Data management method for multi-target biohazard factor immunoassay analyzer

Abstract

The invention discloses a data management method for a multi-target biohazard factor immunoassay analyzer, relates to the technical field of biological monitoring data management, and solves the technical problems that the confidentiality of sample data is difficult to guarantee, only a digital record of a single detection result is provided, multi-factor correlation analysis is lacked, and the accuracy of immunoassay is influenced. By monitoring the data storage saturation coefficient, the operation saturation coefficient and the operation abnormal coefficient of the edge node in real time, the potential operation pressure of the edge node can be found in time, so that corresponding measures are taken for prevention and optimization, the performance of the edge node system is prevented from being reduced, and according to the comprehensive operation pressure analysis result, the safety of the edge node system is improved. Data storage and calculation tasks of edge nodes can be distributed more reasonably, and resources are ensured to be fully utilized. The cloud processing platform can analyze detection configuration parameters, experiment signal data and quality control data in real time when different types of reagent cards are experimented, and the accuracy and reliability of the data are ensured.

Description

Technical Field

[0001] The invention belongs to the field of biological monitoring data management, and in particular is a data management method for a multi-target biological hazard factor immunoanalyzer. Background Art

[0002] In recent years, the field of biotechnology has made significant progress, especially the development of molecular biology, immunology and bioinformatics, which has provided a solid theoretical basis and technical support for the research and development of multi-target biohazard factor immunoanalyzers. These technological advances have enabled researchers to gain a deeper understanding of the structure and function of biomolecules, thereby designing more accurate and efficient immunoassay methods. Immunoassay technology is one of the core technologies for the detection of multi-target biohazard factors. With the continuous innovation of antibody preparation technology, fluorescent labeling technology and quantum dot immunochromatography technology, the sensitivity and accuracy of immunoassay technology have been significantly improved. These innovative technologies provide a more sensitive, accurate and efficient detection method for the research and development of multi-target biohazard factor immunoanalyzers. Multi-target biohazard factor immunoanalyzers are advanced instruments for the detection and analysis of a variety of biohazard factors, and are widely used in public health, food safety, environmental monitoring and other fields. With the rapid development of biotechnology and the increasing demand for biosafety, the importance of data management systems in this field has become increasingly prominent.

[0003] Traditional systems rely on manual export-analysis processes, which delays decision-making; large amounts of sample data are directly uploaded to the cloud, making it difficult to ensure the confidentiality of sample data; at the same time, most data management solutions for immunoassay analyzers analyze various data in sequence and only provide digital records of single test results, lacking multi-factor correlation analysis, which affects the accuracy of immunoassays. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a data management method for a multi-target biohazard factor immunoanalyzer, which is used to solve the technical problems that it is difficult to ensure the confidentiality of sample data, only provide a digital record of a single test result, lack of multi-factor association analysis, and affect the accuracy of immunoanalysis.

[0005] To solve the above problems, the first aspect of the present invention provides a data management method for a multi-target biohazard factor immunoassay analyzer, comprising the following steps: An edge node is set up on the immunoassay analyzer to store sample information and environmental data on the edge, and detection configuration parameters, experimental signal data, and quality control data on the cloud. By building a key-value pair about SampleID, data association between corresponding data is achieved. A unique RFID tag is embedded in the reagent card, and the RFID tag records the reagent card number, SampleID and sample data storage address; Record the device ID, device type, device location and usage frequency in the device management table, and group the immunoassay analyzers into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group according to the different threshold ranges of the device usage frequencies; Detect the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge nodes of each immunoanalyzer group, analyze the comprehensive operation pressure, allocate the data storage and calculation tasks of the edge nodes according to the comprehensive operation pressure analysis results, and modify the corresponding data table; The cloud processing platform performs real-time analysis on the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards during experiments, performs reliability analysis on the experimental data, and conducts re-experiments on experimental data that may have fault detection until there is no fault detection in the experimental data. The corresponding signal data during the experiment is recorded as the final data in the data table of the cloud processing platform.

