A medical big data-based operating system resource dynamic scheduling method

Abstract

The application discloses a kind of operating system resource dynamic scheduling methods based on medical big data, comprising the following steps: collecting data respectively pre-processing generates historical medical operation data and real-time medical operation data;Identify medical data object, generate historical medical task association chain and real-time medical task association chain;Identify historical resource release event and incomplete task, generate historical medical task resource debt item;Based on historical incomplete task, generate historical resource debt evolution data;Identify current resource release event, generate current medical task resource debt item;Generate current resource debt migration chain, and generate current resource debt evolution data;Matching generates resource debt realization prediction data;Generate current available resource data, generate resource debt release scheme and adjust operating system resource.The application utilizes medical big data to predict resource debt and dynamically schedule, realizes accurate prediction, scheduling in time and resource utilization rate improvement.

Description

Technical Field

[0001] This invention relates to the field of operating system resource scheduling technology, and in particular to a method for dynamic scheduling of operating system resources based on medical big data. Background Technology

[0002] Existing medical information systems typically monitor image processing, monitoring and analysis, and diagnostic data calculation tasks by collecting medical business data, task execution data, and operating system resource data. They also combine the real-time occupancy of processor, memory, cache, storage, and network resources to allocate and dynamically adjust resources according to task priority, current load, or historical resource usage. However, existing technologies typically treat the decrease in resource usage after task pause, resource reduction, or cache clearing as a reduction in load. It is difficult to identify unfinished medical data processing content and its subsequent resource requirements, and it is also difficult to track the transfer process of unfinished tasks with medical data objects between subsequent medical tasks. As a result, the resource rebound caused when subsequent tasks resume execution cannot be predicted in advance, which can easily lead to resource allocation delays, task blockages, and a decrease in medical data processing efficiency. Therefore, how to provide a dynamic scheduling method for operating system resources based on medical big data is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0003] One objective of this invention is to propose a dynamic scheduling method for operating system resources based on medical big data. This invention utilizes medical big data to predict resource debt and dynamically schedule it, thereby achieving accurate prediction, timely scheduling, and improved resource utilization.

[0004] A method for dynamic scheduling of operating system resources based on medical big data according to an embodiment of the present invention includes the following steps: Collect historical and real-time raw medical operation data, and preprocess them to generate historical and real-time medical operation data respectively. Identify the medical data objects generated or invoked by each medical task, connect them to the operating system resource usage process corresponding to the same diagnosis and treatment event, and generate historical medical task association chains and real-time medical task association chains. Based on the historical medical task association chain, identify historical resource release events and unfinished tasks, and generate historical medical task resource debt items; Based on historical medical task resource debt items and historical medical task association chains, the process of historical unfinished tasks being taken over and resumed by subsequent medical tasks is tracked to generate historical resource debt evolution data. Based on the real-time medical task association chain, identify the current resource release event, determine and count the amount of unfinished tasks, calculate and generate changes in operating system resource usage, and generate the current medical task resource debt item. Based on the current medical task resource debt items and the real-time medical task association chain, a current resource debt migration chain is generated, and the current medical task resource debt items are connected to generate current resource debt evolution data; The current resource debt evolution data is matched with historical resource debt evolution data to generate resource debt realization prediction data; Based on real-time medical operation data, the system generates current available resource data, combines current available resource data with resource debt fulfillment prediction data to generate a resource debt release plan, and adjusts operating system resources.

[0005] Optionally, the preprocessing includes data cleaning, time alignment, and identifier matching; The historical medical operation raw data includes historical medical business data, historical task execution data, and historical operating system resource data; The raw data for real-time medical operations includes real-time medical business data, real-time task execution data, and real-time operating system resource data.

[0006] Optionally, the generation of the historical medical task association chain and the real-time medical task association chain includes: The system extracts patient identifiers, treatment event identifiers, medical task identifiers, and task start and end times from historical and real-time medical operation data. It identifies the medical data objects generated or invoked by each medical task and generates a consistent medical data object identifier for the same medical data object. Medical tasks are grouped according to patient identifiers and treatment event identifiers, and arranged according to task start time to generate historical treatment event task sequences and real-time treatment event task sequences. The source medical task and subsequent medical tasks that call the same medical data object are identified by the medical data object identifier. The two are connected to generate historical medical task transmission relationship and real-time medical task transmission relationship. The system acquires operating system resource usage records during the execution of each medical task, connects them according to the collection time, generates historical operating system resource usage processes and real-time operating system resource usage processes, connects the medical task transmission relationship and operating system resource usage processes to the corresponding diagnosis and treatment event task sequence, and generates historical medical task association chains and real-time medical task association chains.

[0007] Optionally, the generation of the historical medical task resource debt item includes: Read the task execution status and operating system resource usage process along the historical medical task association chain, identify medical task pauses and various resource reduction behaviors, and generate historical resource release events; Obtain the task processing stage and pending medical data objects corresponding to historical resource release events, determine the data processing content that has not yet been completed, and generate unfinished tasks and the amount of unfinished tasks. Obtain operating system resource usage records before and after historical resource release events, calculate the differences in usage of various resources, and generate historical resource usage change data; The unfinished task volume is correlated with historical resource usage change data to determine the amount of operating system resources reduced for each unfinished task, and a historical resource debt detail is generated. Connect historical resource release events, medical task identifiers, unfinished tasks, unfinished task quantity, medical data object identifiers, task processing stages, and historical resource debt details to generate historical medical task resource debt items.

[0008] Optionally, the generation of the historical resource debt evolution data includes: Obtain unfinished tasks, unfinished task quantities, medical data object identifiers, and historical resource debt details from historical medical task resource debt items, and determine the source tasks of resource debts in the historical medical task association chain; Search for subsequent medical tasks that call the same medical data object along the direction of medical data object transmission, match the medical data processed by them with the unfinished tasks, determine the amount of unfinished tasks to be undertaken and the time of undertaking, and generate historical unfinished task undertaking records. Connect the resource debt source task with the first subsequent medical task, allocate operating system resources according to the proportion of unfinished tasks undertaken to the total unfinished tasks, and generate the first historical resource debt migration record; Process the remaining historical unfinished task acceptance records in sequence, take the previous subsequent medical task as the source of resource debt for this task, allocate the previous operating system resource amount according to the proportion of the current unfinished task amount to the previous unfinished task amount, and generate the remaining historical resource debt migration records. By connecting historical resource and debt migration records according to the medical data object identifier and the time of acceptance, a historical resource and debt migration chain is generated. Obtain operating system resource usage records before and after undertaking subsequent medical tasks, calculate the increase in resource usage and determine the historical resource debt rebound amount, and generate historical resource debt rebound records and historical resource debt rebound trajectories; By connecting historical medical task resource debt items, historical resource debt migration chains, and historical resource debt rebound trajectories, historical resource debt evolution data is generated.

[0009] Optionally, the generation of the current medical task resource debt item includes: Read the task execution status and operating system resource usage process along the real-time medical task association chain, identify medical task pause and resource reduction behaviors, and generate the current resource release event; Determine the medical task corresponding to the current resource release event, obtain the task processing stage and the medical data objects to be processed, identify unfinished tasks and count the number of unfinished tasks; Obtain the operating system resource usage records before and after the current resource release event, calculate the differences in various resource usages, and generate the changes in operating system resource usage; The unfinished task volume is mapped to the difference in various resource usage, generating a resource debt mapping relationship between unfinished tasks and various operating system resources; Connect the current resource release event, medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, changes in operating system resource usage, and the corresponding relationship of resource debt to generate the current medical task resource debt item.

