Clinical test item order processing method and device, equipment and storage medium
Through hybrid intelligent scheduling with information collection, priority evaluation and resource matching combined with adaptive algorithms, the problems of complex orders and dynamic resource scheduling in order processing of clinical trial projects are solved, and the efficiency and optimization effect of resource utilization is improved.
Patent Information
- Application Number
- CN202411167940.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-23
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing clinical trial project order processing methods cannot fully automatedly handle complex order information and dynamic resource scheduling requirements, resulting in inefficiency and unoptimized resource utilization.
Information collection and classification processing, priority evaluation, resource capability evaluation and matching processing are adopted, and hybrid algorithms of adaptive genetic algorithms and adaptive large neighborhood search algorithms are combined for intelligent scheduling and decision-making support, and an optimized scheduling plan is generated.
It significantly improves the efficiency and accuracy of order processing of clinical trial projects, optimizes resource allocation and utilization, and realizes intelligent management of complex orders and dynamic resource scheduling.
Smart Images

Figure CN118969157A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of order processing, and in particular to a method, device, equipment and storage medium for order processing of a clinical trial project. Background Art
[0002] With the rapid development of modern medicine and biotechnology, clinical trials play a vital role in new drug development, efficacy evaluation and safety testing. The management and execution of clinical trial projects involve complex processes and large amounts of data processing, and traditional order processing methods are difficult to meet the growing needs and complexity.
[0003] In the existing technology, the processing of clinical trial project orders usually adopts manual methods to collect information, classify, evaluate priorities and allocate resources. This method is not only inefficient, but also prone to human errors and unable to respond to the dynamic changes of clinical trials in real time. In addition, the scheduling and allocation of resources often rely on experience and subjective judgment, making it difficult to achieve optimal resource utilization and scheduling.
[0004] In order to improve the efficiency and accuracy of clinical trial project order processing, some studies have attempted to introduce information technology and intelligent algorithms. For example, data management systems are used to electronically process orders and perform preliminary classification and screening based on preset rules. However, these methods still have certain limitations and cannot fully automate the processing of complex order information and dynamic resource scheduling requirements. Summary of the invention
[0005] The main purpose of the present invention is to solve the existing technical problem that complex order information and dynamic resource scheduling requirements cannot be fully automatically processed.
[0006] A first aspect of the present invention provides a method for processing an order for a clinical trial project, the method comprising: Collect and classify the received clinical trial project orders to obtain a classified order set; According to the classified order set, each clinical trial project order is evaluated for priority to obtain a dynamic order list with priority scores; Based on the dynamic order list with priority scores, capacity evaluation and matching processing are performed on existing resources to obtain multiple optional resource allocation solutions; According to the multiple optional resource allocation schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used to perform intelligent scheduling and decision support processing to obtain an optimized scheduling plan; The clinical trial project order is processed according to the optimized scheduling plan.
[0007] Optionally, in a first implementation of the first aspect of the present invention, the information collection and classification processing of the received clinical trial project orders to obtain a classified order set includes: Performing information extraction processing on the received clinical trial project order to obtain an order data structure, wherein the order data structure includes test information and drug information; According to the test information in the order data structure, an urgency assessment process is performed on the clinical trial project order to obtain an order urgency index, wherein the test information includes the test type, the number of subjects required, and the expected test period; Based on the order urgency index and the preset classification rules, the clinical trial project orders are preliminarily classified to obtain a preliminary classification result, wherein the preliminary classification result includes emergency orders, regular orders and long-term orders; Performing secondary classification processing on the preliminary classification results according to the drug information in the order data structure to obtain secondary classification results divided by therapeutic areas; The secondary classification results are processed by time window allocation to obtain a classified order set with a processing time window, and the classified order set with the processing time window is integrated to obtain a classified order set.
[0008] Optionally, in a second implementation of the first aspect of the present invention, performing priority evaluation processing on each clinical trial project order according to the classified order set to obtain a dynamic order list with priority scores includes: Performing layer processing on the classified order set, and performing cluster analysis processing on each layer of orders in the order set to obtain a domain correlation clustering result; Based on the domain correlation clustering results, the resource demand of each cluster is quantified to obtain the cluster resource demand index, and the resource pressure assessment is performed according to the cluster resource demand index and the existing resource status to obtain the resource pressure coefficient of each cluster; Perform fuzzy comprehensive evaluation on the project value, sponsor importance, and technical difficulty of each clinical trial project order to obtain an order comprehensive value index, and perform dynamic weight allocation on each clinical trial project order based on the resource pressure coefficient and the order comprehensive value index to obtain a corresponding initial priority score; According to the level and initial priority score of the clinical trial project order, an inter-level priority balancing process is performed to obtain a priority score after inter-level balancing, and a time sensitivity adjustment process is performed on the priority score after inter-level balancing to obtain a priority score that takes time factors into consideration; Based on the priority scores taking the time factor into consideration, a multi-level feedback queue is constructed to obtain a dynamic order list with priority scores.
[0009] Optionally, in a third implementation of the first aspect of the present invention, the capability evaluation and matching processing of existing resources based on the dynamic order list with priority scores to obtain multiple optional resource configuration solutions include: Perform multi-dimensional feature extraction on existing resources to obtain a resource feature matrix, and perform cluster analysis on existing resources based on the resource feature matrix to obtain resource capability clustering results; Based on the resource capability clustering result and the dynamic order list with priority scores, preliminary resource allocation processing is performed to obtain an initial resource allocation plan; Performing load balancing analysis on the initial resource allocation plan to obtain a resource utilization index, and dynamically adjusting the initial resource allocation plan according to the resource utilization index to obtain a balanced resource allocation plan; Based on the balanced resource allocation scheme, performing resource conflict detection processing to obtain a resource conflict list, and performing iterative optimization processing on resource allocation according to the resource conflict list to obtain multiple candidate resource configuration schemes; The multiple candidate resource allocation schemes are subjected to multi-objective evaluation processing to obtain scheme evaluation indicators, and based on the scheme evaluation indicators, the candidate schemes are subjected to Pareto optimal screening processing to obtain multiple optional resource allocation schemes.