[0006] Optionally, in an example of the above aspect, an edge node is set at the immunoassay analyzer end, sample information and environmental data are stored at the edge end, detection configuration parameters, experimental signal data and quality control data are stored in the cloud, and data association between corresponding data is achieved by constructing a key-value pair about SampleID, including the following steps: An edge node is set up at the immunoanalyzer end to obtain sample information of biological samples detected by the edge end, collect detection configuration parameters of the immunoanalyzer, and receive experimental signal data, quality control data, and environmental data of the immunoanalyzer in parallel through the multimodal sensor interface during the biological sample detection process; The collected data is timestamped and the signal baseline is corrected based on the noise filtering technology of wavelet transform. The detection configuration parameters, experimental signal data and quality control data are encrypted and sent to the cloud processing platform to generate the corresponding data table, with SampleID as the foreign key. According to the sample information and environmental data stored on the edge, the sample and environmental data table uses SampleID as the foreign key, and the storage address of the sample and environmental data table corresponding to SampleID is sent to the cloud processing platform; A reagent card management list and a device management table are established in the database of the cloud processing platform. The reagent cards are classified and numbered for each classification. The device affiliation of the reagent card is recorded. The SampleID of the reagent card test sample is added to the reagent card management list. The SampleID list is formed by counting the SampleIDs corresponding to the device test samples, and the corresponding device ID is used as the foreign key of the SampleID list.

[0007] Optionally, in an example of the above aspect, classifying the reagent cards includes: According to the reaction mechanism of the reagent card, it can be classified into: enzyme-linked immunosorbent assay reagent card, fluorescent immunoassay reagent card and chemiluminescent immunoassay reagent card; A reagent card management list is set for each category of reagent cards, and the target type of the corresponding reagent card is marked in the reagent card management list, including: bacterial target, viral target, toxin target and parasite target.

[0008] Optionally, in an example of the above aspect, the device ID, device type, device location and usage frequency are recorded in the device management table, and the immunoassay analyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group according to the usage frequencies of the devices being within different threshold ranges, including the following steps: Record the device ID, device type, device location and usage frequency in the device management table, set the detection time period, and count the usage frequency of the devices in the detection time period; Obtain the usage frequency data of the device in several past detection time periods, calculate the average value to obtain the average frequency data, and perform weighted average of the average frequency data and the usage frequency data of the device in the latest detection time period to obtain the usage frequency data for device grouping; Collect historical usage frequency data of the equipment and rank them, taking the top 20%, top 40%, and top 60% usage frequency data nodes as the corresponding usage frequency thresholds; According to the usage frequencies of the obtained devices being within different threshold ranges, the immunoanalyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group, and the groupings are updated regularly.

[0009] Optionally, in an example of the above aspect, detecting the edge node data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of each immune analyzer group includes the following steps: For the immunoanalyzers in the high-frequency usage group and the second-high-frequency usage group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are tested in each testing time period; For the immunoanalyzers in the medium frequency use group, the memory usage data, CPU usage data and operation abnormality data of the immunoanalyzers are tested at intervals of several test time periods; For the immunoanalyzers in the low-frequency use group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are detected with the receipt of error messages more than three times during the detection period as the trigger condition; The edge node data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient are calculated respectively according to the memory usage data, CPU usage data and operation abnormality data.

[0010] Optionally, in an example of the above aspect, calculating the edge node data storage saturation coefficient, the operation saturation coefficient and the operation abnormality coefficient respectively according to the memory occupancy data, the CPU occupancy data and the operation abnormality data includes the following steps: The data storage saturation coefficient of the edge node is calculated using the following formula: ; Among them, SSI is the data storage saturation coefficient, Ct is the currently occupied storage space, C is the total storage capacity of the node, St is the hourly increment of node data, α and β are the corresponding weights, and the default α=0.6, β=0.4; The operating saturation coefficient of the edge node is calculated using the following formula: ; Among them, OSI is the ‌operation saturation coefficient, CPUt / CPU is the current CPU occupancy rate, Ct / C is the current memory occupancy rate, N is the maximum concurrent task capacity, Nt is the number of tasks in the pending task queue, γ is the queue pressure weight, and the default γ=0.2; The occurrence data of different types of abnormal events in different detection time periods are counted, and the operation abnormality coefficient of the edge node is calculated using the following formula: ; Among them, OAI is the operation abnormality coefficient, Eti is the number of occurrences of the i-th abnormal event in the detection time period, Ebasei is the benchmark frequency of similar abnormal events, which is set to the historical average of similar abnormal events, t0 is the length of the detection time period, Wi is the event severity weight, for example: hardware failure λ=0.1, communication delay λ=0.3, μ is the time attenuation factor, and the default μ=0.1.

