Self-adaptive cooperative processing method and system for medical instrument test data

Abstract

The invention discloses a self-adaptive cooperative processing method and system for medical instrument test data, and relates to the technical field of electric digital data processing. The self-adaptive cooperative processing method for the test data of the medical apparatus comprises the following steps of performing classification-scheduling cooperative delay evaluation; carrying out standardization-weighted fusion collaborative delay evaluation; and carrying out collaborative decision matching delay evaluation. According to the method, delay evaluation is carried out on the classification-scheduling cooperation process of the medical instrument test data to judge whether cache update frequency optimization is carried out or not, then delay evaluation and cooperation feedback are carried out on the standardization-weighted fusion cooperation process, and finally delay evaluation is carried out on the cooperation decision matching process. Delay dynamic perception, accurate regulation and control and closed-loop optimization of the whole data collaboration process in medical instrument multi-device combined monitoring are achieved, and then the problem that in the prior art, due to data collaboration asynchronization in the medical care monitoring multi-device combined monitoring process, the collaborative decision timeliness is not high is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to an adaptive collaborative processing method and system for medical device test data. Background Technology

[0002] In the fields of medical device testing and the research and development of electronic products and peripheral accessories, the development of electronic products and peripheral accessories provides basic hardware support for medical devices. High-precision sensors, signal processing chips, and other electronic products are key components for medical devices to acquire accurate test data. For example, the performance of the sensors in an electrocardiogram (ECG) monitor directly affects the quality of ECG data acquisition. Peripheral accessories, such as data transmission cables and interfaces, also ensure data interaction between medical devices and external equipment.

[0003] In medical device testing scenarios, test data often exhibits characteristics such as multi-source heterogeneity (e.g., sensor data, image data, and text reports) and dynamic changes (due to differences in test scenarios and equipment models). An adaptive collaborative framework, combined with technologies such as data fusion, intelligent analysis, and dynamic scheduling, enables efficient processing of medical device test data. In large-scale distributed medical monitoring systems, multimodal data (e.g., electrocardiograms, electroencephalograms, and blood oxygen saturation) generated by medical monitoring equipment (e.g., ECG monitors, ventilators, and blood gas analyzers) in different areas are stored on multiple nodes. To efficiently process this data, task scheduling is first required to allocate data processing tasks to appropriate computing nodes.

[0004] Existing technologies design universal data access interfaces based on the interface protocols of different testing equipment (such as HL7, DICOM, and MODBUS), automatically identify data formats through protocol conversion middleware (such as an Apache Camel-based adapter component), and achieve standardized access to multi-source data. They also employ an adaptive cleaning rule engine, combined with domain knowledge (such as the ISO standard for medical device testing). The system utilizes 13485 and machine learning algorithms (such as Isolation Forest and K-means clustering) to automatically identify outliers (such as jump data caused by sensor malfunctions) and missing values. It dynamically selects imputation strategies (such as mean imputation and LSTM prediction imputation) based on data distribution characteristics. For different types of data (structured data, images, and text), it employs multimodal feature extraction models to extract features and deep learning fusion methods, such as attention-based multimodal fusion models, to weightedly fuse multi-source features, highlighting key information (such as heart rate variability, blood pressure fluctuations, and blood oxygen trends). It integrates multimodal data from different devices (such as monitors, ventilators, and infusion pumps) for comprehensive analysis and early warning. For example, it fuses and analyzes electrocardiogram (ECG) data, blood oxygen saturation data, and respiratory rate data. When abnormal ECG waveforms, decreased blood oxygen saturation, and increased respiratory rate are detected, it comprehensively judges that the patient may have respiratory failure and issues timely warning signals. It also establishes a collaborative communication mechanism between multiple devices to achieve data sharing and interaction. Through collaborative work between devices, it provides a more comprehensive understanding of the patient's condition and improves the accuracy of disease warnings.

[0005] For example, the medical device data classification method and system based on LLM disclosed in Chinese Invention Patent No. CN117828087B includes: converting standard medical device data into training data based on the training data format; performing semantic and linguistic learning processing on the medical device training data based on the LLM model; using new device terminology data to verify and evaluate the medical device classification model to obtain the optimal medical device classification model; and classifying the medical device data to be classified based on the optimal medical device classification model.

[0006] The above-mentioned technology has at least the following technical problems: In existing technologies, during the data transmission process of multi-device joint monitoring in medical care, nodes are crucial links in the transmission path of medical device test data, and the processor performance of a node determines its data processing speed. If the node's processor performance is inadequate, such as a low clock frequency of the processor's core computing unit, or when it simultaneously undertakes multiple tasks such as data classification and data fusion calculations, the operating system executes tasks according to priority, and tasks share processor resources. When the number of tasks exceeds the processor's parallel processing capacity, or when high-priority tasks (such as ventilator parameter adjustments) frequently switch, low-priority tasks are preempted, and data can only queue for processing. Moreover, instruction execution depends on the clock cycle. When processor performance is insufficient, the clock cycle becomes longer, the instruction completion time increases, the number of instructions executed per second decreases, and processing the same computational task takes longer, making it impossible to process data in a timely manner. Ultimately, this causes data processing delays, affects the synchronization of data collaboration among multiple devices, and reduces the timeliness of collaborative decision-making. Therefore, there is a problem of low timeliness in collaborative decision-making due to asynchrony in data collaboration during multi-device joint monitoring in medical care. Summary of the Invention

[0007] To address the issue of low timeliness in collaborative decision-making caused by asynchronous data collaboration during multi-device joint monitoring in medical and nursing care in existing technologies, this invention provides an adaptive collaborative processing method and system for medical device test data. The technical solution is as follows: On the one hand, an adaptive collaborative processing method for medical device test data is provided, which includes the following steps: Step 1, receiving medical device test data collected from multiple terminal medical monitoring devices in a medical care monitoring scenario, performing preliminary classification using edge collaborative nodes, and uploading it to the central server. Simultaneously, a latency assessment is performed on the classification-scheduling collaborative process of the medical device test data to determine if there is a need for cache update frequency optimization. Cache update frequency optimization is used to reduce the latency interference of cache update frequency on the classification-scheduling collaborative process of medical device test data, thereby achieving load balancing between edge nodes and collaborative processing capabilities for medical device test data; Step 2, if it is determined that the classification-scheduling collaborative process is incomplete, the data will be processed accordingly. If the degree of collaboration process is qualified, the multimodal data gathered at the central server node will be fused with the help of the central server node. At the same time, the delay evaluation of the standardization-weighted fusion collaboration process in the fusion process will be carried out, and the standardization-weighted fusion collaboration feedback will be carried out to achieve computing power collaboration and accuracy collaboration between the central server node and the edge collaboration node in multimodal data fusion processing. Step 3: Based on the results of the standardization-weighted fusion collaboration feedback, the collaboration decision matching of the data collaboration process will be carried out. At the same time, the delay evaluation of the collaboration decision matching process will be carried out based on the number of node collaboration interactions obtained, so as to achieve global collaboration and closed-loop collaboration between the edge collaboration node and the central server in the entire chain of medical device test data processing.