[0010] Optionally, in a fourth implementation of the first aspect of the present invention, the multi-dimensional feature extraction process is performed on the existing resources to obtain a resource feature matrix, and cluster analysis is performed on the existing resources according to the resource feature matrix to obtain a resource capability clustering result, which includes: Performing structural processing on the basic information of existing resources to obtain an initial resource data set, and classifying and coding the existing resources based on the initial resource data set to obtain a resource classification coding table; Based on the resource classification coding table, the historical usage data of existing resources is analyzed and processed in time series to obtain resource dynamic characteristic indicators, and the professional capability information of existing resources is analyzed and processed semantically to obtain resource professional capability vectors; According to the resource dynamic feature index and the resource professional capability vector, feature fusion processing is performed to obtain a multi-dimensional resource feature vector, and dimension reduction processing is performed on the multi-dimensional resource feature vector to obtain a resource feature matrix after dimension reduction; Based on the resource feature matrix after dimension reduction, a hierarchical clustering algorithm is used to perform resource clustering processing to obtain a preliminary clustering result, and a silhouette coefficient analysis is performed on the preliminary clustering result to obtain an optimal number of clusters; According to the optimal number of clusters, the preliminary clustering results are refined and clustered to obtain resource capability clustering results.
[0011] Optionally, in a fifth implementation of the first aspect of the present invention, the hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used to perform intelligent scheduling and decision support processing according to the multiple optional resource configuration schemes to obtain an optimized scheduling plan, which includes: Performing natural number encoding processing on the multiple optional resource configuration schemes to obtain an initial population, and performing fitness calculation processing on individuals in the initial population to obtain individual fitness values; Based on the individual fitness values, the initial population is subjected to a tournament selection operation to obtain a parent individual, and the parent individual is subjected to an enhanced best path random insertion crossover operation to obtain a child individual; According to the fitness value of the offspring individual, an adaptive crossover probability calculation process is performed to obtain a dynamic crossover probability, and based on the dynamic crossover probability, a crossover operation process is performed on the offspring individual to obtain an offspring individual after the crossover; Performing adaptive large neighborhood search operator initialization processing on the offspring individuals after the crossover to obtain an initial removal operator set, an insertion operator set and their corresponding initial weights; Iteratively controlling the optimization process according to the initial removal operator set, the insertion operator set and their corresponding initial weights, repeatedly performing operator selection, local search, and weight update to obtain a final optimized population; The best individual in the final optimized population is decoded to obtain an optimized scheduling plan.
[0012] Optionally, in a sixth implementation of the first aspect of the present invention, the iterative control processing of the optimization process according to the initial removal operator set, the insertion operator set and their corresponding initial weights, repeatedly performing operator selection, local search, and weight update to obtain the final optimized population includes: According to the initial removal operator set, the insertion operator set and their corresponding initial weights, a roulette wheel method is used to perform operator selection processing to obtain a removal operator and an insertion operator for the current iteration; Based on the selected removal operator and insertion operator, a local search process is performed on the offspring individuals to obtain locally optimized individuals; Performing search performance evaluation on the locally optimized individuals to obtain an effect score of the operator; Performing an operator weight update process according to the effect score of the operator usage to obtain an updated removal operator weight and an insertion operator weight; Based on the fitness value of the locally optimized individual, an adaptive mutation probability calculation process is performed to obtain a dynamic mutation probability, and according to the dynamic mutation probability, a mutation operation process is performed on the locally optimized individual to obtain a mutated individual; Performing population updating processing on the mutated individuals to obtain a new generation of population; Based on the new generation population, the updated removal operator weights and the insertion operator weights, the optimization process is iteratively controlled, and operator selection, local search, and weight update are repeatedly performed to obtain a final optimized population.
[0013] A second aspect of the present invention provides an order processing device for a clinical trial project, the order processing device for a clinical trial project comprising: The order classification module is used to collect and classify the received clinical trial project orders to obtain a classified order set; A priority evaluation module, used to perform a priority evaluation process on each clinical trial project order according to the classified order set, and obtain a dynamic order list with priority scores; A resource evaluation module, for evaluating and matching existing resources based on the dynamic order list with priority scores to obtain multiple optional resource allocation solutions; A scheduling module, for performing intelligent scheduling and decision support processing by using a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm according to the multiple optional resource configuration schemes to obtain an optimized scheduling plan; A processing module is used to process the clinical trial project order according to the optimized scheduling plan.
[0014] The third aspect of the present invention provides an order processing device for a clinical trial project, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via lines; the at least one processor calls the instructions in the memory so that the order processing device for the clinical trial project executes the steps of the above-mentioned order processing method for the clinical trial project.
[0015] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the above-mentioned order processing method for clinical trial projects.
[0016] The above-mentioned order processing method, device, equipment and storage medium for clinical trial projects collect information and classify clinical trial project orders to obtain a classified order set; based on the order set, each clinical trial project order is evaluated for priority to obtain a dynamic order list with priority scores; based on the dynamic order list with priority scores, the existing resources are evaluated for capacity and matched to obtain multiple optional resource allocation schemes; based on multiple optional resource allocation schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used for intelligent scheduling and decision support processing to obtain an optimized scheduling plan and process clinical trial project orders. This method realizes intelligent scheduling and decision support for clinical trial project orders, significantly improves the efficiency and accuracy of order processing, and optimizes resource allocation and utilization.
[0017] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.
[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 This is a schematic diagram of a first embodiment of a method for processing an order for a clinical trial project in an embodiment of the present invention; Figure 2 A schematic diagram of an embodiment of an order processing device for clinical trial projects in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of an order processing device for clinical trial projects in an embodiment of the present invention. DETAILED DESCRIPTION
[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0021] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.
[0022] To facilitate understanding of this embodiment, firstly, a method for processing an order for a clinical trial project disclosed in an embodiment of the present invention is described in detail. Figure 1 As shown, the method comprises the following steps: 101. Collect and classify the received clinical trial project orders to obtain a classified order set; In one embodiment of the present invention, the information collection and classification processing of the received clinical trial project orders to obtain a classified order set includes: performing information extraction processing on the received clinical trial project orders to obtain an order data structure, wherein the order data structure includes test information and drug information; performing urgency assessment processing on the clinical trial project orders according to the test information in the order data structure to obtain an order urgency index, wherein the test information includes the test type, the required number of subjects and the expected test cycle; performing preliminary classification processing on the clinical trial project orders based on the order urgency index and preset classification rules to obtain preliminary classification results, wherein the preliminary classification results include emergency orders, regular orders and long-term orders; performing secondary classification processing on the preliminary classification results according to the drug information in the order data structure to obtain secondary classification results divided by therapeutic field; performing time window allocation processing on the secondary classification results to obtain a classified order set with a processing time window, and integrating the classified order sets with the processing time window to obtain a classified order set.