[0011] Optionally, in an example of the above aspect, analyzing the comprehensive operating pressure includes the following steps: For the immunoassay analyzers in the high-frequency use group and the sub-high-frequency use group, the comprehensive operation pressure is analyzed by the following formula based on the edge node data storage saturation coefficient, operation saturation coefficient, and operation abnormality coefficient: ; Among them, Health1 is the comprehensive operation pressure analysis value of the immunoassay analyzer in the high-frequency use group and the sub-high-frequency use group, SSI is the data storage saturation coefficient, OSI is the operation saturation coefficient, OAI is the operation abnormality coefficient, σ is the node health correction factor, σ=1-(Fhardware / Fmax), Fhardware is the current hardware aging score, and Fmax is the initial value of the node hardware aging score; For the immunoassay analyzers in the medium frequency group, the comprehensive operating pressure is analyzed by the following formula: ; Among them, Health2 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the medium frequency use group; For the immunoassay analyzers in the low-frequency use group, the comprehensive operating pressure is analyzed by the following formula: ; Among them, Health3 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the low-frequency use group.

[0012] Optionally, in an example of the above aspect, allocating data storage and computing tasks of edge nodes according to the comprehensive operation pressure analysis result includes the following steps: According to the results of the comprehensive operation pressure analysis, if the comprehensive operation pressure analysis value is greater than the first threshold, the nearest node is selected from the nodes whose comprehensive operation pressure analysis value is less than the second threshold, 10% of the historical storage data is transferred to the selected node, and the last 10% of the tasks in the pending task queue are transferred to the cloud processing platform until the comprehensive operation pressure analysis value of the corresponding node is no greater than the first threshold.

[0013] Optionally, in an example of the above aspect, the cloud processing platform performs reliability analysis on the data during the experiment by performing real-time analysis on the detection configuration parameters, experimental signal data and quality control data during the experiments of reagent cards of different categories, including the following steps: The cloud processing platform obtains reagent cards of different categories, detection configuration parameters for different target types, experimental signal data, historical data of normal detection of quality control data, and historical data when detection failures occur, and adds normal detection and fault detection labels to the historical data; Establish a multimodal deep confidence assessment model, and use different classifications of reagent cards to perform target type detection on bacterial targets, viral targets, toxin targets, and parasite targets to train the corresponding model; The cloud processing platform obtains the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards in real time; Through the trained model, the reliability analysis of the experimental data is performed to identify the data of the corresponding target type detection performed by the corresponding classified reagent card, which is normal detection or fault detection data. If fault detection data appears, the data during the experiment is unreliable and is fault detection data. Otherwise, the data during the experiment is reliable.

[0014] Optionally, in an example of the above aspect, establishing a multimodal depth confidence assessment model includes the following steps: Construct a multimodal deep belief network and set static parameter branch, dynamic signal branch and quality control data branch in the input layer; The static parameter branch is used to input the detection configuration parameters and is set as a fully connected layer. The dynamic signal branch is used to input the experimental signal data and is set as an LSTM network layer. The LSTM network layer processes the time series signal of the experimental signal data, and the sliding window length is set to 30 sampling cycles. The quality control data branch is used to input the quality control data and is set as a time series convolution layer. The cumulative effect of quality control deviation is captured by constructing a time series convolution network, and the convolution kernel width matches the calibration cycle. The fusion layer sets up a cross-modal attention mechanism to weightedly fuse the features of each branch; The output layer uses the Sigmoid activation function, and the output data is the classification of normal detection and fault detection.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention can timely discover the operating pressure of potential edge nodes by real-time monitoring of the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of edge nodes, so as to take corresponding measures for prevention and optimization, and avoid edge node system crash or performance degradation. According to the comprehensive operation pressure analysis results, the data storage and computing tasks of edge nodes can be more reasonably allocated to ensure that resources are fully utilized and avoid waste of resources and overload. This dynamic resource allocation method can be flexibly adjusted according to actual needs, improving the flexibility and response speed of the system. By optimizing resource allocation and reducing system bottlenecks, the overall performance of immunoanalyzer grouping can be significantly improved.