[0008] On the other hand, an adaptive collaborative processing system for medical device test data is provided. This system includes: a collaborative allocation delay assessment module, a collaborative fusion delay assessment module, and a collaborative decision matching delay assessment module. The collaborative allocation delay assessment module receives medical device test data collected from multiple terminal medical monitoring devices in a medical care monitoring scenario. After preliminary classification using edge collaborative nodes, it uploads the data to a central server. Simultaneously, it performs delay assessment on the classification-scheduling collaborative process of the medical device test data through an edge collaboration mechanism to determine if there is a need to optimize the cache update frequency. The collaborative fusion delay assessment module, if the classification-scheduling collaborative process is deemed qualified, uses the central server nodes to fuse the multimodal data aggregated to the central server. It also performs delay assessment on the standardization-weighted fusion collaborative process during the fusion process and provides standardization-weighted fusion collaborative feedback. The collaborative decision matching delay assessment module performs collaborative decision matching for the data collaboration process based on the results of the standardization-weighted fusion collaborative feedback, and simultaneously performs delay assessment on the collaborative decision matching process based on the acquired number of node collaborative interactions.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. By conducting latency assessments on the medical device test data classification-scheduling collaboration process, it is determined whether to optimize the cache update frequency. This reduces the probability of blocking caused by cache updates, allowing data to flow smoothly in the classification and scheduling stages. This ensures timely acquisition of the latest classification information and rapid decision-making, improving the response speed to new data, reducing waiting time, and rationally controlling the use of computing resources. Then, latency assessments and collaborative feedback are conducted on the standardization-weighted fusion collaboration process. This shortens the execution time of each processor instruction, increases the amount of data and instructions processed per unit time, reduces data backlog and congestion, and improves system response speed. Finally, based on the feedback results, collaborative decision matching is performed, and latency assessments are conducted on it to accelerate the transmission of decision information, making the interaction between nodes more compact and efficient, shortening the overall collaborative decision matching time, accelerating the decision matching speed, and thus improving the timeliness of collaborative decisions.

[0010] 2. By decomposing two core time-consuming data types—feature mapping computation and weight adaptation calculation—a collaborative fusion latency index is constructed to achieve accurate quantitative evaluation of the multimodal data fusion process. When latency exceeds limits, hardware performance is optimized in two dimensions by adjusting the instruction cycle count and processor clock frequency through the collaborative fusion latency index deviation mapping parameter: reducing instruction cycles to shorten single instruction latency and increasing clock frequency to reduce data waiting time, directly breaking through the fusion efficiency bottleneck at the underlying computing power level. The optimized closed-loop verification mechanism ensures that the fusion process is always in a highly efficient state. If it passes, it provides high-quality data support for subsequent collaborative decisions; if it fails, it triggers an early warning for continuous optimization. This mechanism effectively solves the problem of collaborative failure caused by processing latency in multimodal medical data fusion, improving the timeliness and accuracy of data fusion.

[0011] 3. By quantifying decision latency through the core indicator of node collaborative interaction count, the quality of collaborative matching can be accurately judged, avoiding the one-sidedness of evaluation by a single indicator. During the optimization phase, the mapping relationship between deviation scores and the adjustment value of the number of parallel nodes is used to specifically increase the number of parallel decision nodes, directly improving the parallel processing capability of tasks, reducing the communication and coordination time between nodes, and fundamentally improving decision efficiency and system response speed. Simultaneously, the optimized secondary verification mechanism forms a closed-loop control; if qualified, it ensures the timeliness of the entire collaborative processing process; if unqualified, it triggers an early warning to initiate deep optimization. This effectively solves the decision latency problem caused by interaction redundancy in multi-node collaboration, ensuring the efficient processing of medical device test data in complex collaborative scenarios and providing real-time and reliable decision support for medical monitoring. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating an adaptive collaborative processing method for medical device test data provided in an embodiment of the present invention; Figure 2 A logical framework diagram of an adaptive collaborative processing method for medical device test data provided in an embodiment of the present invention; Figure 3 A flowchart for classification-scheduling collaborative delay assessment and optimization provided in an embodiment of the present invention; Figure 4 A flowchart for standardized-weighted fusion collaborative delay evaluation and optimization provided in an embodiment of the present invention; Figure 5A flowchart for collaborative decision-making matching delay evaluation and optimization provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an adaptive collaborative processing system for medical device test data provided in an embodiment of the present invention; Figure 7 This is the main interface of an adaptive collaborative processing system for medical device test data provided in an embodiment of the present invention; Figure 8 This invention provides a classification-scheduling collaborative interface in an adaptive collaborative processing system for medical device test data, as provided in an embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0015] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0016] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0017] like Figure 1 The flowchart shown is an adaptive collaborative processing method for medical device test data. The processing flow of this method may include the following steps: Step 1: Receive medical device test data collected from multiple terminal medical monitoring devices in a medical care monitoring scenario. After preliminary classification using edge collaboration nodes (deployed locally at ward servers, edge gateways, etc.), upload the data to the central server. Simultaneously, use the edge collaboration mechanism to evaluate the latency of the classification-scheduling collaboration process of the medical device test data to determine if there is a need for cache update frequency optimization. Cache update frequency optimization is used to reduce the degree of latency interference of cache update frequency on the classification-scheduling collaboration process of medical device test data, so as to achieve load balancing between edge nodes and collaborative capabilities between medical device test data processing.

[0018] Step 2: If the classification-scheduling coordination process is deemed qualified, the multimodal data aggregated to the central server is fused using the central server node. At the same time, the delay of the standardization-weighted fusion coordination process is evaluated, and standardization-weighted fusion coordination feedback is provided to achieve computing power coordination and accuracy coordination between the central server node and the edge coordination node in multimodal data fusion processing.