[0010] Optionally, the generation of the current resource debt evolution data includes: Obtain the medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, and resource debt correspondence from the current medical task resource debt items, and determine the resource debt source task in the real-time medical task association chain; Starting with the resource debt source task, search for subsequent medical tasks that call the same medical data object along the direction of medical data object transmission, match the actual medical data processed with the unprocessed medical data corresponding to the unfinished task, determine the subsequent medical task, the amount of unfinished task to be undertaken, and the undertaking time, generate and arrange the current unfinished task undertaking record, and obtain the current unfinished task undertaking data. Connect the resource debt source task with the first subsequent medical task, and map the operating system resource quantity in the resource debt correspondence to the first subsequent medical task according to the proportion of the unfinished task quantity undertaken to the total unfinished task quantity, thereby generating the first current resource debt migration record; Process the remaining unfinished task assignment records in sequence, take the previous subsequent medical task as the resource debt source task of this time, and map the operating system resource amount in the previous current resource debt migration record to the corresponding subsequent medical task according to the proportion of the current unfinished task amount to the previous unfinished task amount, and generate the remaining current resource debt migration records. Connect the current resource and debt migration records sequentially according to the medical data object identifier and the time of acceptance to generate the current resource and debt migration chain; Connect the current medical task resource debt item to the starting position of the current resource debt migration chain to generate the current resource debt evolution data.

[0011] Optionally, the generation of the resource debt fulfillment forecast data includes: Extract the resource release event type, occurrence time, amount of unfinished tasks, amount of operating system resources, subsequent medical tasks and takeover time from the current resource debt evolution data and historical resource debt evolution data, and arrange them according to the resource debt migration order to generate current resource debt matching data and historical resource debt matching data; Historical resource debt matching data is filtered according to the type of resource release event, and aligned with the current resource debt matching data at each migration position. The difference in unfinished task volume, the difference in operating system resource volume, and the difference in adjacent acceptance time interval are calculated to generate historical resource debt evolution matching results. Obtain historical resource debt rebound records from the matched historical resource debt rebound trajectory, accumulate the historical resource debt rebound amount, determine the arrival time when it reaches the corresponding operating system resource amount, subsequent medical tasks and migration location, and generate historical resource debt repayment reference data; Based on the ratio of unfinished tasks and operating system resources between the current resource debt matching data and the historical resource debt matching data, the historical resource debt repayment reference data is corrected to generate the expected resource debt repayment time and the scale of resource debt rebound. Subsequent medical tasks are determined according to the migration location to generate a set of resource debt repayment tasks. By connecting the expected maturity time of resource debt, the scale of resource debt rebound, and the set of resource debt maturity tasks, resource debt maturity forecast data is generated.

[0012] Optionally, the adjustment of operating system resources corresponding to the medical task includes: The system obtains the usage and total amount of various operating system resources from real-time medical operation data, calculates the available resources, and generates current available resource data. Obtain the operating system resource type and operating system resource quantity corresponding to each medical task from the resource debt repayment task set, and generate the resource data to be supplemented for each medical task; The order of resource debt release is determined according to the expected realization time of resource debt and the execution order of medical tasks. The release time and rate of resource debt are determined by combining the data on resources to be supplemented and the data on currently available resources, and a resource debt release plan is generated. According to the resource debt release plan, the operating system resources for the corresponding medical tasks will be increased sequentially, and the amount of operating system resources will be continuously adjusted according to the resource debt release rate.

[0013] The beneficial effects of this invention are: This invention cleanses, aligns, and identifies historical medical business data, historical task execution data, historical operating system resource data, and corresponding real-time data. This enables the association of medical business processes, medical task execution processes, and operating system resource usage processes under a unified time benchmark. By generating consistent medical data object identifiers for the same medical data object, it connects medical tasks, medical data object transmission relationships, and operating system resource usage processes within the same diagnosis and treatment event. This allows for continuous tracking of data sources, execution order, and resource usage among medical tasks, avoiding isolated scheduling based solely on the immediate resource usage status of a single medical task. This invention further identifies incomplete data processing content when medical tasks are paused, processor resources are reduced, memory resources are reduced, cache is cleared, or storage writes are delayed. It generates medical task resource debt items by combining the amount of incomplete tasks with the difference in resource usage before and after the event, thereby distinguishing the apparent decrease in resource usage from medical tasks that are still pending execution. This avoids misjudging temporarily released operating system resources as the disappearance of medical task resource requirements and improves the accuracy of identifying the actual resource requirements of medical tasks. This invention tracks the process of unfinished tasks being taken over and resumed by subsequent medical tasks, forming a resource debt migration chain and a resource debt rebound trajectory. This reflects the migration and recovery of corresponding operating system resources when unfinished tasks are continuously transferred between different medical tasks along with medical data objects. By matching current resource debt evolution data with historical resource debt evolution data, the expected time of resource debt repayment, the scale of resource debt rebound, and the set of resource debt repayment tasks can be predicted, enabling the identification of potential concentrated resource demands from subsequent medical tasks before the actual rebound. This invention further combines the current availability of processor resources, memory resources, cache resources, storage resources, and network resources to generate a resource debt release scheme that includes resource type, resource quantity, release order, release time, and release rate. It also continuously adjusts the corresponding operating system resources according to the order of medical task execution, thereby reducing resource contention, task blocking, and processing delays when subsequent medical tasks resume execution, improving the continuity of medical data processing, operating system resource utilization efficiency, and the timeliness and stability of dynamic scheduling of medical tasks. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a dynamic scheduling method for operating system resources based on medical big data proposed in this invention. Figure 2This is a schematic diagram of the medical task association chain in the operating system resource dynamic scheduling method based on medical big data proposed in this invention. Detailed Implementation