[0023] Specifically, the received clinical trial project orders are comprehensively extracted and processed to obtain a structured order data structure. This data structure not only contains trial information, such as trial type, number of subjects required, and expected trial period, but also covers drug information, such as drug name, mechanism of action, and indications. For example, for a Phase II clinical trial order of a new anticancer drug, its trial information may include "Phase II trial", "100 subjects required", and "expected trial period of 12 months", while drug information may include "new targeted drug XYZ", "selectively inhibit specific cancer cell signaling pathways", and "used to treat advanced pancreatic cancer". Next, the system will evaluate the urgency of the clinical trial project order based on the trial information in the order data structure. This process will take into account multiple factors, such as the urgency of the trial type (for example, Phase I trials are usually more urgent than Phase III trials), the number of subjects required (a smaller sample size may mean a faster recruitment process), and the expected trial period (a shorter period may require a faster start). Through a comprehensive analysis of these factors, the system will calculate an order urgency index for each order. Based on this order urgency index and the pre-set classification rules, the system then performs a preliminary classification process on the clinical trial project orders. This process divides the orders into three categories: urgent orders, regular orders, and long-term orders. For example, a new drug phase I trial with a high urgency index may be classified as an urgent order, while a large-scale phase III trial may be classified as a regular order or a long-term order. However, classification based on urgency alone may not be precise enough, so the system will further perform secondary classification processing on the preliminary classification results based on the drug information in the order data structure. This processing process is mainly divided according to therapeutic areas, such as cardiovascular diseases, tumors, and nervous system diseases. For example, in the urgent order category, it may be further subdivided into subcategories such as "urgent-tumor" and "urgent-cardiovascular". This secondary classification can help resource allocation more accurately, because different therapeutic areas may require different professional resources and equipment. After the classification is completed, the system will perform time window allocation processing on the secondary classification results. The purpose of this step is to assign a suitable processing time window to each order to facilitate subsequent resource allocation and scheduling. The allocation of time windows takes into account the urgency of the order, the characteristics of the treatment area, and the overall availability of resources. For example, an order classified as "urgent-oncology" may be assigned an earlier time window, while an order for "long-term-chronic disease" may be assigned to a later time window. Finally, the system will integrate these classified order sets with processing time windows to form the final classified order set. This integration process is not just a simple merger, but also involves the priority balance between orders and the preliminary planning of resource allocation.For example, the system might ensure that orders in different therapeutic areas are given a certain chance to be processed during each time period, to prevent orders in a particular area from being shelved for a long time.
[0024] 102. According to the classified order set, perform priority evaluation processing on each clinical trial project order to obtain a dynamic order list with priority scores; In one embodiment of the present invention, the priority evaluation process is performed on each clinical trial project order according to the classified order set to obtain a dynamic order list with priority scores, including: layering the classified order set, and performing cluster analysis on each layer of orders in the order set to obtain a domain-related clustering result; based on the domain-related clustering result, quantifying the resource demand of each cluster to obtain a cluster resource demand index, and performing resource pressure evaluation based on the cluster resource demand index and the existing resource status to obtain a resource pressure coefficient of each cluster; and evaluating the project value, application and management of each clinical trial project order. The importance of the clinical trial project and the technical difficulty are fuzzily evaluated to obtain the comprehensive value index of the order, and based on the resource pressure coefficient and the comprehensive value index of the order, each clinical trial project order is dynamically weighted to obtain the corresponding initial priority score; according to the level and initial priority score of the clinical trial project order, an inter-layer priority balancing process is performed to obtain the priority score after inter-layer balancing, and the time sensitivity adjustment process is performed on the priority score after inter-layer balancing to obtain the priority score considering the time factor; based on the priority score considering the time factor, a multi-level feedback queue is constructed to obtain a dynamic order list with priority scores.
[0025] Specifically, the classified order set is stratified, usually into three layers: high priority, medium priority, and low priority. This stratification is based on the previous urgency assessment results. For example, urgent orders may be classified as high priority, regular orders as medium priority, and long-term orders as low priority. Next, the system will perform cluster analysis on the orders in each layer to obtain domain relevance clustering results. This cluster analysis is mainly based on the therapeutic field of the order, such as clustering all tumor-related clinical trial projects into one category and cardiovascular disease-related projects into another category. The purpose of this is to better evaluate and allocate resource requirements in specific fields. Next, based on these domain relevance clustering results, the system will quantify the resource requirements of each cluster. This process takes into account the characteristics of the clinical trial projects in each cluster, such as the complexity of the trial, the required professional equipment and personnel, etc., to derive a cluster resource requirement index. For example, a cluster containing multiple complex early-stage tumor clinical trials may have a high resource requirement index because such trials usually require more professional equipment and highly skilled researchers. Then, the system will compare this cluster resource demand index with the existing resource status, conduct resource pressure assessment, and finally obtain the resource pressure coefficient of each cluster. If the resource demand of a cluster far exceeds the current available resources, then the resource pressure coefficient of this cluster will be very high. At the same time, the system will perform fuzzy comprehensive evaluation on the project value, sponsor importance and technical difficulty of each clinical trial project order. This process uses fuzzy logic to deal with these factors that are difficult to accurately quantify, and finally obtains an order comprehensive value index. For example, a clinical trial of an innovative drug for a rare disease initiated by a well-known pharmaceutical company may get a higher comprehensive value index because it has high project value (may bring major medical breakthroughs), high sponsor importance (well-known company), and high technical difficulty (rare disease research). With the resource pressure coefficient and order comprehensive value index, the system can dynamically assign weights to each clinical trial project order to obtain an initial priority score. This process balances resource pressure and project value. For example, even if a project has a high comprehensive value, if the resource pressure of its cluster is already very high, then its initial priority score may be appropriately reduced. However, based solely on the initial priority score, some orders in certain tiers may not be processed for a long time, so the system will balance the priorities between tiers according to the tier to which the order belongs and the initial priority score. This process ensures that even orders in the low-priority tier have a chance to be processed, thereby obtaining a priority score after balancing between tiers. For example, the system may periodically increase the scores of some low-priority orders with a long waiting time so that they have a chance to be processed. Next, the system will adjust the time sensitivity of the priority scores after balancing between tiers.This step takes into account the timeliness of the order, such as the expected start date of the clinical trial, the duration of drug patent protection, and other factors. For example, if the expected start date of a clinical trial project is approaching, its priority score may be significantly increased. In this way, the system finally obtains a priority score that takes time factors into account. Finally, based on these priority scores that take time factors into account, the system builds a multi-level feedback queue. This queue is not static, but can be dynamically adjusted based on factors such as the addition of new orders and changes in the status of existing orders. For example, if a high-priority order has not been processed for a long time, the system may automatically promote it to a higher priority queue. In this way, the system finally obtains a dynamic order list with priority scores.