[0019] Step 3: Based on the results of the standardized-weighted fusion collaborative feedback, perform collaborative decision matching for the data collaboration process (including the classification-scheduling collaboration process and the standardized-weighted fusion collaboration process). At the same time, perform a delay evaluation on the collaborative decision matching process based on the number of node collaborative interactions obtained, so as to achieve global and closed-loop collaboration between edge collaborative nodes and the central server in the entire chain of medical device test data processing.

[0020] like Figure 2 The diagram illustrates the logical framework of an adaptive collaborative processing method for medical device test data. First, relying on edge collaboration nodes, it receives and initially classifies medical device test data from multiple terminals. The edge collaboration mechanism then performs a latency assessment on the classification-scheduling collaborative process of the medical device monitoring data to determine if the cache update frequency needs optimization. This ensures efficient and adaptable data allocation, preventing delays from impacting subsequent processes. If the classification-scheduling collaborative assessment is successful, a central server aggregates multimodal data for a standardization-weighted fusion collaborative process. This process also undergoes latency assessment and collaborative feedback on standardization and weighted fusion, providing a high-quality data foundation for subsequent decisions. Based on the fusion collaborative feedback results, the data collaboration process of classification-scheduling and standardization-weighted fusion is advanced for collaborative decision matching. The latency of this process is assessed to determine if interactive response efficiency needs optimization. This ensures that the entire process, from data allocation and fusion to decision matching, operates under controllable latency and high-efficiency collaboration, improving the overall efficiency of medical device monitoring data processing and application.

[0021] In this embodiment, the core lies in the deep linkage of a three-level collaborative mechanism, which breaks through the limitations of existing technologies that only optimize a single processing link, and effectively solves the problem of insufficient decision-making timeliness caused by asynchronous data collaboration in the joint monitoring of multiple medical monitoring devices.

[0022] Specifically, this method breaks down data processing into three progressive stages: classification-scheduling collaboration, standardization-weighted fusion collaboration, and collaborative decision matching. Each stage is optimized independently yet supports each other: the classification-scheduling stage reduces data transmission latency at edge nodes from the source through time synchronization determination and queue length optimization, laying the foundation for subsequent collaboration; the standardization-weighted fusion stage achieves synergy in computing power and accuracy between central and edge nodes through fusion latency assessment and processor performance optimization, ensuring the efficiency of multimodal data fusion; and the collaborative decision matching stage strengthens global collaboration across the entire link through interaction number control and node parallel optimization.

[0023] The three-tiered process forms a complete link of "problem prediction - real-time optimization - closed-loop verification". This not only reduces the delay interference in each link, but also improves the response speed of multi-device data from collection to decision-making through the collaborative linkage between links. This ensures the real-time and accuracy of key test data in medical monitoring, provides efficient data support for clinical monitoring decisions, and enhances the system's adaptability and stability to complex medical scenarios.

[0024] like Figure 3 The flowchart shown is for classification-scheduling collaborative delay assessment and optimization. The design logic of the flowchart is as follows: the process starts with the initial classification of edge collaborative nodes, followed by scheduling collaborative delay assessment. If the assessment is unqualified, the cache update frequency is adjusted. After optimization, classification and scheduling collaborative delay assessment are performed again. If the assessment is still unqualified, classification-scheduling collaborative early warning is triggered. Otherwise, multimodal data fusion is performed.

[0025] Specifically, the edge collaboration mechanism is used to evaluate the delay of the classification-scheduling collaboration process for medical device test data. This includes: determining the time synchronization of classification-scheduling collaboration and evaluating the delay of the classification-scheduling collaboration process. The specific process for determining the time synchronization of classification-scheduling collaboration is as follows: If the average difference in the acquired data synchronization timestamps is not greater than a preset average difference in the data synchronization timestamps, it indicates that the time synchronization of the medical device test data meets the requirements of classification-scheduling collaboration, and a delay evaluation of the classification-scheduling collaboration process is performed. The average difference in the data synchronization timestamps represents the average difference between the timestamp recorded by the edge collaboration node when receiving the data and the timestamp generated by the terminal medical monitoring device when collecting the data. The data synchronization timestamps are obtained through a high-precision crystal oscillator combined with a time counter. The preset average difference in the data synchronization timestamps is represented by the sum of the average differences in the historical data synchronization timestamps in the historical classification-scheduling collaboration processes in the database. If the acquired average difference in the data synchronization timestamps is greater than the preset average difference in the data synchronization timestamps, the task queue length of the edge node is optimized. This optimization reduces the degree of delay interference of the edge node task queue length on the classification-scheduling collaboration process of medical device test data. Specifically, the optimization of the edge node task queue length involves the following steps: The average synchronization timestamp difference of the acquired data is input into the database. A mapping relationship is established between this value and the corresponding reduction in queue length for edge node tasks in the database to obtain the actual reduction in queue length. The original task queue length of the edge node is used as a baseline, and this reduction is subtracted to calculate the optimized final queue length. For example, when the average synchronization timestamp difference is 50ms, the corresponding actual queue length reduction is 8 items. If the current original task queue length of the edge node is 20 items, the optimized final queue length is 20 items minus 8 items, i.e., 12 items. By reducing the number of tasks processed in batches, the queue length of the next edge node task is shortened, reducing the difference between the terminal acquisition time and the edge reception time. After the queue length is reduced, if the monitored average synchronization timestamp difference is not greater than the preset average synchronization timestamp difference, the edge node task queue length optimization is completed, and a delay assessment of the classification-scheduling coordination process is performed; otherwise, a data time alignment calibration warning is issued.

[0026] like Figure 4 The flowchart shown is for standardized-weighted fusion collaborative delay assessment and optimization. The design logic of the flowchart is as follows: The process starts when the central server completes multimodal data fusion. First, a standardized-weighted fusion delay assessment is performed. If the assessment fails, the performance of the node processor needs to be optimized. After optimization, the assessment is performed again. If the assessment still fails, a standardized-weighted fusion collaborative early warning is triggered. Otherwise, the collaborative decision matching stage is entered.

[0027] The specific process for delay evaluation in the classification-scheduling coordination process is as follows: the acquired coordination allocation delay data is compared with the preset coordination allocation delay data in the database to obtain the difference comparison results. At the same time, the comparison results are summed and averaged to obtain the coordination allocation delay index. The coordination allocation delay data includes the average time for modal feature extraction and the average time for node allocation calculation. Both the modal feature extraction time and the node allocation calculation time are monitored by the high-precision timer built into the edge computing node. The average time for modal feature extraction represents the average time required for the medical device test data sample to complete feature extraction from the start of processing when the original data is processed (such as time domain features, frequency domain features, morphological features, etc.). The average time for node allocation calculation represents the average time required to dynamically allocate the feature-extracted medical device test data to different edge computing nodes.