[0015] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0016] refer to Figures 1-2 A method for dynamic scheduling of operating system resources based on medical big data includes the following steps: Historical medical operation raw data and real-time medical operation raw data are collected, and preprocessed to generate historical medical operation data and real-time medical operation data respectively. The historical medical operation raw data includes historical medical business data, historical task execution data and historical operating system resource data. The real-time medical operation raw data includes real-time medical business data, real-time task execution data and real-time operating system resource data. Based on historical and real-time medical operation data, medical data objects generated or invoked by each medical task are identified, and a consistent medical data object identifier is generated for the same medical data object. Based on patient identifier, diagnosis and treatment event identifier, medical data object identifier, and medical task identifier, the medical tasks corresponding to the same diagnosis and treatment event, the medical data object transmission relationship, and the operating system resource usage process of each medical task are connected to generate historical medical task association chains and real-time medical task association chains. The operating system resources include processor resources, memory resources, cache resources, storage resources, and network resources. Based on the historical medical task association chain, identify historical resource release events, determine the corresponding unfinished tasks and count the number of unfinished tasks, calculate the difference in resource usage before and after the occurrence of the historical resource release event, generate historical resource usage change data, and generate historical medical task resource debt items by combining the number of unfinished tasks and the historical resource usage change data. The historical resource release events include medical task suspension, processor resource reduction, memory resource reduction, cache clearing and storage delayed writing. Based on historical medical task resource debt items and historical medical task association chains, the process of unfinished tasks corresponding to historical medical task resource debt items being taken over by subsequent medical tasks and the recovery execution process after being taken over are identified, generating historical resource debt evolution data, which includes historical medical task resource debt items, historical resource debt migration chains, and historical resource debt rebound trajectories. Based on the real-time medical task association chain, the current resource release event is identified, the unfinished tasks corresponding to the current resource release event are determined and the number of unfinished tasks is counted, the difference in resource usage before and after the current resource release event occurs is calculated, the change in operating system resource usage is generated, and the current medical task resource debt item is generated by combining the number of unfinished tasks and the change in operating system resource usage. The current resource release event includes medical task suspension, processor resource reduction, memory resource reduction, cache clearing and storage delayed writing. Based on the current medical task resource debt items and the real-time medical task association chain, the transfer process of the unfinished tasks corresponding to the current medical task resource debt items between subsequent medical tasks is tracked to generate the current resource debt migration chain. The current medical task resource debt items and the current resource debt migration chain are then connected to generate the current resource debt evolution data. The current resource debt evolution data is matched with historical resource debt evolution data to generate resource debt repayment prediction data, which includes the expected repayment time of resource debt, the scale of resource debt rebound, and the set of resource debt repayment tasks. Based on real-time medical operation data, current available resource data is generated. Combining the current available resource data and resource debt fulfillment prediction data, a resource debt release plan is generated. The operating system resources corresponding to the medical tasks in the resource debt fulfillment task set are adjusted according to the resource debt release plan. The resource debt release plan includes operating system resource type, operating system resource quantity, resource debt release order, resource debt release time, and resource debt release rate.

[0017] In this embodiment, the preprocessing includes data cleaning, time alignment, and identifier matching; The historical medical operation raw data includes historical medical business data, historical task execution data, and historical operating system resource data; The raw data for real-time medical operations includes real-time medical business data, real-time task execution data, and real-time operating system resource data.

[0018] In this embodiment, the generation of historical medical task association chains and real-time medical task association chains includes: The patient identifier, diagnosis and treatment event identifier, medical task identifier, task start time and task end time are obtained from historical medical operation data and real-time medical operation data, respectively. The medical data objects generated or called by each medical task are identified, and a consistent medical data object identifier is generated for the same medical data object. Specifically, the patient identifier, treatment event identifier, data type, acquisition device identifier, acquisition batch, and data sequence number corresponding to the medical data object are read and concatenated in the order of the fields to form an object feature string. The patient identifier P20260018, treatment event identifier E035, data type CT, acquisition device identifier CT03, acquisition batch B06, and data sequence number 126 are concatenated to form the object feature string "P20260018-E035-CT-CT03-B06-126". The object feature string is hashed to generate a medical data object identifier. When medical tasks generate or call medical data objects with the same field content, they obtain a consistent medical data object identifier. Medical tasks corresponding to the same medical event are grouped according to patient identifier and medical event identifier, and arranged according to task start time to generate historical medical event task sequence and real-time medical event task sequence; Based on the medical data object identifier, the medical task that generates the medical data object is identified as the source medical task, and the medical task whose start time is later than the source medical task and which calls the medical data object is identified as the subsequent medical task. The source medical task and the subsequent medical task are connected to generate historical medical task transmission relationship and real-time medical task transmission relationship. According to the medical task identifier, obtain the operating system resource usage records of each medical task from the task start time to the task end time, and connect them according to the collection time to generate historical operating system resource usage process and real-time operating system resource usage process; connect the historical medical task transmission relationship and historical operating system resource usage process to the historical diagnosis and treatment event task sequence to generate historical medical task association chain; connect the real-time medical task transmission relationship and real-time operating system resource usage process to the real-time diagnosis and treatment event task sequence to generate real-time medical task association chain.

[0019] In this embodiment, the generation of historical medical task resource debt items includes: Read the task execution status and operating system resource usage process of each medical task along the historical medical task association chain, identify the occurrence time and corresponding medical task of medical task pause, processor resource reduction, memory resource reduction, cache clearing and storage delayed write, and generate historical resource release events; Obtain the task processing stage and pending medical data objects corresponding to the medical tasks when historical resource release events occur, determine the data processing content that has not yet been completed according to the medical data object identifier, generate unfinished tasks, and count the amount of pending data corresponding to the unfinished tasks to generate the amount of unfinished tasks. Using the occurrence time of historical resource release events as the dividing point, resource usage records before and after the occurrence of the event are obtained from the operating system resource usage process of the corresponding medical task. The usage difference of processor resources, memory resources, cache resources, storage resources and network resources is calculated to generate historical resource usage change data. Based on the medical data object identifier and task processing stage corresponding to the unfinished tasks, the amount of unfinished tasks is matched with historical resource usage change data to determine the amount of operating system resources reduced for each unfinished task due to historical resource release events, and a historical resource debt detail is generated. Specifically, based on the task processing stage, the amount of images to be processed and the amount of memory resources occupied in the three consecutive acquisition cycles before the occurrence of the historical resource release event are obtained. The amount of images to be processed are 270, 285, and 300, respectively, and the corresponding memory resource occupancy is 2.70GB, 2.85GB, and 3.00GB, respectively. Dividing these values ​​by the data yields a unit image memory resource of 10MB. The amount of unfinished tasks corresponding to medical data object identifiers D01 and D02 are 300 and 200, respectively, resulting in memory resource requirements of 3.00GB and 2.00GB. The amount of memory resource reduction in the historical resource occupancy change data is 4.00GB. The 4.00GB is allocated as 2.40GB and 1.60GB according to the ratio of 3.00 and 2.00, generating memory resource debt details corresponding to D01 and D02. The same process is used to generate historical resource debt details corresponding to other operating system resource types. Based on the medical task identifier and the occurrence time of historical resource release events, connect historical resource release events, medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, task processing stage, and historical resource debt details to generate historical medical task resource debt items.