[0026] 103. Based on the dynamic order list with priority scores, capability evaluation and matching processing are performed on existing resources to obtain multiple optional resource allocation solutions; In one embodiment of the present invention, the capability assessment and matching processing of existing resources based on the dynamic order list with priority scores to obtain multiple optional resource allocation schemes includes: performing multi-dimensional feature extraction processing on existing resources to obtain a resource feature matrix, and performing cluster analysis processing on existing resources based on the resource feature matrix to obtain resource capability clustering results; performing preliminary resource allocation processing based on the resource capability clustering results and the dynamic order list with priority scores to obtain an initial resource allocation scheme; performing load balancing analysis processing on the initial resource allocation scheme to obtain a resource utilization index, and dynamically adjusting the initial resource allocation scheme based on the resource utilization index to obtain a balanced resource allocation scheme; performing resource conflict detection processing based on the balanced resource allocation scheme to obtain a resource conflict list, and iteratively optimizing resource allocation based on the resource conflict list to obtain multiple candidate resource allocation schemes; performing multi-objective evaluation processing on the multiple candidate resource allocation schemes to obtain scheme evaluation indicators, and performing Pareto optimal screening processing on the candidate schemes based on the scheme evaluation indicators to obtain multiple optional resource allocation schemes.
[0027] Specifically, the existing resources are subjected to multi-dimensional feature extraction to obtain a resource feature matrix. This matrix contains multiple features of each resource, such as the professional field of the research center, the equipment level, and the experience of the personnel. For example, for a specific research center, its features may include "oncology professional level: 0.9", "number of PET-CT equipment: 2", "years of clinical trial experience: 10", etc. Then, the system will cluster the existing resources based on this resource feature matrix using a clustering analysis algorithm (such as K-means or hierarchical clustering) to obtain a resource capacity clustering result. This result may divide the resources into several main categories, such as "advanced tumor research center", "cardiovascular specialty hospital", "comprehensive tertiary hospital", etc. With the resource capacity clustering results and the dynamic order list with priority scores, the system can perform preliminary resource allocation processing to obtain an initial resource allocation plan. This process attempts to match high-priority orders to the most suitable resource cluster. For example, a high-priority tumor immunotherapy clinical trial project may be preferentially allocated to resources in the "advanced tumor research center" category. However, this initial allocation may result in some resources being overused while others are idle. Therefore, the system will perform a load balancing analysis on this initial resource allocation plan to obtain a resource utilization index. This index reflects the usage of each resource and may include "working time percentage" and "equipment utilization rate". For example, the system may find that the utilization rate of the PET-CT equipment in a research center has reached 95%, while the utilization rate of similar equipment in another center is only 50%. Based on these resource utilization indicators, the system will dynamically adjust the initial resource allocation plan to obtain a balanced resource allocation plan. This process may reallocate some orders originally assigned to high-load resources to low-load resources to achieve overall load balancing. However, even after load balancing, conflicts may still exist in the resource allocation plan. For example, two high-priority projects may be scheduled to use the same equipment at the same time. Therefore, based on the balanced resource allocation plan, the system will perform resource conflict detection and obtain a resource conflict list. This list may include all conflicts in time or resources, such as "Project A and Project B need to use PET-CT-1 equipment at the same time on July 1st." With this conflict list, the system can perform iterative optimization to try to resolve these conflicts. This process may involve adjusting project time, replacing resources, and even reallocating projects. Through multiple iterations, the system eventually obtains multiple candidate resource allocation schemes, each of which is a feasible, conflict-free resource allocation scheme. Finally, the system will perform multi-objective evaluation on these candidate resource allocation schemes to obtain scheme evaluation indicators. These indicators may include "overall resource utilization", "high priority project satisfaction", "project completion time", etc.For example, plan A may perform best in terms of resource utilization, while plan B is better at meeting the needs of high-priority projects. Based on these evaluation indicators, the system will perform Pareto optimal screening and obtain multiple optional resource allocation plans. Pareto optimality means that without compromising any goal, the state of other goals can no longer be improved. It means that these finally selected plans have achieved a balance in various evaluation indicators, and no plan is worse than other plans in all indicators. This complex resource evaluation and matching process can fully consider multiple factors such as resource characteristics, order priorities, load balancing, and multi-objective optimization, thereby obtaining a series of high-quality resource allocation plans. For example, the final options may include: Plan 1 gives priority to meeting the needs of all high-priority projects, but the overall resource utilization is low; Plan 2 performs best in resource utilization, but may require a slight delay in some medium-priority projects; Plan 3 achieves a good balance in all indicators. In this way, decision makers can choose the most suitable one from these optimized plans based on the current specific situation and strategic focus.
[0028] Furthermore, the multi-dimensional feature extraction process is performed on the existing resources to obtain a resource feature matrix, and the existing resources are clustered and analyzed based on the resource feature matrix to obtain resource capability clustering results, including: structuring the basic information of the existing resources to obtain an initial resource data set, and classifying and coding the existing resources based on the initial resource data set to obtain a resource classification coding table; based on the resource classification coding table, time series analysis is performed on the historical usage data of the existing resources to obtain resource dynamic feature indicators, and semantic analysis is performed on the professional capability information of the existing resources to obtain a resource professional capability vector; based on the resource dynamic feature indicators and the resource professional capability vector, feature fusion is performed to obtain a multi-dimensional resource feature vector, and dimension reduction is performed on the multi-dimensional resource feature vector to obtain a resource feature matrix after dimension reduction; based on the resource feature matrix after dimension reduction, a hierarchical clustering algorithm is used to perform resource clustering to obtain a preliminary clustering result, and a contour coefficient analysis is performed on the preliminary clustering result to obtain an optimal number of clusters; based on the optimal number of clusters, the preliminary clustering result is refined clustering to obtain a resource capability clustering result.