[0028] The results of the difference comparison include the ratio of the average time spent on modal feature extraction to the preset average time spent on modal feature extraction, the ratio of the average time spent on node allocation calculation to the preset average time spent on node allocation calculation, and the preset collaborative allocation delay data, which includes the preset average time spent on modal feature extraction and the preset average time spent on node allocation calculation. The preset average time spent on modal feature extraction is represented by the sum of the average times spent on modal feature extraction in the historical classification-scheduling collaborative process in the database, and the preset average time spent on node allocation calculation is represented by the sum of the average times spent on node allocation calculation in the historical classification-scheduling collaborative process in the database. The collaborative allocation delay index is used to quantify the degree of influence of collaborative allocation delay data on the classification-scheduling collaborative process of medical device test data.

[0029] In this embodiment, when optimizing the edge node task queue length, the average difference in data synchronization timestamps is used as input to the edge queue management optimization algorithm. Based on a pre-defined mapping relationship between this difference and the corresponding reduction in the edge node task queue length (this mapping relationship is constructed based on existing algorithms, which comprehensively consider system load, data processing capabilities, and other factors, and derive the input-output correspondence through extensive experiments and data analysis), the actual reduction in queue length is calculated. Subsequently, the number of batch processing tasks is reduced to shorten the next edge node task queue length, thereby reducing the difference between terminal acquisition and edge reception time and improving time synchronization.

[0030] It's important to understand that the collaborative allocation latency index increases with both the average time spent on modal feature extraction and the average time spent on node allocation computation. Modal feature extraction, as a front-end step in data processing, suffers from delays in data preparation if its average time is too long. This directly compresses the time window for subsequent node allocation computation, forcing the computation to proceed hastily. This may increase the risk of computational errors due to insufficient time, indirectly lengthening the overall latency. Conversely, an increase in the average time spent on node allocation computation not only consumes more time itself but also causes subsequent processes to stall while waiting for it to complete, further accumulating latency. These two factors, inversely related and mutually reinforcing, jointly drive up the collaborative allocation latency index.

[0031] This example helps to accurately pinpoint the root cause of latency by considering the interaction between the average time spent on modal feature extraction and the average time spent on node allocation computation. Optimizing the feature extraction algorithm reduces data complexity and decreases the average time spent on modal feature extraction; rationally planning node resources and simplifying the allocation algorithm shortens the average time spent on node allocation computation. This effectively reduces the collaborative allocation latency exponent, improves the efficiency of medical device test data classification and scheduling collaboration, ensures timely data processing, provides faster and more accurate support for medical decision-making, and avoids further increases in latency due to resource conflicts, thereby improving the overall timeliness of the collaborative process.

[0032] Furthermore, to determine whether there is a need for cache update frequency optimization, the specific process is as follows: if the obtained collaborative allocation delay index is not greater than the preset collaborative allocation delay index in the database, it indicates that the classification-scheduling collaborative process is qualified, and the delay evaluation of the standardization-weighted fusion collaborative process is performed; otherwise, it indicates that the classification-scheduling collaborative process is unqualified, and the cache update frequency is optimized. The preset collaborative allocation delay index is represented by the result of summing and averaging the historical collaborative allocation delay indices in the historical classification-scheduling collaborative processes in the database.

[0033] The cache update frequency optimization involves the following steps: The result of summing the obtained collaborative allocation delay index deviation and the cache update frequency deviation is input into the database. This result is then matched with the corresponding cache update frequency adjustment value in the database to obtain the actual cache update frequency adjustment value. Based on this actual cache update frequency adjustment value, the PID control algorithm outputs a specific adjustment command for the cache update frequency, dynamically correcting the system's cache update mechanism to reduce edge node congestion caused by excessive frequency and improve the matching degree between scheduling tasks and classification outputs. The collaborative allocation delay index deviation represents the difference between the obtained collaborative allocation delay index and the preset collaborative allocation delay index in the database. The cache update frequency deviation represents the difference between the obtained cache update frequency and the preset cache update frequency in the database. The preset cache update frequency is represented by the average of the historical cache update frequencies obtained in the historical classification-scheduling collaborative process in the database. After cache update frequency optimization, if the newly obtained collaborative allocation delay index is not greater than the preset collaborative allocation delay index in the database, the cache update frequency optimization is completed, and a delay assessment of the standardization-weighted fusion collaborative process is performed; otherwise, a classification-scheduling collaborative early warning is issued.

[0034] It's important to understand that the result of summing the obtained collaborative allocation latency index deviation and cache update frequency deviation is input into the database because the collaborative allocation latency index deviation reflects task scheduling efficiency, while the cache update frequency deviation reflects data timeliness management. These two are coupled: excessively frequent cache updates increase latency, and excessively high latency may necessitate adjusting the update frequency. Summing comprehensively quantifies the system deviation resulting from the combined effect of both. When matched with historical mapping relationships in the database, the actual adjustment value can be obtained more accurately, balancing scheduling efficiency and data validity. This avoids unintended consequences from adjusting a single parameter, improving the overall comprehensiveness and accuracy of the optimization.

[0035] Specifically, the specific adjustment instructions for the cache update frequency are as follows: Since the obtained collaborative allocation latency index is greater than the preset collaborative allocation latency index in the database, it indicates that the current cache update frequency is too high. At this time, the actual cache update frequency adjustment value is the amount by which the cache update frequency needs to be reduced. The specific adjustment instructions output by the PID (Proportional-Integral-Derivative) control algorithm are: extend the cache update cycle and increase the threshold for triggering updates. For example, increase the update cycle from the original duration by 20%. By reducing the number of updates per unit time, the cache operation load of edge nodes is reduced, the computing power congestion is alleviated, and the basic matching between cache data and scheduling tasks is ensured, thereby reducing collaborative allocation latency and optimizing system performance.

[0036] In this embodiment, by comparing the collaborative allocation delay index with a preset value, the system can accurately identify whether the classification-scheduling collaborative process is qualified, avoiding a continuous increase in collaborative delay due to unreasonable cache updates, and ensuring the stability and efficiency of the collaborative process. Utilizing a PID control algorithm, combined with the collaborative allocation delay index deviation and cache update frequency deviation, the system dynamically outputs a reduction in the cache update frequency. This dynamic adjustment method can flexibly change the cache update frequency according to actual conditions, effectively reducing edge node computing power congestion caused by excessively high frequencies and improving system resource utilization.