[0020] In this embodiment, the generation of historical resource debt evolution data includes: Obtain medical task identifiers, unfinished tasks, unfinished task quantities, medical data object identifiers, and historical resource debt details from historical medical task resource debt items, and determine the source task of resource debt in the historical medical task association chain according to the medical task identifier; Search for subsequent medical tasks that call the same medical data object along the transmission direction of medical data objects in the historical medical task association chain, match the medical data actually processed by the subsequent medical tasks with the unprocessed medical data corresponding to the unfinished tasks, determine the subsequent medical tasks, the amount of unfinished tasks to be accepted, and the acceptance time, generate historical unfinished task acceptance records, and arrange them according to the acceptance time. Specifically, the first to 300 medical images corresponding to the unfinished task are obtained according to the medical data object identifier D01, and the first to 180 medical images actually processed by the subsequent medical task T12 are obtained. The 180 medical images that overlap are identified as the amount of unfinished task to be undertaken. The acquisition time corresponding to the first medical image read by T12 is identified as the undertaking time. The subsequent medical task T12, the medical data object identifier D01, the 180 unfinished task to be undertaken and the undertaking time are connected to generate a historical record of unfinished task undertaking. Connect the resource debt source task with the subsequent medical task in the first historical unfinished task undertaking record, allocate the operating system resource amount in the historical resource debt details according to the proportion of the unfinished task undertaking amount to the total unfinished task amount, and generate the first historical resource debt migration record. Specifically, the 180 unfinished tasks in the first historical unfinished task acceptance record are divided by the 300 unfinished tasks to obtain the first acceptance ratio of 0.60. The 2.40GB memory resource in the historical resource debt details corresponding to D01 is multiplied by 0.60 to obtain 1.44GB memory resource. The resource debt source task, subsequent medical task T12, medical data object identifier D01, 180 unfinished tasks, acceptance time, and 1.44GB memory resource are connected to generate the first historical resource debt migration record. The remaining historical unfinished task acceptance records are processed sequentially according to the acceptance time. The subsequent medical tasks in the previous historical resource debt migration record are used as resource debt source tasks. The operating system resource amount in the previous historical resource debt migration record is allocated according to the proportion of the current unfinished task amount to the previous unfinished task amount, and the remaining historical resource debt migration records are generated. Specifically, based on the acceptance time, the next historical unfinished task acceptance record is obtained. Subsequent medical task T18 reads the medical data object D01 generated or updated by the previous subsequent medical task T12, and accepts 120 medical images from it. T12 is identified as the source task of this resource debt. Dividing 120 by 180 yields the acceptance ratio of 0.6667. Multiplying the 1.44GB memory resource amount in the previous historical resource debt migration record by 0.6667 yields 0.96GB of memory resource amount. Connect T12, T18, medical data object identifier D01, 120 unfinished tasks, the acceptance time, and 0.96GB of memory resources to generate the second historical resource debt migration record. Subsequent medical task T23 reads the medical data object D01 generated or updated by T18 and accepts 75 medical images from it. Divide 75 by 120 to get the acceptance ratio of 0.625, and multiply 0.96GB by 0.625 to get 0.60GB, generating the third historical resource debt migration record. By sequentially connecting historical resource and debt migration records according to the medical data object identifier and the time of acceptance, a historical resource and debt migration chain is generated. Specifically, historical resource debt migration records are collected according to the medical data object identifier D01 and arranged according to the time of acceptance. The first historical resource debt migration record records the resource debt source task T08, the subsequent medical task T12 and 1.44GB of memory resources. The second historical resource debt migration record records the resource debt source task T12, the subsequent medical task T18 and 0.96GB of memory resources. The third historical resource debt migration record records the resource debt source task T18, the subsequent medical task T23 and 0.60GB of memory resources. The same subsequent medical task and resource debt source task in adjacent records are connected in sequence to generate a historical resource debt migration chain composed of T08, T12, T18 and T23 in sequence. Obtain the operating system resource usage records before and after taking over unfinished tasks in subsequent medical tasks, calculate the usage increment of the same operating system resource type, determine the usage increment corresponding to the operating system resource amount in the historical resource debt migration record as the historical resource debt rebound amount, and connect the subsequent medical task, the time of acceptance, the operating system resource type, the historical resource debt rebound amount and the historical resource debt migration record to generate the historical resource debt rebound record; Specifically, the memory resource usage of the subsequent medical task T12 before it took over the unfinished task for three consecutive collection cycles is obtained as 6.10GB, 6.20GB, and 6.30GB. The memory resource usage before taking over is calculated as 6.20GB. Then, the memory resource usage of the subsequent medical task T12 for three consecutive collection cycles after taking over is obtained as 7.20GB, 7.64GB, and 7.40GB. Subtracting 6.20GB from these values ​​yields memory resource usage increments of 1.00GB, 1.44GB, and 1.20GB, respectively. 1.44GB is determined as the historical resource debt rebound amount corresponding to T12. The same process is used to obtain the historical resource debt rebound amounts of 0.96GB and 0.60GB corresponding to T18 and T23, respectively. These are then connected to the subsequent medical task, the time of taking over, the memory resource type, the historical resource debt rebound amount, and the historical resource debt migration record to generate a historical resource debt rebound record. The historical resource debt rebound records are then connected according to the migration order of T12, T18, and T23 in the historical resource debt migration chain to generate a historical resource debt rebound trajectory. By connecting historical medical task resource debt items, historical resource debt migration chains, and historical resource debt rebound trajectories, historical resource debt evolution data is generated.

[0021] In this embodiment, the generation of the current medical task resource debt item includes: Read the task execution status and operating system resource usage process of each medical task along the real-time medical task association chain, identify the occurrence time of medical task pause, processor resource reduction, memory resource reduction, cache clearing and storage delayed write, record the corresponding medical task identifier, and generate the current resource release event; Based on the current resource release event, determine the corresponding medical task, obtain the task processing stage and pending medical data objects of the corresponding medical task when the current resource release event occurs, determine the data processing content that has not yet been completed as incomplete tasks, and count the amount of pending medical data corresponding to incomplete tasks to generate the amount of incomplete tasks. Using the time of the current resource release event as the dividing point, the resource usage records before and after the event are obtained from the operating system resource usage process of the corresponding medical task. The usage differences of processor resources, memory resources, cache resources, storage resources and network resources are calculated respectively to generate the operating system resource usage change. Based on the task processing stage and the amount of medical data to be processed corresponding to the unfinished tasks, the amount of unfinished tasks is mapped to the differences in various resource usages in the operating system resource usage changes, generating a resource debt mapping relationship between each unfinished task and processor resources, memory resources, cache resources, storage resources and network resources; Specifically, based on the task processing stage, the number of images to be processed in the three consecutive acquisition cycles before the current resource release event is obtained as 324, 342, and 360, respectively, along with the corresponding memory resource usage of 3.24GB, 3.42GB, and 3.60GB. These are divided to obtain a unit image memory resource of 10MB. The number of unfinished tasks corresponding to medical data object identifiers D11 and D12 are 360 ​​and 240, respectively, resulting in memory resource requirements of 3.60GB and 2.40GB. The memory resource reduction in the operating system resource usage change is 5.50GB. Based on the ratio of 3.60 to 2.40, the 5.50GB is allocated as 3.30GB and 2.20GB, generating the memory resource debt correspondence for D11 and D12. The same process is used to generate the resource debt correspondence for the other operating system resource types. Based on the occurrence time of the current resource release event and the medical task identifier, connect the current resource release event, medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, changes in operating system resource usage, and the corresponding relationship of resource debt to generate the current medical task resource debt item.