[0029] Specifically, the system will perform structured processing on the basic information of existing resources to obtain an initial resource data set. This data set may include the name of the research center, geographical location, number of beds, list of professional equipment, as well as the professional background and work experience of researchers. For example, for a specific research center, its initial data may include "name: First Affiliated Hospital of Medical University", "location: Haidian District, Beijing", "number of beds: 500", "main equipment: PET-CT scanner, high-throughput sequencer", etc. Next, the system will classify and encode the existing resources based on this initial resource data set to obtain a resource classification code table. This step classifies and encodes resources according to different dimensions, such as classification by geographical location (such as: Beijing-01, Shanghai-02, etc.), classification by professional field (such as: oncology-A, cardiovascular-B, etc.), classification by equipment type (such as: imaging equipment-I, laboratory equipment-L, etc.). This classification and coding can help the system manage and retrieve resource information more efficiently. With the resource classification code table, the system can perform time series analysis and processing on the historical usage data of existing resources based on this to obtain resource dynamic characteristic indicators. This process analyzes the usage of each resource over the past period of time, including frequency of use, duration of use, idle time, etc. For example, the system may find that the average daily usage time of a PET-CT scanner in the past three months was 6 hours, and the peak usage rate reached 90%. These dynamic feature indicators can reflect the actual utilization of resources and potential bottlenecks. At the same time, the system will also perform semantic analysis on the professional capability information of existing resources to obtain the resource professional capability vector. This step analyzes the professional descriptions of research centers and researchers, extracts keywords and concepts, and converts them into numerical vectors. For example, a researcher's professional capability vector may include values such as "oncology: 0.9, immunotherapy: 0.8, clinical trial management: 0.7", reflecting the researcher's professional level in different fields. Next, the system will perform feature fusion processing based on the resource dynamic feature indicators and resource professional capability vector to obtain a multi-dimensional resource feature vector. This fusion process combines the dynamic usage of resources with static professional capabilities to form a comprehensive resource description. However, this multi-dimensional feature vector may have a high dimensionality, which is not conducive to subsequent analysis and calculation. Therefore, the system will perform dimensionality reduction processing on this multi-dimensional resource feature vector to obtain a resource feature matrix after dimensionality reduction. Commonly used dimensionality reduction methods include principal component analysis (PCA) or t-SNE, which can greatly reduce the dimension of data while retaining the main information. With the reduced resource feature matrix, the system can use a hierarchical clustering algorithm to perform resource clustering processing and obtain preliminary clustering results. The hierarchical clustering algorithm will gradually cluster similar resources together to form a tree-like clustering structure.For example, the system may first cluster all oncology-related research centers together, and then further cluster centers that are good at immunotherapy into one group and those that are good at targeted therapy into another group within this large category. However, the preliminary clustering results may not be optimal, so the system will perform silhouette coefficient analysis on the preliminary clustering results to obtain the optimal number of clusters. The silhouette coefficient is a method to evaluate clustering quality. It measures the similarity of each sample (here, each resource) to its own cluster and the difference from other clusters. By calculating the average silhouette coefficient under different numbers of clusters, the system can find a balance point at which the resources within the cluster are similar enough and the resources between different clusters are different enough. Finally, based on the optimal number of clusters obtained, the system will perform refined clustering on the preliminary clustering results to obtain the final resource capacity clustering results. This refinement process may adjust the attribution of some resources to ensure that each cluster is compact and meaningful. For example, the final clustering results may include several major categories such as "Advanced Tumor Research Center", "Cardiovascular Specialty Hospital", and "Comprehensive Tertiary Hospital", and the resources within each category are relatively similar in capabilities and characteristics.
[0030] 104. According to the multiple optional resource allocation schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used to perform intelligent scheduling and decision support processing to obtain an optimized scheduling plan; In one embodiment of the present invention, the method of performing intelligent scheduling and decision support processing based on the multiple optional resource configuration schemes by using a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm to obtain an optimized scheduling plan includes: performing natural number encoding processing on the multiple optional resource configuration schemes to obtain an initial population, and performing fitness calculation processing on individuals in the initial population to obtain individual fitness values; based on the individual fitness values, performing a tournament selection operation processing on the initial population to obtain parent individuals, and performing an enhanced optimal path random insertion crossover operation processing on the parent individuals to obtain offspring individuals; and performing a random insertion crossover operation processing on the offspring individuals according to the fitness values of the offspring individuals. The fitness value of is used to perform adaptive crossover probability calculation processing to obtain a dynamic crossover probability, and based on the dynamic crossover probability, a crossover operation is performed on the offspring individuals to obtain the offspring individuals after the crossover; an adaptive large neighborhood search operator initialization processing is performed on the offspring individuals after the crossover to obtain an initial removal operator set, an insertion operator set and their corresponding initial weights; an iterative control processing is performed on the optimization process according to the initial removal operator set, the insertion operator set and their corresponding initial weights, and operator selection, local search and weight update are repeatedly performed to obtain a final optimized population; the best individual in the final optimized population is decoded to obtain an optimized scheduling plan.