[0037] Optimizing the cache update frequency improves the matching degree between scheduled tasks and classification outputs, making data flow more smoothly in the classification-scheduling collaboration. If the re-acquired collaborative allocation latency index is qualified, it can smoothly enter the latency evaluation of the standardization-weighted fusion collaboration process, providing strong support for the optimization of the entire medical device test data processing workflow and improving overall processing efficiency and accuracy.

[0038] like Figure 5 The flowchart shown is for collaborative decision matching delay evaluation and optimization. The design logic of the flowchart is as follows: the process starts with collaborative decision matching, and then enters the collaborative interaction number delay evaluation stage to determine whether the interaction efficiency meets the standard. If the evaluation result is not qualified, the interaction response efficiency needs to be optimized. After the optimization is completed, the collaborative interaction number delay evaluation is performed again. If the evaluation is still not qualified, a collaborative decision matching warning is triggered; otherwise, the process ends.

[0039] Specifically, a delay assessment is performed on the standardized-weighted fusion collaboration process during the fusion process. The specific procedure is as follows: the acquired collaborative fusion delay data is compared with the preset collaborative fusion delay data in the database one by one to obtain the difference comparison results. At the same time, the difference comparison results are summed and the average value is calculated to obtain the collaborative fusion delay index. The collaborative fusion delay data includes the feature mapping operation time and the weight adaptation calculation time, both of which are monitored by a high-precision timer. The feature mapping operation time represents the time consumed in the process of transforming the standardized multimodal data into data of the same dimension during the fusion of medical device test data. The weight adaptation calculation time represents the time consumed by the central server in the process of multimodal data fusion to transform different modal features. The computation time consumed by weighting is included in the comparison results of various differences, such as the ratio of the acquired feature mapping computation time to the preset feature mapping computation time, the ratio of the acquired weight adaptation computation time to the preset weight adaptation computation time, and the preset collaborative fusion delay data, which includes the preset feature mapping computation time and the preset weight adaptation computation time. The preset feature mapping computation time is represented by the sum and average of the historical feature mapping computation times in the historical standardization-weighted fusion collaborative process in the database, and the preset weight adaptation computation time is represented by the sum and average of the historical weight adaptation computation times in the historical standardization-weighted fusion collaborative process in the database. The collaborative fusion delay index is used to quantify the degree of influence of the collaborative fusion delay data on the standardization-weighted fusion collaborative process.

[0040] In this embodiment, it is important to understand that the collaborative fusion latency index increases with the increase of feature mapping computation time and weight adaptation computation time. Feature mapping is a prerequisite for weight adaptation. Only after the feature extraction of the input data is completed can weight adaptation be calculated based on these features. The increase in feature mapping computation time will directly accumulate to the overall process and constrain the start time of weight adaptation. Furthermore, in resource-constrained scenarios such as edge nodes, feature mapping and weight adaptation usually share computing resources. In this case, the time consumption of the two will be mutually constrained due to resource allocation. If feature mapping occupies a large amount of computing resources, the available resources for weight adaptation will be compressed, which will lead to an increase in the weight adaptation computation time.

[0041] This example leverages the interdependence between standardization and weighted fusion (standardization is a prerequisite for weighted fusion) to reduce unnecessary waiting or redundant calculations through collaborative optimization, thereby minimizing resource waste. By dynamically adjusting the resource preparation for weighted fusion based on the real-time progress of standardization (e.g., pre-allocating memory), idle resources during the waiting period for standardization results can be avoided. Dynamic resource allocation balances the computational load of both processes. For instance, if data standardization takes longer (e.g., format conversion for high-resolution images), the corresponding weighted fusion resource allocation can be temporarily reduced (e.g., decreasing the number of iterations for weight calculation), prioritizing the fusion efficiency of other data that completes standardization quickly and minimizing the overall process efficiency reduction.

[0042] Furthermore, the standardized-weighted fusion collaborative feedback process is as follows: the obtained collaborative fusion delay index is compared with the preset collaborative fusion delay index in the database. If the obtained collaborative fusion delay index is not greater than the preset collaborative fusion delay index, it indicates that the standardized-weighted fusion collaborative process is qualified, and a delay assessment of the collaborative decision matching process is performed. If the obtained collaborative fusion delay index is greater than the preset collaborative fusion delay index, it indicates that the standardized-weighted fusion collaborative process is unqualified, and node processor performance optimization is performed. The preset collaborative fusion delay index is represented by the sum and average of historical collaborative fusion delay indices in historical standardized-weighted fusion collaborative processes in the database.

[0043] The node processor performance optimization involves the following steps: Based on the mapping relationship between the acquired collaborative fusion latency index deviation and the corresponding instruction cycle number adjustment value and node processor clock frequency adjustment value in the database, the decrease value of the instruction cycle number and the increase value of the node processor clock frequency are obtained. The current instruction cycle number and node processor clock frequency are obtained. The decrease value of the instruction cycle number is subtracted from the current instruction cycle number to obtain the actual instruction cycle number. The increase value of the node processor clock frequency is added to the current node processor clock frequency to obtain the actual node processor clock frequency. By reducing the instruction cycle number, the execution time of each instruction is shortened, thereby reducing the time consumed in the entire data processing stage. At the same time, by increasing the node processor clock frequency, the pause and waiting time of medical device test data is reduced. The collaborative fusion latency index deviation represents the difference between the acquired collaborative fusion latency index and the preset collaborative fusion latency index in the database. After the node processor performance optimization, if the re-acquired collaborative fusion latency index is not greater than the preset collaborative fusion latency index in the database, the node processor performance optimization is completed and the latency evaluation of collaborative decision matching is performed; otherwise, a standardized-weighted fusion collaborative early warning is performed.

[0044] In this embodiment, the determination of the instruction cycle reduction value and the node processor clock frequency increase value can be achieved by combining dynamic voltage and frequency scaling technology with historical optimization data. Through the mapping relationship between the collaborative fusion latency index deviation and the adjustment value stored in the database, when the deviation is positive, the Dynamic Voltage and Frequency Scaling (DVFS) algorithm is invoked to generate the initial instruction cycle reduction value (e.g., shortening single-cycle time based on instruction pipeline optimization) and clock frequency increase value (e.g., increasing the crystal oscillator output frequency). Simultaneously, dynamic correction is performed based on the current processor load rate: when the load is too high, the frequency increase is appropriately reduced to avoid overheating; when the load is low, the cycle reduction is increased to accelerate processing. The final output adjustment value must meet the processor hardware safety threshold, and precise control of the instruction cycle and frequency is achieved through the hardware abstraction layer interface.