[0016] The cloud processing platform of the present invention can analyze the detection configuration parameters, experimental signal data and quality control data of reagent cards of different classifications in real time to ensure the accuracy and reliability of the data. For experimental data that may have fault detection, the platform will conduct another experiment until there is no fault in the data detection, thereby further improving the reliability of the data. Through the automated analysis of the cloud processing platform, the possibility of manual intervention and errors is reduced, and the experimental efficiency is improved. The platform can intelligently manage the experimental process, including data entry, analysis, storage and report generation, thereby realizing the automated management of the experiment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 This is a schematic diagram of the classification of reagent cards of the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0020] See also Figure 1-Figure 2 The first aspect of the present invention provides a data management method for a multi-target biohazard factor immunoassay analyzer, comprising the following steps: An edge node is set up on the immunoassay analyzer to store sample information and environmental data on the edge, and detection configuration parameters, experimental signal data, and quality control data on the cloud. By building a key-value pair about SampleID, data association between corresponding data is achieved. A unique RFID tag is embedded in the reagent card, and the RFID tag records the reagent card number, SampleID and sample data storage address; Record the device ID, device type, device location and usage frequency in the device management table, and group the immunoassay analyzers into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group according to the different threshold ranges of the device usage frequencies; Detect the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge nodes of each immunoanalyzer group, analyze the comprehensive operation pressure, allocate the data storage and calculation tasks of the edge nodes according to the comprehensive operation pressure analysis results, and modify the corresponding data table; The cloud processing platform performs real-time analysis on the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards during experiments, performs reliability analysis on the experimental data, and conducts re-experiments on experimental data that may have fault detection until there is no fault detection in the experimental data. The corresponding signal data during the experiment is recorded as the final data in the data table of the cloud processing platform.

[0021] Specifically, in this embodiment, an edge node is set at the immune analyzer end; Edge node configuration, set to: Hardware: NVIDIA Jetson Xavier, 16GB memory + 256GB SSD; Data compression: Apache Parquet column storage is used, with a compression ratio of 5:1; The sample information and environmental data are stored at the edge, and the detection configuration parameters, experimental signal data, and quality control data are stored in the cloud. By constructing key-value pairs about SampleID, data association between corresponding data is achieved; ‌Secure transport protocol, set to:‌ Edge → Cloud: MQTT over TLS 1.3, batch transmission every 5 minutes; Key field encryption: The sample ID is encrypted using the national encryption SM4 algorithm.

[0022] Data delay tests were performed on the cloud and edge, and the results are shown in Table 1 below; Table 1: Cloud and edge performance test table

[0023] By making the edge lightweight, only small-scale data with high-frequency access (such as the SampleID mapping table) is retained to reduce the storage pressure of edge devices. Cloud storage is easy to expand on demand. Experimental signals and quality control data are usually large in volume. Cloud storage can be elastically expanded on demand to avoid excessive investment in local hardware.

[0024] Environmental data (such as temperature and humidity) and sample metadata (such as ID and type) are stored at the edge, supporting real-time reading and fast response, avoiding cloud transmission delays, and suitable for instant decision-making during the experiment.

[0025] Large-scale data such as experimental signals are processed centrally in the cloud, and the high-performance computing capabilities of cloud resources are used to complete complex analysis, taking into account both efficiency and flexibility.

[0026] The RFID tag records the reagent card number, SampleID and sample data storage address. In this embodiment, the RFID reagent card setting includes: Reagent card number: RC-001-A; SampleID: SMP-20240515-007; Sample data storage address: Edge node path: EdgeNode-3 / Samples / SMP-20240515-007.json; SampleID is used to automatically match cloud quality control data, such as Cloud / QC / SMP-20240515-007.csv, with edge sample metadata.

[0027] The device ID, device type, device location and usage frequency are recorded in the device management table. The structure of the device management table is shown in Table 2 below: Table 2: Equipment Management Table

[0028] The device log is used to count the average daily usage of each immunoassay analyzer in real time. For example, the device IA-003 has tested 63 samples in the past 7 days → the average daily usage frequency = 9 times.

[0029] The system automatically runs the grouping algorithm every morning to update the device grouping labels.