[0031] Specifically, multiple optional resource allocation schemes are first encoded with natural numbers to obtain an initial population. For example, each resource allocation scheme may be encoded as a sequence of numbers, where each number represents a specific resource allocation decision. Assuming there is a scheme containing 5 clinical trial projects and 3 research centers, it may be encoded as [1,2,3,1,2], indicating that project 1 is allocated to research center 1, project 2 is allocated to research center 2, and so on. Next, the system will perform fitness calculations on each individual in this initial population (i.e., each encoded resource allocation scheme) to obtain an individual fitness value. The fitness value reflects the pros and cons of each scheme, and may consider multiple factors such as resource utilization, project completion time, and cost. For example, a scheme that can achieve high resource utilization while ensuring timely completion of high-priority projects may obtain a higher fitness value. Based on these individual fitness values, the system performs a tournament selection operation to obtain parent individuals. In the tournament selection, the system may randomly select several individuals for comparison and select the one with the highest fitness as the parent. This process simulates natural selection and ensures that excellent schemes have a higher chance of being retained and passed on. Then, the system performs an enhanced best path random insertion crossover operation on these parent individuals to obtain offspring individuals. This operation may involve inserting certain allocation decisions from one solution into another solution to produce a new and potentially better solution. For example, if solution A performs well in processing high-priority projects, while solution B is better in resource utilization, the crossover operation may produce a new solution that combines the two advantages. Next, the system performs an adaptive crossover probability calculation based on the fitness value of the offspring individuals to obtain a dynamic crossover probability. The purpose of this step is to dynamically adjust the frequency of crossover operations based on the current state of the population. For example, if the population diversity is low, the system may increase the crossover probability to increase variation; conversely, if many excellent solutions have been generated, the system may reduce the crossover probability to maintain these good solutions. Based on this dynamic crossover probability, the system performs a crossover operation on the offspring individuals to obtain offspring individuals after crossover. At this point, the algorithm introduces the element of adaptive large neighborhood search (ALNS). The system performs an adaptive large neighborhood search operator initialization process on the offspring individuals after crossover to obtain the initial removal operator set, insertion operator set and their corresponding initial weights. These operators may include removal operators such as "randomly remove an item's assignment" and "remove an item from the research center with the highest load", and insertion operators such as "insert the item into the nearest available time slot" and "insert the item into the research center with the lowest load". Each operator has an initial weight that reflects the probability of it being selected for use. Next, the system enters an iterative optimization loop. In each iteration, the system first selects a removal operator and an insertion operator based on the weights of the operators.For example, it may choose "remove a project from the research center with the highest load" as the removal operator, and "insert the project into the research center with the lowest load" as the insertion operator. Then, the system uses this pair of operators to perform a local search on the current solution and try to improve the quality of the solution. If the new solution is better than the original solution (i.e., the fitness is higher), the system accepts the new solution and increases the weight of the operator pair used; otherwise, the system may accept the new solution with a certain probability (which helps to jump out of the local optimum) and slightly reduce the weight of the operator pair used. This process will be repeated until the preset termination condition is reached (such as reaching the maximum number of iterations, or there is no improvement in multiple consecutive iterations). In this way, the system is able to strike a balance between global search (through crossover and mutation of genetic algorithms) and local optimization (through ALNS). Finally, the system decodes the best individuals in the final optimized population to obtain an optimized scheduling plan. This scheduling plan may list in detail the start time of each clinical trial project, the research center and specific resources used, and the expected completion time. For example, the final scheduling plan may show: "Project A (high priority) will start on July 1st at Research Center 1, using PET-CT-1 equipment, and is expected to be completed on August 15th; Project B (medium priority) will start on July 15th at Research Center 2, using MRI-2 equipment, and is expected to be completed on September 30th...", etc. This plan not only takes into account the priority of the project and the availability of resources, but also balances multiple goals through a complex optimization process, such as maximizing resource utilization and minimizing project delays.
[0032] Furthermore, the optimization process is iteratively controlled according to the initial removal operator set, the insertion operator set and their corresponding initial weights, and the operator selection, local search and weight update are repeatedly performed to obtain the final optimized population, including: according to the initial removal operator set, the insertion operator set and their corresponding initial weights, the roulette wheel method is used to select operators to obtain removal operators and insertion operators for the current iteration; based on the selected removal operators and insertion operators, local search is performed on offspring individuals to obtain locally optimized individuals; search performance evaluation is performed on the locally optimized individuals to obtain effect scores of operator use; based on According to the effect score of the operator, the operator weight is updated to obtain the updated removal operator weight and insertion operator weight; based on the fitness value of the locally optimized individual, the adaptive mutation probability is calculated to obtain the dynamic mutation probability, and according to the dynamic mutation probability, the locally optimized individual is mutated to obtain the mutated individual; the mutated individual is updated to obtain a new generation population; based on the new generation population and the updated removal operator weight and insertion operator weight, the optimization process is iteratively controlled, and the operator selection, local search, and weight update are repeatedly executed to obtain the final optimized population.
[0033] Specifically, the roulette wheel method is first used to perform operator selection processing to obtain the removal operator and insertion operator for the current iteration. The roulette wheel method simulates a weighted random selection process, and the higher the weight, the greater the probability of the operator being selected. For example, suppose there are three removal operators: random removal (weight 0.3), worst removal (weight 0.5), and related removal (weight 0.2). The system will randomly select a removal operator based on these weights. The same method is also used to select the insertion operator. After the operator is selected, the system will perform local search processing on the offspring individuals based on this pair of operators to obtain the locally optimized individuals. In the specific scenario of clinical trial project scheduling, this may mean first removing some project allocations from the current plan (using the selected removal operator) and then reallocating these projects (using the selected insertion operator). For example, if the "worst removal" and "greedy insertion" operator pair is selected, the system may first remove several project allocations in the current plan that result in the lowest resource utilization, and then try to reinsert these projects into positions that can maximize resource utilization. After completing the local search, the system will evaluate the search performance of the locally optimized individuals and obtain the effect score of the operator use. This score reflects the contribution of the selected operator to improving the quality of the solution. For example, if the overall fitness of the solution is improved by 10% after using a pair of operators, then this pair of operators may obtain a higher effect score. Based on these effect scores, the system updates the operator weights to obtain updated removal operator weights and insertion operator weights. Operators with good performance will receive increased weights, while operators with poor performance will receive reduced weights. This adaptive mechanism enables the algorithm to gradually "learn" which operators are more effective during the search process, thereby improving the search efficiency of subsequent iterations. Next, the system performs adaptive mutation probability calculation based on the fitness value of the locally optimized individuals to obtain a dynamic mutation probability. The purpose of this step is to dynamically adjust the degree of mutation according to the quality of the current solution. For example, if the current solution is already very good, the system may reduce the mutation probability to avoid destroying the good solution; if the current solution quality is not high, the system may increase the mutation probability to increase the chance of exploring new solutions. Then, the system performs mutation operations on the locally optimized individuals based on this dynamic mutation probability to obtain the mutated individuals. In the specific scenario of clinical trial project scheduling, mutation may manifest as a random change in the allocation of certain projects. For example, a project may be randomly selected and adjusted from the original research center to another research center. This operation helps maintain the diversity of the population and prevents the algorithm from falling into a local optimal solution. The mutated individuals are updated to obtain a new generation of population. This process usually retains some of the best solutions in the current population while introducing newly generated solutions to balance "utilization" (retaining good solutions) and "exploration" (introducing new solutions).Finally, based on this new generation population, the updated removal operator weights and insertion operator weights, the system iteratively controls the optimization process, repeatedly performing steps such as operator selection, local search, and weight update. This process continues until a certain termination condition is reached, such as reaching the maximum number of iterations, or there is no significant improvement in multiple consecutive iterations. In this way, the system finally obtains the final optimized population. To illustrate this process, let's take a specific example: suppose that in a certain iteration, the system selects the operator pair of "load balancing removal" (removing a project from the research center with the highest load) and "earliest available time insertion" (inserting the project into the earliest available time slot). After using this pair of operators, the system successfully moved a high-priority project from an overloaded research center to another research center with a lower load, which not only improved the overall resource utilization balance, but also shortened the estimated completion time of the project. This successful local search will cause the weight of this pair of operators to increase. In the next iteration, the probability of this pair of operators being selected will be higher. At the same time, the system may slightly reduce the probability of mutation because the current solution is already quite good. Then, the system may make small random adjustments (mutations) to the allocation of some other projects to obtain a new generation of population. This process is repeated over and over again, with each iteration potentially bringing small improvements, eventually converging to a high-quality solution.