[0045] In this example of weighted fusion collaboration, standardization is a fundamental step. Quickly completing standardization can reduce data preprocessing time, enabling subsequent weighted fusion to proceed as early as possible. Overall, it shortens the time cycle from data input to fusion result output. Furthermore, the optimized processor can quickly and accurately complete the weighted calculation of large amounts of data, avoiding delays in fusion result output due to computational latency and ensuring real-time acquisition of fusion information.

[0046] Furthermore, a delay assessment of collaborative decision matching is performed based on the acquired number of node collaborative interactions. This includes: collaborative decision matching delay determination and collaborative decision matching delay optimization. Specifically, the collaborative decision matching delay determination process is as follows: if the acquired number of node collaborative interactions is not greater than the preset number of node collaborative interactions in the database, the collaborative decision matching is considered successful, and an adaptive collaborative processing assessment of the medical device test data is completed. This adaptive collaborative processing assessment includes classification-scheduling collaborative delay assessment, standardization-weighted fusion collaborative delay assessment, and collaborative decision matching assessment. If the acquired number of node collaborative interactions is greater than the preset number of node collaborative interactions in the database, the collaborative decision matching is considered unsuccessful, and collaborative decision matching delay optimization is performed. The number of node collaborative interactions represents the number of times different decision nodes need to communicate, share data, and coordinate their work to complete the data collaboration task during collaborative decision matching.

[0047] For example, a node sending a data request to another node and receiving a response, or multiple nodes jointly participating in a collaborative computing step, are all counted as one or more interactions. This number reflects the frequency and closeness of collaboration between nodes. The preset number of node collaborative interactions is represented by the sum and average of historical node collaborative interactions in the historical collaborative decision-making matching process in the database. The number of node collaborative interactions indicates the number of times different decision-making nodes need to communicate, share data, and coordinate their work to complete the data collaboration task during the collaborative decision-making matching process, and is obtained through monitoring by a network performance monitor. For example, a node sending a data request to another node and receiving a response, or multiple nodes jointly participating in a collaborative computing step, are all counted as one or more interactions, and this number reflects the frequency and closeness of collaboration between nodes.

[0048] Specifically, the collaborative decision-making matching delay optimization involves the following steps: The acquired collaborative decision-making matching delay deviation score is input into the database, and a mapping relationship is established between this score and the corresponding node parallel quantity adjustment value in the database. The current node parallel quantity is then added to the node parallel quantity adjustment value to obtain the actual node parallel quantity. By increasing the parallel quantity of decision nodes, the efficiency of decision task processing and the overall system response speed are improved. The collaborative decision-making matching delay deviation score represents the result of adding the node collaborative interaction number deviation score and the node parallel quantity deviation score. The node collaborative interaction number deviation score represents the ratio of the acquired node collaborative interaction number deviation to the preset node collaborative interaction number deviation, and the node parallel quantity deviation score represents the ratio of the acquired node parallel quantity deviation to the preset node parallel quantity deviation. After collaborative decision-making matching delay optimization, if the re-acquired node collaborative interaction number is not greater than the preset node collaborative interaction number in the database, the collaborative decision-making matching is considered successful, and the adaptive collaborative processing evaluation of medical device test data is completed. Otherwise, a collaborative decision-making matching warning is issued.

[0049] In this embodiment, by quantifying the deviation score between the number of node collaborative interactions and the number of parallel operations, a precise mapping between the deviation and the node adjustment value is established. This allows for targeted increases in the number of parallel operations, directly improving the efficiency of decision-making tasks and the system response speed, and effectively shortening the matching latency. Simultaneously, the optimized results are used to determine the collaborative processing effect through a threshold of the number of node interactions, achieving closed-loop verification. Passing the threshold ensures the high efficiency and accuracy of medical device test data processing; failing to pass triggers a timely warning and secondary optimization, ensuring that collaborative decision-making always adapts to the dynamic changes in test data. This improves the stability, adaptability, and data processing quality of the medical device testing system, providing reliable support for equipment performance evaluation.

[0050] like Figure 6The diagram illustrates the structure of an adaptive collaborative processing system for medical device test data. The system includes: a collaborative allocation delay evaluation module, a collaborative fusion delay evaluation module, and a collaborative decision matching delay evaluation module. The collaborative allocation delay evaluation module receives medical device test data collected from multiple terminal medical monitoring devices in a medical care monitoring scenario. After preliminary classification using edge collaborative nodes, the data is uploaded to the central server. Simultaneously, it evaluates the delay of the classification-scheduling collaborative process of the medical device test data through an edge collaboration mechanism to determine if there is a need to optimize the cache update frequency. The collaborative fusion delay evaluation module, if the classification-scheduling collaborative process is deemed satisfactory, uses the central server nodes to fuse the multimodal data aggregated to the central server. It also evaluates the delay of the standardization-weighted fusion collaborative process during the fusion process and provides standardization-weighted fusion collaborative feedback. The collaborative decision matching delay evaluation module performs collaborative decision matching for the data collaboration process based on the results of the standardization-weighted fusion collaborative feedback, and evaluates the delay of the collaborative decision matching process based on the acquired number of node collaborative interactions.

[0051] In this embodiment, a closed-loop processing mechanism of "edge-center-end-link" is constructed through the collaborative linkage of the collaborative allocation delay assessment module, the collaborative fusion delay assessment module, and the collaborative decision matching delay assessment module, realizing end-to-end optimization of medical device test data from collection and classification to fusion decision-making. The collaborative allocation delay assessment module ensures the real-time performance and load balancing of data processing at edge nodes; the collaborative fusion delay assessment module improves the accuracy and computing power adaptability of multimodal data fusion; and the collaborative decision matching delay assessment module maximizes end-to-end efficiency through global collaboration. The combination of these three modules reduces latency interference at each stage and enhances the collaborative capabilities between nodes, meeting the high requirements of real-time performance and reliability for data processing in medical monitoring scenarios.