[0030] Detect the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge nodes of each immunoanalyzer group, analyze the comprehensive operation pressure, allocate the data storage and calculation tasks of the edge nodes according to the comprehensive operation pressure analysis results, and modify the corresponding data table; By real-time monitoring of the data storage saturation coefficient, operation saturation coefficient, and operation abnormality coefficient of edge nodes, potential operating pressure of edge nodes can be discovered in time, so that corresponding measures can be taken for prevention and optimization to avoid edge node system crashes or performance degradation. According to the results of the comprehensive operation pressure analysis, the data storage and computing tasks of edge nodes can be allocated more reasonably to ensure that resources are fully utilized and to avoid resource waste and overload. This dynamic resource allocation method can be flexibly adjusted according to actual needs, improving the flexibility and response speed of the system. By optimizing resource allocation and reducing system bottlenecks, the overall performance of the immunoanalyzer grouping can be significantly improved, including data processing speed and response time.

[0031] Analyze the comprehensive operating pressure, and allocate data storage and computing tasks to edge nodes based on the results of the comprehensive operating pressure analysis, so as to timely discover and deal with potential problems and avoid increased maintenance costs due to edge node system crashes or performance degradation; at the same time, reasonable resource allocation and optimized system performance can also help reduce the frequency and cost of hardware and software updates.

[0032] The cloud processing platform performs real-time analysis on the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards during experiments, performs reliability analysis on the experimental data, and conducts re-experiments on experimental data that may have fault detection until there is no fault detection in the experimental data. The corresponding signal data during the experiment is recorded as the final data in the data table of the cloud processing platform.

[0033] The cloud processing platform can analyze the detection configuration parameters, experimental signal data and quality control data of reagent card experiments of different categories in real time to ensure the accuracy and reliability of the data. For experimental data that may have faulty detection, the platform will conduct another experiment until there is no fault in the data detection, thereby further improving the reliability of the data. Through the automated analysis of the cloud processing platform, the possibility of manual intervention and errors is reduced, and the experimental efficiency is improved. The platform can intelligently manage the experimental process, including data entry, analysis, storage and report generation, thereby realizing the automated management of experiments. The cloud processing platform provides a safe and reliable data storage environment to ensure that experimental data will not be lost or tampered with. Cloud storage also means that users can access experimental data through the Internet anytime and anywhere, improving the availability and flexibility of data.

[0034] In one embodiment of the present invention, an edge node is set at the immunoassay analyzer end, sample information and environmental data are stored at the edge end, detection configuration parameters, experimental signal data and quality control data are stored in the cloud, and data association between corresponding data is achieved by constructing a key-value pair about SampleID, including the following steps: An edge node is set up at the immunoanalyzer end to obtain sample information of biological samples detected by the edge end, collect detection configuration parameters of the immunoanalyzer, and receive experimental signal data, quality control data, and environmental data of the immunoanalyzer in parallel through the multimodal sensor interface during the biological sample detection process; The collected data is timestamped and the signal baseline is corrected based on the noise filtering technology of wavelet transform. The detection configuration parameters, experimental signal data and quality control data are encrypted and sent to the cloud processing platform to generate the corresponding data table, with SampleID as the foreign key. According to the sample information and environmental data stored on the edge, the sample and environmental data table uses SampleID as the foreign key, and the storage address of the sample and environmental data table corresponding to SampleID is sent to the cloud processing platform; A reagent card management list and a device management table are established in the database of the cloud processing platform. The reagent cards are classified and numbered for each classification. The device affiliation of the reagent card is recorded. The SampleID of the reagent card test sample is added to the reagent card management list. The SampleID list is formed by counting the SampleIDs corresponding to the device test samples, and the corresponding device ID is used as the foreign key of the SampleID list.

[0035] Furthermore, the reagent cards are classified into: According to the reaction mechanism of the reagent card, it can be classified into: enzyme-linked immunosorbent assay reagent card, fluorescent immunoassay reagent card and chemiluminescent immunoassay reagent card; A reagent card management list is set for each category of reagent cards, and the target type of the corresponding reagent card is marked in the reagent card management list, including: bacterial target, viral target, toxin target and parasite target.

[0036] Specifically, in this embodiment, the target types are divided into: Bacterial Target: Antibodies or antigens that are specific to a specific bacterium.

