption index lower limit to obtain the duration prediction emergency judgment result; If the historical processing difficulty index is above the historical processing difficulty lower limit and the resource consumption index is above the resource consumption index lower limit, the duration forecast emergency judgment result is that the duration forecast is urgent, otherwise the duration forecast is not urgent; The duration prediction urgency determination result includes the duration prediction being urgent and the duration prediction being not urgent.

5. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 1, characterized in that: The analysis of the emergency judgment result based on the duration prediction obtains the preliminary judgment result of data segmentation, and the specific steps include: Obtain the duration prediction emergency judgment result. If the duration prediction emergency judgment result is that the duration prediction is urgent, the data segmentation preliminary judgment result is preliminary execution of segmentation. If the duration prediction emergency judgment result is that the duration prediction is not urgent, the data segmentation preliminary judgment result is preliminary non-execution of segmentation.

6. The method for predicting and scheduling surgery duration based on machine learning according to claim 1, wherein: The step of obtaining the data transmission channel effect parameter and analyzing to obtain the data transmission channel effect index specifically includes: Acquire data transmission channel effect parameters, wherein the data transmission channel effect parameters include an average network upload speed, an average bandwidth, an average CPU utilization rate, an average channel occupancy rate, and a channel bit error rate within a second preset time period; Obtain the ideal channel effect data set preset in the medical information platform, and analyze it with the data transmission channel effect parameters to obtain the data transmission channel effect index; The ideal channel effect data set includes an ideal network upload speed, an ideal bandwidth, and a channel bit error rate limit value; The data transmission channel effect index is used to introduce influence weights after comparing the average network upload speed with the ideal network upload speed, comparing the average bandwidth with the ideal bandwidth, and comparing the channel bit error rate with the channel bit error rate limit value, and combining the average CPU utilization and the average channel occupancy introduced with the influence weights to obtain the data transmission channel effect index.

7. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 1, characterized in that: The steps of obtaining the final data segmentation judgment result based on the preliminary judgment result of data segmentation and the data transmission channel effect index analysis, and performing corresponding data processing to obtain the characteristic data packet include: Obtaining a data transmission channel effect threshold preset in the medical information platform, and comparing it with the data transmission channel effect index to obtain a data transmission channel quality assessment result, wherein the data transmission channel quality assessment result includes a data transmission channel quality that is qualified and a data transmission channel quality that is unqualified; Based on the preliminary judgment result of data segmentation and the data transmission channel effect index analysis, a final data segmentation judgment result is obtained, wherein the final data segmentation judgment result includes data segmentation and data non-segmentation; If the preliminary judgment result of data segmentation is that the data segmentation is not to be performed and the data transmission channel quality assessment result is that the data transmission channel quality is qualified, then the final judgment result of data segmentation is that the data is not segmented, otherwise the data is segmented; If the final data segmentation judgment result is data segmentation, a data segmentation standard is obtained by analysis, and data segmentation processing is performed based on the data segmentation standard to obtain a feature data packet, wherein the data segmentation standard includes a first data segmentation standard, a second data segmentation standard, and a third data segmentation standard; If the preliminary determination result of data segmentation is that segmentation is initially performed and the data transmission channel quality assessment result is that the data transmission channel quality is qualified, a first data segmentation standard is obtained by analysis, and data segmentation processing is performed based on the first data segmentation standard to obtain a characteristic data packet; If the preliminary determination result of data segmentation is that segmentation is initially performed and the data transmission channel quality assessment result is that the data transmission channel quality is unqualified, a second data segmentation standard is obtained by analysis, and data segmentation processing is performed based on the second data segmentation standard to obtain a characteristic data packet; If the preliminary determination result of data segmentation is that segmentation is not performed and the data transmission channel quality assessment result is that the data transmission channel quality is unqualified, the third data segmentation standard is analyzed and data segmentation processing is performed based on the third data segmentation standard to obtain a characteristic data packet.

8. The method for predicting and scheduling surgery duration based on machine learning according to claim 7, wherein: The analysis obtains a first data segmentation standard, and performs data segmentation processing based on the first data segmentation standard to obtain a feature data packet, and the specific steps include: Based on the analysis of historical processing difficulty indicators and resource consumption indicators, a comprehensive emergency assessment indicator is obtained; Obtain each preset comprehensive emergency assessment indicator interval in the medical information platform and the single data transmission standard corresponding to each comprehensive emergency assessment indicator interval, and match them with the comprehensive emergency assessment indicator. If the comprehensive emergency assessment indicator is within a preset comprehensive emergency assessment indicator interval, obtain the single data transmission standard corresponding to the interval as the first data segmentation standard; Performing data segmentation processing based on a first reference standard for data segmentation to obtain a characteristic data packet; The comprehensive emergency assessment index is used to combine the historical processing difficulty index with the historical processing difficulty impact weight, and then process the result with the resource consumption index and the resource consumption impact weight to obtain the comprehensive emergency assessment index.

9. The method for predicting and scheduling surgery duration based on machine learning as claimed in claim 8, characterized in that: The analysis obtains the second data segmentation standard, and performs data segmentation processing based on the second data segmentation standard to obtain a characteristic data packet, and the specific steps include: Based on the analysis of comprehensive emergency evaluation indicators and data transmission channel effect indicators, the processing evaluation efficiency indicators are obtained; Obtaining a preset treatment evaluation efficiency index threshold in the medical information platform, performing difference processing with the treatment evaluation efficiency index to obtain a treatment evaluation efficiency difference, and performing comprehensive processing based on the treatment evaluation efficiency difference to obtain a treatment evaluation efficiency deviation; Obtain each processing evaluation efficiency deviation interval preset in the medical information platform and the single data transmission standard corresponding to the processing evaluation efficiency deviation interval; if the processing evaluation efficiency deviation is within a preset processing evaluation efficiency deviation interval, obtain the single data transmission standard corresponding to the interval as the second data segmentation standard; A second data segmentation standard is obtained based on the analysis, and data segmentation processing is performed based on the second data segmentation standard to obtain a characteristic data packet; The processing evaluation efficiency index is used to characterize the comprehensive complexity of data processing by the comprehensive emergency evaluation index and the data transmission channel effect index.

10. The method for predicting and scheduling surgery duration based on machine learning according to claim 1, wherein: The method of selecting a corresponding data transmission protocol based on the emergency determination result of the duration prediction, thereby transmitting the feature data packet to the surgery duration prediction module, specifically comprises the following steps: Analyze the emergency judgment result based on the duration prediction to obtain the corresponding data transmission protocol, which includes the UDP protocol and the TCP protocol; If the result of the duration prediction emergency judgment is that the duration prediction is urgent, the UDP protocol is used to upload the predicted feature data to the surgery duration prediction module. If the result of the duration prediction emergency judgment is that the duration prediction is not urgent, the TCP protocol is used to upload the predicted feature data to the surgery duration prediction module.

Images