[0052] It should be added that, such as Figure 7The image shows the main interface of an adaptive collaborative processing system for medical device test data provided in this embodiment of the invention. The upper left corner displays the system name: Medical Device Collaborative Monitoring and Processing System. Navigation options include Home, Device Management, Collaborative Operation Monitoring, Adaptive Processing, Alarm Management, System Settings, and Help and Support, facilitating user switching between different functional modules. Device types and monitoring statuses are displayed, including ventilators, ECG devices, monitors, and blood glucose meters. A "Add More Devices" button is present on the right, prompting users to view more device information. Device linkage parameter optimization suggestions are displayed in a table format, containing four columns: device name, linked devices, latency index, and optimization suggestions. A "Next Page" button is located in the lower right corner, indicating that the data may be displayed across multiple pages. A line graph shows the changing trend of the collaborative allocation latency index: the horizontal axis represents time, and the vertical axis represents the latency rate percentage. The graph shows the changes in the latency rate at different time points, using different colors to distinguish between high efficiency, low efficiency, and warning states, helping users intuitively understand the stability of data transmission.

[0053] like Figure 8 As shown, this is a classification-scheduling collaborative interface in an adaptive collaborative processing system for medical device test data provided by an embodiment of the present invention. The system name of the medical device collaborative monitoring and processing system is displayed in the upper left corner, next to a navigation bar containing options such as Home, Device Management, Collaborative Operation Monitoring, Adaptive Processing, Alarm Management, System Settings, and Help and Support. Users can switch to different functional modules through these options. The top center of the page displays Collaborative Operation Monitoring, clearly indicating the current page. Below the title are three buttons: Return to Home, Refresh Data, and Export Report, facilitating user navigation. The middle of the page displays the data synchronization status, showing the average data synchronization time deviation. The system displays the difference in synchronization status, indicated by green and red indicators. Task queue scaling allows users to adjust system performance using these parameters. Optimization details show task queue scaling adjustments and batch task processing volume. Module feature extraction and node allocation calculation monitoring displays the names of different devices, module feature extraction / node allocation calculation time, thresholds, and status. Cooperative allocation performance displays the cooperative allocation latency index, and an orange "Optimize Latency" button on the right allows users to perform optimization operations. The optimization and alarm area displays the device's optimization status, optimization parameters, and optimization goals in a table format. A "Next Page" button is located at the bottom right corner of the table, allowing users to navigate through different pages.

[0054] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. It will be clearly understood by those skilled in the art that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0055] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An adaptive collaborative processing method for medical device test data, characterized in that, Includes the following steps: Step 1: Receive medical device test data collected from multiple terminal medical monitoring devices in a medical care monitoring scenario. After preliminary classification using edge collaboration nodes, upload the data to the central server. Simultaneously, perform a delay assessment on the classification-scheduling collaboration process of the medical device test data to determine if there is a need for cache update frequency optimization. The cache update frequency optimization is used to reduce the degree of delay interference of the cache update frequency on the classification-scheduling collaboration process of the medical device test data, so as to achieve the collaborative capability between load balancing among edge nodes and medical device test data processing. Step 2: If the classification-scheduling coordination process is deemed qualified, the multimodal data aggregated to the central server is fused using the central server node. At the same time, the delay of the standardization-weighted fusion coordination process is evaluated, and the standardization-weighted fusion coordination feedback is performed to achieve computing power coordination and accuracy coordination between the central server node and the edge coordination node in multimodal data fusion processing. Step 3: Based on the results of the standardized-weighted fusion collaborative feedback, collaborative decision matching is performed in the data collaboration process. At the same time, the delay evaluation of the collaborative decision matching process is performed based on the number of node collaborative interactions obtained, so as to realize global collaboration and closed-loop collaboration between edge collaborative nodes and central server in the entire chain of medical device test data processing.

2. The adaptive collaborative processing method for medical device test data as described in claim 1, characterized in that, The aforementioned delay assessment of the classification-scheduling coordination process for medical device test data specifically includes: determining the time synchronization of classification-scheduling coordination and assessing the delay of the classification-scheduling coordination process; The specific process for determining the time synchronization of the classification-scheduling coordination is as follows: If the average time-of-synchronization timestamp difference of the acquired data is not greater than the preset average time-of-synchronization timestamp difference, it indicates that the time synchronization of the medical device test data meets the requirements of classification-scheduling coordination, and the delay assessment of the classification-scheduling coordination process is carried out. If the average difference in the acquired data synchronization timestamps is greater than the preset average difference in the data synchronization timestamps, then the edge node task queue length is optimized. The edge node task queue length optimization is used to reduce the delay interference of the edge node task queue length on the classification-scheduling coordination process of medical device test data. The optimization of the edge node task queue length involves the following steps: The average synchronization timestamp difference of the acquired data is input into the database. The mapping relationship between the data and the queue length reduction value of the corresponding edge node task in the database is used to obtain the actual queue length reduction value. Based on the original task queue length of the edge node, the acquired queue length reduction value is subtracted to calculate the optimized final queue length. By reducing the number of batch processing tasks, the queue length of the next edge node task is shortened, thereby reducing the difference between the terminal acquisition time and the edge reception time. After the queue length is reduced, if the average data synchronization timestamp difference is not greater than the preset average data synchronization timestamp difference, the edge node task queue length optimization is completed and the delay assessment of the classification-scheduling coordination process is performed; otherwise, a data time alignment calibration warning is issued.

3. The adaptive collaborative processing method for medical device test data as described in claim 2, characterized in that, The delay assessment process for the classification-scheduling coordination process is as follows: The obtained collaborative allocation delay data is compared with the preset collaborative allocation delay data in the database to obtain the difference results. At the same time, the collaborative allocation delay index is obtained by summing and averaging the difference results. The collaborative allocation delay data includes the average time for modal feature extraction and the average time for node allocation computation. The average time for modal feature extraction represents the average time required for medical device test data samples to complete feature extraction from the start of processing. The average time for node allocation computation represents the average time required to dynamically allocate the feature-extracted medical device test data to different edge computing nodes. The collaborative allocation delay index is used to quantify the impact of collaborative allocation delay data on the classification-scheduling collaborative process of medical device test data.