[0022] In this embodiment, the generation of current resource debt evolution data includes: Obtain the medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, and resource debt correspondence from the current medical task resource debt items, and determine the resource debt source task in the real-time medical task association chain according to the medical task identifier; Starting with the resource debt source task, search for subsequent medical tasks that call the same medical data object along the medical data object transmission direction in the real-time medical task association chain. Match the medical data actually processed by the subsequent medical tasks with the unprocessed medical data corresponding to the unfinished tasks to determine the subsequent medical tasks, the amount of unfinished tasks to be undertaken, and the undertaking time. Generate the current unfinished task undertaking record and arrange the current unfinished task undertaking record according to the undertaking time to generate the current unfinished task undertaking data. Specifically, the first to 360 medical images corresponding to the current unfinished task are obtained according to the medical data object identifier D11, and the first to 180 medical images actually processed by the subsequent medical task T21 are obtained. The 180 medical images that overlap are identified as the unfinished task volume. The acquisition time corresponding to the first medical image read by T21 is identified as the acceptance time. The subsequent medical task T21, the medical data object identifier D11, the 180 unfinished task volumes and the acceptance time are connected to generate the first current unfinished task acceptance record. Connect the resource debt source task with the subsequent medical task in the first current unfinished task acceptance record, and map the operating system resource quantity in the resource debt correspondence to the corresponding subsequent medical task according to the proportion of the accepted unfinished task quantity to the total unfinished task quantity, thus generating the first current resource debt migration record; Specifically, the 180 unfinished tasks in the first current unfinished task acceptance record are divided by the 360 ​​unfinished tasks to obtain the first current acceptance ratio of 0.50. The 3.30GB memory resource in the resource debt correspondence relationship corresponding to D11 is multiplied by 0.50 to obtain 1.65GB memory resource. The resource debt source task, subsequent medical task T21, medical data object identifier D11, 180 unfinished tasks, acceptance time and 1.65GB memory resource are connected to generate the first current resource debt migration record. Process the remaining unfinished task acceptance records in sequence according to the acceptance time, take the subsequent medical tasks in the previous current resource debt migration record as the resource debt source tasks, and map the operating system resource amount in the previous current resource debt migration record to the corresponding subsequent medical tasks according to the proportion of the current unfinished task amount to the previous unfinished task amount, and generate the remaining current resource debt migration records. Specifically, based on the acceptance time, the next currently unfinished task acceptance record is obtained. Subsequent medical task T24 reads the medical data object D11 generated or updated by the previous subsequent medical task T21, and accepts 120 medical images from it. T21 is identified as the source task of this resource debt. Dividing 120 by 180 gives the acceptance ratio of 0.6667. Multiplying the 1.65GB memory resource in the previous current resource debt migration record by 0.6667 gives 1.10GB of memory resource. Connect T21, T24, medical data object identifier D11, 120 unfinished tasks, the acceptance time, and 1.10GB of memory resources to generate the second current resource debt migration record. Subsequent medical task T27 reads the medical data object D11 generated or updated by T24 and accepts 60 medical images from it. Divide 60 by 120 to get the acceptance ratio of 0.50, and multiply 1.10GB by 0.50 to get 0.55GB, generating the third current resource debt migration record. Connect the current resource and debt migration records sequentially according to the medical data object identifier and the time of acceptance to generate the current resource and debt migration chain; Specifically, current resource debt migration records are collected according to the medical data object identifier D11 and arranged according to the time of acceptance. The first current resource debt migration record records the resource debt source task T17, the subsequent medical task T21 and 1.65GB of memory resources. The second current resource debt migration record records the resource debt source task T21, the subsequent medical task T24 and 1.10GB of memory resources. The third current resource debt migration record records the resource debt source task T24, the subsequent medical task T27 and 0.55GB of memory resources. The same subsequent medical task and resource debt source task in adjacent records are connected in sequence to generate a current resource debt migration chain composed of T17, T21, T24 and T27 in sequence. Connect the current medical task resource debt item to the starting position of the current resource debt migration chain to generate the current resource debt evolution data.

[0023] In this embodiment, the generation of resource debt fulfillment forecast data includes: The resource release event type, the time of occurrence of the resource release event, the amount of unfinished tasks, the amount of operating system resources, the subsequent medical tasks and the time of acceptance are obtained from the current resource debt evolution data and the historical resource debt evolution data, respectively. These are then arranged in the order of resource debt migration to generate current resource debt matching data and historical resource debt matching data. Historical resource debt matching data is filtered according to the type of resource release event. The filtered results are aligned with the current resource debt matching data at each migration position. The difference in the amount of unfinished tasks, the difference in the amount of operating system resources, and the difference in the time interval between adjacent unfinished tasks at the corresponding migration positions are calculated in turn to generate historical resource debt evolution matching results. Specifically, the amount of unfinished tasks accepted in each resource debt migration record is taken as the amount of unfinished tasks at the corresponding migration position. The current resource debt matching data is aligned with the historical resource debt matching data corresponding to the same resource release event type at each migration position. The current amount of unfinished tasks and the historical amount of unfinished tasks at the first migration position are both 180, resulting in a difference value of 0 for the amount of unfinished tasks. The current memory resource amount is 1.65GB, and the historical memory resource amount is 1.44GB. The difference of 0.21GB is divided by the sum of the two, 3.09GB, to obtain... The difference in operating system resource quantity is 0.0680. The adjacent acceptance time intervals are 12 and 11 collection cycles, respectively. Dividing the difference 1 by the sum of the two, 23, yields an acceptance time interval difference of 0.0435. Averaging the three difference values, the first migration position difference value is 0.0372. Averaging all migration position difference values, the two sets of historical resource debt matching data yield 0.044 and 0.126, respectively. The historical resource debt evolution data corresponding to 0.044 is used to generate the historical resource debt evolution matching result. Based on the historical resource debt evolution matching results, historical resource debt rebound records are obtained from the corresponding historical resource debt rebound trajectories. The historical resource debt rebound amount is accumulated according to the migration order. The arrival time when the accumulated historical resource debt rebound amount reaches the corresponding operating system resource amount, subsequent medical tasks and migration positions are determined. The time interval from the occurrence time of the historical resource release event to the arrival time is calculated to generate historical resource debt realization reference data. Specifically, based on memory resource type, the operating system resource amounts corresponding to the three migration positions in the historical resource debt matching data are obtained as 1.44GB, 0.96GB, and 0.60GB, and added together to obtain the corresponding operating system resource amount of 3.00GB. The historical resource debt rebound amounts corresponding to the three migration positions are obtained as 1.44GB, 0.96GB, and 0.60GB from the historical resource debt rebound trajectory. These amounts are accumulated sequentially according to the migration order to obtain 1.44GB, 2.40GB, and 3.00GB. The third migration position corresponding to the first time the accumulated historical resource debt rebound amount reaches 3.00GB is determined as the arrival position. The arrival time and subsequent medical tasks corresponding to this migration position are obtained. The 23 collection cycles between the occurrence time of the historical resource release event and the arrival time are calculated. The 23 collection cycles, subsequent medical tasks, and migration positions are connected to generate historical resource debt realization reference data. Based on the ratio of unfinished tasks and operating system resources between the current resource debt matching data and the corresponding historical resource debt matching data, the time interval and historical resource debt rebound amount in the historical resource debt repayment reference data are corrected. The corrected time interval is then superimposed onto the current resource release event time to generate the expected resource debt repayment time and the resource debt rebound scale. Subsequent medical tasks are determined in the current resource debt migration chain according to the migration location. The subsequent medical tasks, operating system resource types, and operating system resource quantities are then connected to generate a set of resource debt repayment tasks. Specifically, the memory resources corresponding to the three migration locations in the current resource debt matching data (1.65GB, 1.10GB, and 0.55GB) are added together to obtain 3.30GB. The memory resources corresponding to the historical resource debt matching data (1.44GB, 0.96GB, and 0.60GB) are added together to obtain 3.00GB. The current unfinished task volume (360) is divided by the historical unfinished task volume (300) to obtain a task volume ratio of 1.20. The operating system resource volume ratio (1.10) is obtained by dividing 3.30GB by 3.00GB. The average of the two ratios is used to obtain a time correction factor of 1.15. The 23 collection cycles are multiplied by 1.15 to obtain 26.45 collection cycles. The corresponding duration is superimposed onto the current resource release event time to generate the expected resource debt repayment time. The 3.00GB historical resource debt rebound amount is multiplied by 1.10 to generate a 3.30GB memory resource rebound scale. According to the three migration positions in the historical resource debt repayment reference data, the subsequent medical tasks T21, T24 and T27 are determined in the current resource debt migration chain. 1.44GB, 0.96GB and 0.60GB are multiplied by 1.10 respectively to obtain 1.584GB, 1.056GB and 0.660GB. The corresponding subsequent medical tasks, memory resource types and memory resource amounts are connected to generate a resource debt repayment task set. By connecting the expected maturity time of resource debt, the scale of resource debt rebound, and the set of resource debt maturity tasks, resource debt maturity forecast data is generated.