[0034] 105. Process the clinical trial project order according to the optimized scheduling plan.
[0035] In one embodiment of the present invention, a specific start time, research center, and required resources are assigned to each clinical trial project according to the scheduling plan. For example, Project A may be scheduled to start on July 1 in the oncology department of a tertiary hospital, using a specific PET-CT device. Then, the system will send a detailed project scheduling notification to the relevant research centers and researchers, including a project schedule, resource usage plan, and specific operating guidelines. At the same time, the system will also generate a detailed execution plan for each project, including key information such as a subject recruitment schedule, a drug supply plan, and data collection nodes. In addition, the system will establish a real-time monitoring mechanism to track the progress of each project, including the actual start time, resource usage, subject enrollment speed, etc., and compare this information with the original plan. If a significant deviation is found, the system will trigger an alarm and make adjustment suggestions. For example, when the subject recruitment speed is slower than expected, it may be recommended to increase recruitment efforts or extend recruitment time.
[0036] In this embodiment, by collecting information and classifying clinical trial project orders, a classified order set is obtained; according to the order set, each clinical trial project order is evaluated for priority to obtain a dynamic order list with priority scores; based on the dynamic order list with priority scores, the existing resources are evaluated for capacity and matched to obtain multiple optional resource allocation schemes; according to multiple optional resource allocation schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used for intelligent scheduling and decision support processing to obtain an optimized scheduling plan and process clinical trial project orders. This method realizes intelligent scheduling and decision support for clinical trial project orders, significantly improves the efficiency and accuracy of order processing, and optimizes resource allocation and utilization.
[0037] The above describes the order processing method for clinical trial projects in the embodiment of the present invention. The following describes the order processing device for clinical trial projects in the embodiment of the present invention. Figure 2 An embodiment of the order processing device for clinical trial projects in the embodiment of the present invention includes: The order classification module 201 is used to collect and classify the received clinical trial project orders to obtain a classified order set; A priority evaluation module 202, configured to perform a priority evaluation process on each clinical trial project order according to the classified order set, and obtain a dynamic order list with priority scores; A resource evaluation module 203 is used to perform capacity evaluation and matching processing on existing resources based on the dynamic order list with priority scores to obtain multiple optional resource configuration solutions; The scheduling module 204 is used to perform intelligent scheduling and decision support processing according to the multiple optional resource configuration schemes by using a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm to obtain an optimized scheduling plan; The processing module 205 is used to process the clinical trial project order according to the optimized scheduling plan.
[0038] In an embodiment of the present invention, the order processing device of the clinical trial project runs the order processing method of the clinical trial project, and the order processing device of the clinical trial project obtains a classified order set by collecting information and classifying the clinical trial project orders; performs priority evaluation processing on each clinical trial project order according to the order set to obtain a dynamic order list with priority scores; performs capacity evaluation and matching processing on existing resources based on the dynamic order list with priority scores to obtain multiple optional resource configuration schemes; and uses a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm to perform intelligent scheduling and decision support processing according to multiple optional resource configuration schemes, obtains an optimized scheduling plan, and processes the clinical trial project orders. The method realizes intelligent scheduling and decision support for clinical trial project orders, significantly improves the efficiency and accuracy of order processing, and optimizes resource allocation and utilization.
[0039] above Figure 2 The order processing device for clinical trial projects in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The order processing device for clinical trial projects in the embodiment of the present invention is described in detail from the perspective of hardware processing.
[0040] Figure 3 3 is a schematic diagram of the structure of an order processing device for a clinical trial project provided by an embodiment of the present invention. The order processing device 300 for the clinical trial project may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 may be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the order processing device 300 for the clinical trial project. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the order processing device 300 for the clinical trial project to implement the steps of the order processing method for the clinical trial project described above.
[0041] The order processing device 300 for clinical trial projects may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The structure of the order processing equipment for clinical trial projects shown does not constitute a limitation on the order processing equipment for clinical trial projects provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.
[0042] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein instructions are stored in the computer, and when the instructions are executed on a computer, the computer executes the steps of the order processing method for the clinical trial project.
[0043] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.
[0044] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.
[0045] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for processing an order for a clinical trial project, characterized in that: The order processing method for the clinical trial project includes: Collect and classify the received clinical trial project orders to obtain a classified order set; According to the classified order set, each clinical trial project order is evaluated for priority to obtain a dynamic order list with priority scores; Based on the dynamic order list with priority scores, capacity evaluation and matching processing are performed on existing resources to obtain multiple optional resource allocation solutions; According to the multiple optional resource allocation schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used to perform intelligent scheduling and decision support processing to obtain an optimized scheduling plan; The clinical trial project order is processed according to the optimized scheduling plan.
2. The order processing method for clinical trial projects according to claim 1, characterized in that: The information collection and classification of the received clinical trial project orders is performed to obtain a classified order set including: Performing information extraction processing on the received clinical trial project order to obtain an order data structure, wherein the order data structure includes test information and drug information; According to the test information in the order data structure, an urgency assessment process is performed on the clinical trial project order to obtain an order urgency index, wherein the test information includes the test type, the number of subjects required, and the expected test period; Based on the order urgency index and the preset classification rules, the clinical trial project orders are preliminarily classified to obtain a preliminary classification result, wherein the preliminary classification result includes emergency orders, regular orders and long-term orders; Performing secondary classification processing on the preliminary classification results according to the drug information in the order data structure to obtain secondary classification results divided by therapeutic areas; The secondary classification results are processed by time window allocation to obtain a classified order set with a processing time window, and the classified order set with the processing time window is integrated to obtain a classified order set.
3. The order processing method for clinical trial projects according to claim 1, characterized in that: The step of performing priority evaluation on each clinical trial project order according to the classified order set to obtain a dynamic order list with priority scores includes: Performing layer processing on the classified order set, and performing cluster analysis processing on each layer of orders in the order set to obtain a domain correlation clustering result; Based on the domain correlation clustering results, the resource demand of each cluster is quantified to obtain the cluster resource demand index, and the resource pressure assessment is performed according to the cluster resource demand index and the existing resource status to obtain the resource pressure coefficient of each cluster; Perform fuzzy comprehensive evaluation on the project value, sponsor importance, and technical difficulty of each clinical trial project order to obtain an order comprehensive value index, and perform dynamic weight allocation on each clinical trial project order based on the resource pressure coefficient and the order comprehensive value index to obtain a corresponding initial priority score; According to the level and initial priority score of the clinical trial project order, an inter-level priority balancing process is performed to obtain a priority score after inter-level balancing, and a time sensitivity adjustment process is performed on the priority score after inter-level balancing to obtain a priority score that takes time factors into consideration; Based on the priority scores taking the time factor into consideration, a multi-level feedback queue is constructed to obtain a dynamic order list with priority scores.