[0037] Viral Target: An antibody or antigen that is specific to a specific virus.

[0038] Toxin targets: Detection of specific toxins, such as bacterial toxins, fungal toxins, etc.

[0039] Parasite target: An antibody or antigen that is directed against a specific parasite.

[0040] The multimodal sensing interface receives optical signals (fluorescence intensity), electrochemical signals (current / voltage), mass spectrometry data (m / z value), etc. in parallel, aligns the timestamps of the collected data, and corrects the signal baseline based on the noise filtering technology of wavelet transform.

[0041] In one embodiment of the present invention, the device ID, device type, device location and usage frequency are recorded in the device management table, and the immunoassay analyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group according to the usage frequencies of the devices being within different threshold ranges, including the following steps: Record the device ID, device type, device location and usage frequency in the device management table, set the detection time period, and count the usage frequency of the devices in the detection time period; Obtain the usage frequency data of the device in several past detection time periods, calculate the average value to obtain the average frequency data, and perform weighted average of the average frequency data and the usage frequency data of the device in the latest detection time period to obtain the usage frequency data for device grouping; Collect historical usage frequency data of the equipment and rank them, taking the top 20%, top 40%, and top 60% usage frequency data nodes as the corresponding usage frequency thresholds; According to the usage frequencies of the obtained devices being within different threshold ranges, the immunoanalyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group, and the groupings are updated regularly.

[0042] In one embodiment of the present invention, detecting the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge node of each immune analyzer group includes the following steps: For the immunoanalyzers in the high-frequency usage group and the second-high-frequency usage group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are tested in each testing time period; For the immunoanalyzers in the medium frequency use group, the memory usage data, CPU usage data and operation abnormality data of the immunoanalyzers are tested at intervals of several test time periods; For the immunoanalyzers in the low-frequency use group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are detected with the receipt of error messages more than three times during the detection period as the trigger condition; The edge node data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient are calculated respectively according to the memory usage data, CPU usage data and operation abnormality data.

[0043] In one embodiment of the present invention, the edge node data storage saturation coefficient, the operation saturation coefficient and the operation abnormality coefficient are calculated respectively according to the memory occupancy data, the CPU occupancy data and the operation abnormality data, including the following steps: The data storage saturation coefficient of the edge node is calculated using the following formula: ; Among them, SSI is the data storage saturation coefficient, Ct is the currently occupied storage space, C is the total storage capacity of the node, St is the hourly increment of node data, α and β are the corresponding weights, and the default α=0.6, β=0.4; The operating saturation coefficient of the edge node is calculated using the following formula: ; Among them, OSI is the ‌operation saturation coefficient, CPUt / CPU is the current CPU occupancy rate, Ct / C is the current memory occupancy rate, N is the maximum concurrent task capacity, Nt is the number of tasks in the pending task queue, γ is the queue pressure weight, and the default γ=0.2; The occurrence data of different types of abnormal events in different detection time periods are counted, and the operation abnormality coefficient of the edge node is calculated using the following formula: ; Among them, OAI is the operation abnormality coefficient, Eti is the number of occurrences of the i-th abnormal event in the detection time period, Ebasei is the benchmark frequency of similar abnormal events, which is set to the historical average of similar abnormal events, t0 is the length of the detection time period, Wi is the event severity weight, for example: hardware failure λ=0.1, communication delay λ=0.3, μ is the time attenuation factor, and the default μ=0.1.

[0044] In one embodiment of the present invention, analyzing the comprehensive operating pressure includes the following steps: For the immunoassay analyzers in the high-frequency use group and the sub-high-frequency use group, the comprehensive operation pressure is analyzed by the following formula based on the edge node data storage saturation coefficient, operation saturation coefficient, and operation abnormality coefficient: ; Among them, Health1 is the comprehensive operation pressure analysis value of the immunoassay analyzer in the high-frequency use group and the sub-high-frequency use group, SSI is the data storage saturation coefficient, OSI is the operation saturation coefficient, OAI is the operation abnormality coefficient, σ is the node health correction factor, σ=1-(Fhardware / Fmax), Fhardware is the current hardware aging score, and Fmax is the initial value of the node hardware aging score; For the immunoassay analyzers in the medium frequency group, the comprehensive operating pressure is analyzed by the following formula: ; Among them, Health2 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the medium frequency use group; For the immunoassay analyzers in the low-frequency use group, the comprehensive operating pressure is analyzed by the following formula: ; Among them, Health3 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the low-frequency use group.