[0024] In this embodiment, adjusting the operating system resources corresponding to the medical task includes: The system obtains the usage and total amount of processor resources, memory resources, cache resources, storage resources, and network resources from real-time medical operation data, calculates the availability of various operating system resources according to the collection time, and generates current available resource data. Obtain the operating system resource type and operating system resource quantity corresponding to each medical task from the resource debt repayment task set, and generate the resource data to be supplemented for each medical task; The resource debt release order is determined according to the expected realization time of resource debt and the execution sequence of each medical task. The resource debt release time and resource debt release rate are determined by combining the resource data to be supplemented for each medical task and the currently available resource data. The resource debt release plan is generated by connecting the resource data to be supplemented for each medical task, the resource debt release order, the resource debt release time and the resource debt release rate. Specifically, the duration from the current moment to the expected fulfillment time of resource debt is divided into 21 resource release cycles. The resource debt release order is determined according to the execution sequence of T21, T24, and T27. The resource release cycles 1 to 7 are designated as the resource debt release time corresponding to T21, the resource release cycles 8 to 14 are designated as the resource debt release time corresponding to T24, and the resource release cycles 15 to 21 are designated as the resource debt release time corresponding to T27. 1.584GB, 1.056GB, and 0.660GB are divided by 7 to obtain the memory resource amounts of 0.2263GB, 0.1509GB, and 0.0943GB corresponding to each resource release cycle. The memory resource amounts are all less than the currently available memory resource amounts within the corresponding resource release cycle. The memory resource amounts corresponding to each cycle are determined as the resource debt release rate. The resource data to be supplemented for each medical task, the resource debt release order, the resource debt release time, and the resource debt release rate are connected to generate a resource debt release plan. According to the resource debt release plan, the operating system resources of each medical task in the resource debt repayment task set will be increased sequentially at the corresponding resource debt release time, and the amount of operating system resources will be continuously adjusted according to the resource debt release rate.

[0025] Example 1: To verify the feasibility of this invention in practice, it was applied to a resource scheduling scenario of the medical business operating system in the information center of a large general hospital. This information center simultaneously runs tasks such as medical image reconstruction, image analysis, vital sign monitoring, laboratory data processing, electronic medical record writing, and medical data archiving. Different medical tasks share processor resources, memory resources, cache resources, storage resources, and network resources, and medical data generated from the same diagnostic event is continuously accessed by multiple medical tasks.

[0026] The original operating system primarily scheduled tasks based on the amount of resources currently being used by medical tasks. When a medical task was paused, resources were reduced, cache was cleared, or storage writes were delayed, the system typically interpreted the decrease in resource usage as a reduction in task load and allocated the freed resources to other medical tasks. However, some medical tasks, even when paused, still had unprocessed medical data. This data would be processed by subsequent medical tasks, causing the previously reduced resource demand to reappear, resulting in a sudden surge in memory usage, processor queuing, and storage write blockage.

[0027] During implementation, historical and real-time raw medical operation data were continuously collected for 30 days. The collected data included patient identifiers, treatment event identifiers, medical task identifiers, task start and end times, task execution status, task processing stage, medical data objects, and various operating system resource usage records. Resource usage data was collected at 5-second intervals, resulting in approximately 1,860,000 medical task execution records, 420,000 medical data object records, and over 20,000,000 operating system resource usage records. The collected data was then cleaned, time-aligned, and identified uniformly to generate historical and real-time medical operation data.

[0028] Identify the medical data objects generated or invoked by each medical task, and generate medical data object identifiers by combining patient identifiers, treatment event identifiers, data types, device identifiers, data batches, and data sequences. Group medical tasks within the same treatment event according to patient identifiers and treatment event identifiers, and arrange them according to task start time to generate a treatment event task sequence.

[0029] The medical task originating from the medical data object is identified based on its identifier, along with subsequent medical tasks that call the same medical data object, thus generating a medical task transmission relationship. Simultaneously, operating system resource usage records during the execution of each medical task are obtained, forming an operating system resource usage process. The medical task transmission relationship and the operating system resource usage process are then connected to the diagnosis and treatment event task sequence, generating historical and real-time medical task association chains.

[0030] In the historical medical task chain, medical task T08 performed medical image reconstruction processing. When this task was paused due to the initiation of a high-priority medical task, 300 tomographic images remained unprocessed. Before the pause, memory usage was 5.80GB, which decreased to 3.40GB after the pause, a reduction of 2.40GB; processor usage decreased from 14 compute cores to 6 compute cores; and cache usage decreased from 1.20GB to 0.30GB.

[0031] The pause behavior of medical task T08 is identified as a historical resource release event. The 300 unprocessed tomographic images are identified as incomplete tasks. The reduced memory resources of 2.40GB are matched with incomplete tasks to generate historical medical task resource debt items.

[0032] Tracking along the direction of medical data object transfer, medical task T12 received 180 incomplete images, corresponding to a migration memory resource of 1.44GB; medical task T18 received another 120 images, corresponding to a migration memory resource of 0.96GB; and medical task T23 received another 75 images, corresponding to a migration memory resource of 0.60GB. Connecting these resource debt migration records generates a historical resource debt migration chain.

[0033] The memory usage increases for medical tasks T12, T18, and T23 after taking over unfinished tasks were 1.44GB, 0.96GB, and 0.60GB, respectively, consistent with the corresponding resource debt migration amounts. This indicates that the reduced resource demand when medical task T08 was paused did not disappear but migrated to subsequent medical tasks along with the unfinished task. Connecting the resource debt rebound records generates historical resource debt rebound trajectories and historical resource debt evolution data.

[0034] In the real-time medical task chain, medical task T17 reduced its resources due to the initiation of a high-priority task, at which time 360 ​​medical images remained unprocessed. Its memory usage decreased from 6.40GB to 3.10GB, a reduction of 3.30GB, thereby generating the current medical task resource debt item.

[0035] Continuing to track subsequent medical tasks, Medical Task T21 handles 180 incomplete images, corresponding to 1.65GB of memory resources; Medical Task T24 handles 120 images, corresponding to 1.10GB of memory resources; and Medical Task T27 handles 60 images, corresponding to 0.55GB of memory resources. Connecting the migration records of each current resource debt generates the current resource debt migration chain and current resource debt evolution data.