4. The adaptive collaborative processing method for medical device test data as described in claim 3, characterized in that, The specific process for determining whether there is a need to optimize the cache update frequency is as follows: If the obtained collaborative allocation delay index is not greater than the preset collaborative allocation delay index in the database, it indicates that the classification-scheduling collaborative process is qualified, and the delay evaluation of the standardization-weighted fusion collaborative process is performed; otherwise, it indicates that the classification-scheduling collaborative process is unqualified, and the cache update frequency is optimized. The specific steps for optimizing the cache update frequency are as follows: The result of summing the obtained collaborative allocation delay index deviation and cache update frequency deviation is input into the database and matched with the corresponding cache update frequency adjustment value in the database to obtain the actual cache update frequency adjustment value. Based on the input actual cache update frequency adjustment value, the PID control algorithm outputs a specific adjustment instruction for the cache update frequency, which is used to dynamically correct the system's cache update mechanism. After optimizing the cache update frequency, if the re-acquired collaborative allocation latency index is not greater than the preset collaborative allocation latency index in the database, the cache update frequency optimization is completed and the latency assessment of the standardized-weighted fusion collaborative process is performed; otherwise, a standardized-weighted fusion collaborative warning is issued.

5. The adaptive collaborative processing method for medical device test data as described in claim 1, characterized in that, The specific process for delaying the evaluation of the standardized-weighted fusion collaboration process during the fusion process is as follows: The obtained collaborative fusion delay data is compared with the preset collaborative fusion delay data in the database one by one to obtain the difference comparison results. At the same time, the difference comparison results are summed and the average value is calculated to obtain the collaborative fusion delay index. The collaborative fusion delay data includes feature mapping computation time and weight adaptation computation time. The feature mapping computation time represents the time consumed in the process of transforming standardized multimodal data into data of the same dimension during the fusion of medical device test data. The weight adaptation computation time represents the computation time consumed by the central server in the process of multimodal data fusion to assign weights to different modal features. The collaborative fusion delay index is used to quantify the degree of impact of collaborative fusion delay data on the standardized-weighted fusion collaborative process.

6. The adaptive collaborative processing method for medical device test data as described in claim 5, characterized in that, The standardized-weighted fusion collaborative feedback process is as follows: The obtained collaborative fusion delay index is compared with the preset collaborative fusion delay index in the database. If the obtained collaborative fusion delay index is not greater than the preset collaborative fusion delay index, it indicates that the standardized-weighted fusion collaboration process is qualified, and the delay assessment of the collaborative decision matching process is carried out. If the obtained collaborative fusion delay index is greater than the preset collaborative fusion delay index, it indicates that the standardized-weighted fusion collaborative process is unqualified, and node processor performance optimization is performed.

7. The adaptive collaborative processing method for medical device test data as described in claim 6, characterized in that, The specific steps for optimizing the node processor performance are as follows: Based on the mapping relationship between the obtained collaborative fusion latency index deviation and the corresponding instruction cycle number adjustment value and node processor clock frequency adjustment value in the database, the instruction cycle number decrease value and node processor clock frequency increase value are obtained. The current instruction cycle number and node processor clock frequency are obtained. The instruction cycle number decrease value is subtracted from the current instruction cycle number to obtain the actual instruction cycle number. The node processor clock frequency is added to the node processor clock frequency increase value to obtain the actual node processor clock frequency. By reducing the number of instruction cycles to shorten the execution time of each instruction, the time consumed in the entire data processing stage can be reduced. At the same time, by increasing the clock frequency of the node processor, the pause and waiting time of medical device test data can be reduced. After the node processor performance is optimized, if the re-acquired collaborative fusion latency index is not greater than the preset collaborative fusion latency index in the database, the node processor performance optimization is completed and the latency assessment of collaborative decision matching is performed; otherwise, a standardized-weighted fusion collaborative warning is issued.

8. The adaptive collaborative processing method for medical device test data as described in claim 1, characterized in that, The delay evaluation of the collaborative decision matching process based on the number of node collaborative interactions obtained specifically includes: collaborative decision matching delay determination and collaborative decision matching delay optimization. The specific process for determining the delay in collaborative decision-making matching is as follows: If the number of node collaborative interactions obtained is not greater than the number of node collaborative interactions preset in the database, it indicates that the collaborative decision matching is qualified and the adaptive collaborative processing evaluation of medical device test data is completed. The adaptive collaborative processing evaluation includes classification-scheduling collaborative delay evaluation, standardization-weighted fusion collaborative delay evaluation, and collaborative decision matching evaluation. If the number of node collaborative interactions obtained is greater than the preset number of node collaborative interactions in the database, it indicates that the collaborative decision matching is unqualified, and collaborative decision matching delay optimization is performed. The number of node collaborative interactions represents the number of times that different decision nodes need to communicate with each other, share data, and coordinate their work in order to complete the data collaboration task in collaborative decision matching.

9. The adaptive collaborative processing method for medical device test data as described in claim 8, characterized in that, The specific steps for optimizing the collaborative decision-making matching delay are as follows: The obtained collaborative decision matching delay deviation score is input into the database and mapped to the corresponding node parallel quantity adjustment value in the database to obtain the node parallel quantity adjustment value. The current node parallel quantity is added to the node parallel quantity adjustment value to obtain the actual node parallel quantity. By increasing the parallel quantity of decision nodes, the efficiency of decision task processing and the overall system response speed are improved. After optimization of the collaborative decision matching delay, if the number of reacquired node collaborative interactions is not greater than the preset number of node collaborative interactions in the database, it indicates that the collaborative decision matching is qualified and the adaptive collaborative processing evaluation of medical device test data is completed; otherwise, a collaborative decision matching warning is issued.

10. A system applying the adaptive collaborative processing method for medical device test data as described in any one of claims 1-9, characterized in that, include: The module includes a collaborative allocation delay assessment module, a collaborative fusion delay assessment module, and a collaborative decision matching delay assessment module. The collaborative allocation delay assessment module is used to receive medical device test data collected from multiple terminal medical monitoring devices in medical care monitoring scenarios, perform preliminary classification using edge collaborative nodes, and upload the data to the central server. At the same time, it performs delay assessment on the classification-scheduling collaborative process of medical device test data through the edge collaborative mechanism to determine whether there is a need to optimize the cache update frequency. The collaborative fusion delay assessment module is used to, if the classification-scheduling collaborative process is deemed qualified, use the central server node to fuse the multimodal data aggregated to the central server, and at the same time assess the delay of the standardization-weighted fusion collaborative process during the fusion process, and provide standardization-weighted fusion collaborative feedback. The collaborative decision matching delay assessment module is used to perform collaborative decision matching in the data collaboration process based on the results of standardized-weighted fusion collaborative feedback, and at the same time to assess the delay of the collaborative decision matching process based on the number of node collaborative interactions obtained.

Images