4. The order processing method for clinical trial projects according to claim 1, characterized in that: Based on the dynamic order list with priority scores, the existing resources are evaluated and matched to obtain multiple optional resource allocation solutions including: Perform multi-dimensional feature extraction on existing resources to obtain a resource feature matrix, and perform cluster analysis on existing resources based on the resource feature matrix to obtain resource capability clustering results; Based on the resource capability clustering result and the dynamic order list with priority scores, preliminary resource allocation processing is performed to obtain an initial resource allocation plan; Performing load balancing analysis on the initial resource allocation plan to obtain a resource utilization index, and dynamically adjusting the initial resource allocation plan according to the resource utilization index to obtain a balanced resource allocation plan; Based on the balanced resource allocation scheme, performing resource conflict detection processing to obtain a resource conflict list, and performing iterative optimization processing on resource allocation according to the resource conflict list to obtain multiple candidate resource configuration schemes; The multiple candidate resource allocation schemes are subjected to multi-objective evaluation processing to obtain scheme evaluation indicators, and based on the scheme evaluation indicators, the candidate schemes are subjected to Pareto optimal screening processing to obtain multiple optional resource allocation schemes.
5. The order processing method for clinical trial projects according to claim 4, characterized in that: The multi-dimensional feature extraction process is performed on the existing resources to obtain a resource feature matrix, and the cluster analysis process is performed on the existing resources according to the resource feature matrix to obtain the resource capability clustering result, which includes: Performing structural processing on the basic information of existing resources to obtain an initial resource data set, and classifying and coding the existing resources based on the initial resource data set to obtain a resource classification coding table; Based on the resource classification coding table, the historical usage data of existing resources is analyzed and processed in time series to obtain resource dynamic characteristic indicators, and the professional capability information of existing resources is analyzed and processed semantically to obtain resource professional capability vectors; According to the resource dynamic feature index and the resource professional capability vector, feature fusion processing is performed to obtain a multi-dimensional resource feature vector, and dimension reduction processing is performed on the multi-dimensional resource feature vector to obtain a resource feature matrix after dimension reduction; Based on the resource feature matrix after dimension reduction, a hierarchical clustering algorithm is used to perform resource clustering processing to obtain a preliminary clustering result, and a silhouette coefficient analysis is performed on the preliminary clustering result to obtain an optimal number of clusters; According to the optimal number of clusters, the preliminary clustering results are refined and clustered to obtain resource capability clustering results.
6. The order processing method for clinical trial projects according to claim 1, characterized in that: According to the multiple optional resource configuration schemes, a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm is used to perform intelligent scheduling and decision support processing, and the optimized scheduling plan includes: Performing natural number encoding processing on the multiple optional resource configuration schemes to obtain an initial population, and performing fitness calculation processing on individuals in the initial population to obtain individual fitness values; Based on the individual fitness values, the initial population is subjected to a tournament selection operation to obtain a parent individual, and the parent individual is subjected to an enhanced best path random insertion crossover operation to obtain a child individual; According to the fitness value of the offspring individual, an adaptive crossover probability calculation process is performed to obtain a dynamic crossover probability, and based on the dynamic crossover probability, a crossover operation process is performed on the offspring individual to obtain an offspring individual after the crossover; Performing adaptive large neighborhood search operator initialization processing on the offspring individuals after the crossover to obtain an initial removal operator set, an insertion operator set and their corresponding initial weights; Iteratively controlling the optimization process according to the initial removal operator set, the insertion operator set and their corresponding initial weights, repeatedly performing operator selection, local search, and weight update to obtain a final optimized population; The best individual in the final optimized population is decoded to obtain an optimized scheduling plan.
7. The order processing method for clinical trial projects according to claim 6, characterized in that: The optimization process is iteratively controlled according to the initial removal operator set, the insertion operator set and their corresponding initial weights, and the operator selection, local search, and weight update are repeatedly performed to obtain the final optimized population, including: According to the initial removal operator set, the insertion operator set and their corresponding initial weights, a roulette wheel method is used to perform operator selection processing to obtain a removal operator and an insertion operator for the current iteration; Based on the selected removal operator and insertion operator, a local search process is performed on the offspring individuals to obtain locally optimized individuals; Performing search performance evaluation on the locally optimized individuals to obtain an effect score of the operator; Performing an operator weight update process according to the effect score of the operator usage to obtain an updated removal operator weight and an insertion operator weight; Based on the fitness value of the locally optimized individual, an adaptive mutation probability calculation process is performed to obtain a dynamic mutation probability, and according to the dynamic mutation probability, a mutation operation process is performed on the locally optimized individual to obtain a mutated individual; Performing population updating processing on the mutated individuals to obtain a new generation of population; Based on the new generation population, the updated removal operator weights and the insertion operator weights, the optimization process is iteratively controlled, and operator selection, local search, and weight update are repeatedly performed to obtain a final optimized population.
8. An order processing device for a clinical trial project, characterized in that: The order processing device for the clinical trial project includes: The order classification module is used to collect and classify the received clinical trial project orders to obtain a classified order set; A priority evaluation module, used to perform a priority evaluation process on each clinical trial project order according to the classified order set, and obtain a dynamic order list with priority scores; A resource evaluation module, for evaluating and matching existing resources based on the dynamic order list with priority scores to obtain multiple optional resource allocation solutions; A scheduling module, for performing intelligent scheduling and decision support processing by using a hybrid algorithm of an adaptive genetic algorithm and an adaptive large neighborhood search algorithm according to the multiple optional resource configuration schemes to obtain an optimized scheduling plan; A processing module is used to process the clinical trial project order according to the optimized scheduling plan.
9. An order processing device for a clinical trial project, characterized in that: The order processing device for the clinical trial project comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the order processing device for the clinical trial project to execute the steps of the order processing method for the clinical trial project as described in any one of claims 1-7.
10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the steps of the order processing method for the clinical trial project as described in any one of claims 1-7 are implemented.