[0036] Matching the current resource debt evolution data with historical resource debt evolution data, comparing the amount of unfinished tasks, operating system resource consumption, and the time interval between adjacent task acceptances, yielded a matching difference value of 0.044. The corresponding historical resource debt experienced 23 collection cycles from release to completion, with a cumulative memory resource rebound of 3.00GB. After adjustments based on the proportion of unfinished tasks, resource consumption, and acceptance time intervals, it is predicted that the current resource debt will be fulfilled in approximately 26 collection cycles, with an estimated memory resource rebound of 3.30GB. Medical tasks T21, T24, and T27 have been identified as resource debt fulfillment tasks.

[0037] At this point, the operating system has 4.20GB of available memory, 18 available processor cores, and 1.80GB of available cache. To avoid impacting other medical tasks by adding 3.30GB of memory at once, a resource debt release plan is generated according to the execution order of the medical tasks.

[0038] The 1.65GB memory resources for medical task T21, 1.10GB memory resources for medical task T24, and 0.55GB memory resources for medical task T27 were supplemented over 7 acquisition cycles. Simultaneously, processor resources, cache resources, storage resources, and network resources were adjusted. After the adjustments were completed, none of the three subsequent medical tasks experienced a sudden increase in memory usage, a surge in the task queue, or storage write blocking.

[0039] Statistical analysis was conducted on actual operational data for 30 days before and after implementation. Before implementation, there were 74 instances of sudden increases in resource usage, 31 of which caused timeouts in medical task processing. The average waiting time for medical imaging tasks was 18.6 seconds, peak memory utilization was 93.2%, and there were 42 instances of storage delay write backlog. After adopting this invention, the number of sudden increases in resource usage decreased to 19, the number of medical task processing timeouts decreased to 8, the average waiting time decreased to 6.4 seconds, peak memory utilization decreased to 81.5%, and the number of storage delay write backlogs decreased to 11. The average effective utilization rate of operating system resources increased from 61.8% to 72.4%.

[0040] As demonstrated by the aforementioned operational data, this invention can identify unfinished tasks that persist even after medical tasks have been paused or resources have been reduced. It tracks the migration of corresponding operating system resources between subsequent medical tasks and predicts in advance the resource availability time, the scale of resource rebound, and the corresponding medical tasks. By gradually replenishing operating system resources with currently available resources, it can reduce medical task congestion caused by concentrated resource rebounds, thereby improving medical data processing efficiency and operating system stability.

[0041] Table 1. Comparison of the overall performance of resource scheduling methods in different medical operating systems

[0042] As shown in Table 1, the fixed resource quota method cannot adjust resources in a timely manner according to changes in the running status of medical tasks. There were 86 instances of sudden increases in resource usage, 39 instances of medical task timeouts, and the average effective utilization rate of operating system resources was only 58.7%. The task priority scheduling method can prioritize important medical tasks, but it mainly reallocates resources based on task priority and cannot identify data processing content that is still incomplete after tasks are paused. Therefore, there were still 74 instances of sudden increases in resource usage.

[0043] The real-time load threshold scheduling method can adjust resources based on current processor utilization, memory utilization, and storage load, reducing the number of resource usage spikes to 43 and the average waiting time for medical imaging tasks to 11.2 seconds. However, this method can only adjust resources after they reach the corresponding threshold, making it difficult to determine whether temporarily released resources will migrate to subsequent medical tasks along with unfinished tasks. Therefore, there are still 18 medical task processing timeouts and 24 storage delay write backlogs.

[0044] The method of this invention reduced the number of resource consumption spikes to 19, a 55.8% decrease compared to the real-time load threshold scheduling method; the number of medical task processing timeouts decreased from 18 to 8, a 55.6% reduction; and the average waiting time for medical imaging tasks decreased from 11.2 seconds to 6.4 seconds, a 42.9% reduction. This is because the invention can identify unfinished tasks generated when medical tasks are paused or resources are reduced, and generate medical task resource debt for the corresponding resource requirements, thereby avoiding the misinterpretation of temporarily reduced resource consumption as the disappearance of resource requirements.

[0045] The peak memory utilization rate of the method in this invention is 81.5%, which is 6.4 percentage points lower than that of the real-time load threshold scheduling method; the number of storage latency write backlogs decreased from 24 to 11, a reduction of 54.2%. This is because the invention tracks the migration process of resource debt between subsequent medical tasks along the direction of medical data object transmission, and predicts the resource debt fulfillment time, the scale of resource rebound, and the corresponding medical tasks. This allows processor resources, memory resources, cache resources, storage resources, and network resources to be replenished in stages in advance, reducing concentrated resource rebounds.

[0046] The method of this invention achieves an average effective utilization rate of 72.4% for operating system resources, which is 4.9 percentage points higher than the real-time load threshold scheduling method. The average resource scheduling response time is reduced from 10.4 seconds to 6.8 seconds, a reduction of 34.6%. This indicates that the present invention does not reduce resource peaks by retaining idle resources for a long time, but rather generates a resource debt release plan based on the current available resource data, so that the resource replenishment time and resource replenishment rate match the actual needs of subsequent medical tasks.

[0047] In summary, this invention transforms the traditional passive adjustment after resource shortage into dynamic scheduling before resource demand arises by constructing a medical task association chain, identifying medical task resource debts, tracking resource debt migration, and predicting the resource debt fulfillment process. This reduces medical task timeouts, sudden increases in resource consumption, and storage write backlogs, while improving operating system resource utilization and the stability of medical data processing.

[0048] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for dynamic scheduling of operating system resources based on medical big data, characterized in that, Includes the following steps: Collect historical and real-time raw medical operation data, and preprocess them to generate historical and real-time medical operation data respectively. Identify the medical data objects generated or invoked by each medical task, connect them to the operating system resource usage process corresponding to the same diagnosis and treatment event, and generate historical medical task association chains and real-time medical task association chains. Based on the historical medical task association chain, identify historical resource release events and unfinished tasks, and generate historical medical task resource debt items; Based on historical medical task resource debt items and historical medical task association chains, the process of historical unfinished tasks being taken over and resumed by subsequent medical tasks is tracked to generate historical resource debt evolution data. Based on the real-time medical task association chain, identify the current resource release event, determine and count the amount of unfinished tasks, calculate and generate changes in operating system resource usage, and generate the current medical task resource debt item. Based on the current medical task resource debt items and the real-time medical task association chain, a current resource debt migration chain is generated, and the current medical task resource debt items are connected to generate current resource debt evolution data; The current resource debt evolution data is matched with historical resource debt evolution data to generate resource debt realization prediction data; Based on real-time medical operation data, the system generates current available resource data, combines current available resource data with resource debt fulfillment prediction data to generate a resource debt release plan, and adjusts operating system resources.

2. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The preprocessing includes data cleaning, time alignment, and identifier matching; The historical medical operation raw data includes historical medical business data, historical task execution data, and historical operating system resource data; The raw data for real-time medical operations includes real-time medical business data, real-time task execution data, and real-time operating system resource data.

3. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the historical medical task association chain and the real-time medical task association chain includes: The system extracts patient identifiers, treatment event identifiers, medical task identifiers, and task start and end times from historical and real-time medical operation data. It identifies the medical data objects generated or invoked by each medical task and generates a consistent medical data object identifier for the same medical data object. Medical tasks are grouped according to patient identifiers and treatment event identifiers, and arranged according to task start time to generate historical treatment event task sequences and real-time treatment event task sequences. The source medical task and subsequent medical tasks that call the same medical data object are identified by the medical data object identifier. The two are connected to generate historical medical task transmission relationship and real-time medical task transmission relationship. The system acquires operating system resource usage records during the execution of each medical task, connects them according to the collection time, generates historical operating system resource usage processes and real-time operating system resource usage processes, connects the medical task transmission relationship and operating system resource usage processes to the corresponding diagnosis and treatment event task sequence, and generates historical medical task association chains and real-time medical task association chains.

4. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the historical medical task resource debt includes: Read the task execution status and operating system resource usage process along the historical medical task association chain, identify medical task pauses and various resource reduction behaviors, and generate historical resource release events; Obtain the task processing stage and pending medical data objects corresponding to historical resource release events, determine the data processing content that has not yet been completed, and generate unfinished tasks and the amount of unfinished tasks. Obtain operating system resource usage records before and after historical resource release events, calculate the differences in usage of various resources, and generate historical resource usage change data; The unfinished task volume is correlated with historical resource usage change data to determine the amount of operating system resources reduced for each unfinished task, and a historical resource debt detail is generated. Connect historical resource release events, medical task identifiers, unfinished tasks, unfinished task quantity, medical data object identifiers, task processing stages, and historical resource debt details to generate historical medical task resource debt items.

5. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the historical resource debt evolution data includes: Obtain unfinished tasks, unfinished task quantities, medical data object identifiers, and historical resource debt details from historical medical task resource debt items, and determine the source tasks of resource debts in the historical medical task association chain; Search for subsequent medical tasks that call the same medical data object along the direction of medical data object transmission, match the medical data processed by them with the unfinished tasks, determine the amount of unfinished tasks to be undertaken and the time of undertaking, and generate historical unfinished task undertaking records. Connect the resource debt source task with the first subsequent medical task, allocate operating system resources according to the proportion of unfinished tasks undertaken to the total unfinished tasks, and generate the first historical resource debt migration record; Process the remaining historical unfinished task acceptance records in sequence, take the previous subsequent medical task as the source of resource debt for this task, allocate the previous operating system resource amount according to the proportion of the current unfinished task amount to the previous unfinished task amount, and generate the remaining historical resource debt migration records. By connecting historical resource and debt migration records according to the medical data object identifier and the time of acceptance, a historical resource and debt migration chain is generated. Obtain operating system resource usage records before and after undertaking subsequent medical tasks, calculate the increase in resource usage and determine the historical resource debt rebound amount, and generate historical resource debt rebound records and historical resource debt rebound trajectories; By connecting historical medical task resource debt items, historical resource debt migration chains, and historical resource debt rebound trajectories, historical resource debt evolution data is generated.

6. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the current medical task resource debt includes: Read the task execution status and operating system resource usage process along the real-time medical task association chain, identify medical task pause and resource reduction behaviors, and generate the current resource release event; Determine the medical task corresponding to the current resource release event, obtain the task processing stage and the medical data objects to be processed, identify unfinished tasks and count the number of unfinished tasks; Obtain the operating system resource usage records before and after the current resource release event, calculate the differences in various resource usages, and generate the changes in operating system resource usage; The unfinished task volume is mapped to the difference in various resource usage, generating a resource debt mapping relationship between unfinished tasks and various operating system resources; Connect the current resource release event, medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, changes in operating system resource usage, and the corresponding relationship of resource debt to generate the current medical task resource debt item.

7. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the current resource debt evolution data includes: Obtain the medical task identifier, unfinished tasks, unfinished task quantity, medical data object identifier, and resource debt correspondence from the current medical task resource debt items, and determine the resource debt source task in the real-time medical task association chain; Starting with the resource debt source task, search for subsequent medical tasks that call the same medical data object along the direction of medical data object transmission, match the actual medical data processed with the unprocessed medical data corresponding to the unfinished task, determine the subsequent medical task, the amount of unfinished task to be undertaken, and the undertaking time, generate and arrange the current unfinished task undertaking record, and obtain the current unfinished task undertaking data. Connect the resource debt source task with the first subsequent medical task, and map the operating system resource quantity in the resource debt correspondence to the first subsequent medical task according to the proportion of the unfinished task quantity undertaken to the total unfinished task quantity, thereby generating the first current resource debt migration record; Process the remaining unfinished task assignment records in sequence, take the previous subsequent medical task as the resource debt source task of this time, and map the operating system resource amount in the previous current resource debt migration record to the corresponding subsequent medical task according to the proportion of the current unfinished task amount to the previous unfinished task amount, and generate the remaining current resource debt migration records. Connect the current resource and debt migration records sequentially according to the medical data object identifier and the time of acceptance to generate the current resource and debt migration chain; Connect the current medical task resource debt item to the starting position of the current resource debt migration chain to generate the current resource debt evolution data.

8. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The generation of the resource debt fulfillment forecast data includes: Extract the resource release event type, occurrence time, amount of unfinished tasks, amount of operating system resources, subsequent medical tasks and takeover time from the current resource debt evolution data and historical resource debt evolution data, and arrange them according to the resource debt migration order to generate current resource debt matching data and historical resource debt matching data; Historical resource debt matching data is filtered according to the type of resource release event, and aligned with the current resource debt matching data at each migration position. The difference in unfinished task volume, the difference in operating system resource volume, and the difference in adjacent acceptance time interval are calculated to generate historical resource debt evolution matching results. Obtain historical resource debt rebound records from the matched historical resource debt rebound trajectory, accumulate the historical resource debt rebound amount, determine the arrival time when it reaches the corresponding operating system resource amount, subsequent medical tasks and migration location, and generate historical resource debt repayment reference data; Based on the ratio of unfinished tasks and operating system resources between the current resource debt matching data and the historical resource debt matching data, the historical resource debt repayment reference data is corrected to generate the expected resource debt repayment time and the scale of resource debt rebound. Subsequent medical tasks are determined according to the migration location to generate a set of resource debt repayment tasks. By connecting the expected maturity time of resource debt, the scale of resource debt rebound, and the set of resource debt maturity tasks, resource debt maturity forecast data is generated.

9. The method for dynamic scheduling of operating system resources based on medical big data according to claim 1, characterized in that, The adjustments to the operating system resources corresponding to the medical task include: The system obtains the usage and total amount of various operating system resources from real-time medical operation data, calculates the available resources, and generates current available resource data. Obtain the operating system resource type and operating system resource quantity corresponding to each medical task from the resource debt repayment task set, and generate the resource data to be supplemented for each medical task; The order of resource debt release is determined according to the expected realization time of resource debt and the execution order of medical tasks. The release time and rate of resource debt are determined by combining the data on resources to be supplemented and the data on currently available resources, and a resource debt release plan is generated. According to the resource debt release plan, the operating system resources for the corresponding medical tasks will be increased sequentially, and the amount of operating system resources will be continuously adjusted according to the resource debt release rate.

Images