[0045] In one embodiment of the present invention, data storage and computing tasks of edge nodes are allocated according to the comprehensive operation pressure analysis results, including the following steps: According to the results of the comprehensive operation pressure analysis, if the comprehensive operation pressure analysis value is greater than the first threshold, the nearest node is selected from the nodes whose comprehensive operation pressure analysis value is less than the second threshold, 10% of the historical storage data is transferred to the selected node, and the last 10% of the tasks in the pending task queue are transferred to the cloud processing platform until the comprehensive operation pressure analysis value of the corresponding node is no greater than the first threshold.

[0046] Specifically, a large number of operating data of immune analyzers at edge nodes are obtained, and the comprehensive operating pressure analysis values ​​of the immune analyzers are calculated. The corresponding comprehensive operating pressure analysis values ​​are obtained from the data where the edge node data delay is significantly increased and the response time is prolonged, and the average value is taken as the first threshold value. The comprehensive operating pressure analysis values ​​of the immune analyzers at the edge end where the edge node data delay time is less than 50ms are statistically counted, and the average value is taken as the second threshold value.

[0047] In one embodiment of the present invention, the cloud processing platform performs real-time analysis on the detection configuration parameters, experimental signal data and quality control data of different categories of reagent card experiments, and performs reliability analysis on the experimental data, including the following steps: The cloud processing platform obtains reagent cards of different categories, detection configuration parameters for different target types, experimental signal data, historical data of normal detection of quality control data, and historical data when detection failures occur, and adds normal detection and fault detection labels to the historical data; Establish a multimodal deep confidence assessment model, and use different classifications of reagent cards to perform target type detection on bacterial targets, viral targets, toxin targets, and parasite targets to train the corresponding model; The cloud processing platform obtains the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards in real time; Through the trained model, the reliability analysis of the experimental data is performed to identify the data of the corresponding target type detection performed by the corresponding classified reagent card, which is normal detection or fault detection data. If fault detection data appears, the data during the experiment is unreliable and is fault detection data. Otherwise, the data during the experiment is reliable.

[0048] Furthermore, a multimodal deep confidence assessment model is established, including the following steps: Construct a multimodal deep belief network and set static parameter branch, dynamic signal branch and quality control data branch in the input layer; The static parameter branch is used to input the detection configuration parameters and is set as a fully connected layer. The dynamic signal branch is used to input the experimental signal data and is set as an LSTM network layer. The LSTM network layer processes the timing signals of the experimental signal data, such as fluorescence intensity fluctuations and robotic arm acceleration. The sliding window length is set to 30 sampling cycles. The quality control data branch is used to input the quality control data and is set as a time series convolution layer. The cumulative effect of quality control deviations is captured by constructing a time series convolution network, and the convolution kernel width matches the calibration cycle. The fusion layer sets up a cross-modal attention mechanism to weightedly fuse the features of each branch; The output layer uses the Sigmoid activation function, and the output data is the classification of normal detection and fault detection.

[0049] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A data management method for a multi-target biohazard factor immunoassay analyzer, characterized in that: The following steps are involved: An edge node is set up on the immunoassay analyzer to store sample information and environmental data on the edge, and detection configuration parameters, experimental signal data, and quality control data on the cloud. By building a key-value pair about SampleID, data association between corresponding data is achieved. A unique RFID tag is embedded in the reagent card, and the RFID tag records the reagent card number, SampleID and sample data storage address; Record the device ID, device type, device location and usage frequency in the device management table, and group the immunoassay analyzers into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group according to the different threshold ranges of the device usage frequencies; Detect the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge nodes of each immunoanalyzer group, analyze the comprehensive operation pressure, allocate the data storage and calculation tasks of the edge nodes according to the comprehensive operation pressure analysis results, and modify the corresponding data table; The cloud processing platform performs real-time analysis on the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards during experiments, performs reliability analysis on the experimental data, and conducts re-experiments on experimental data that may have fault detection until there is no fault detection in the experimental data. The corresponding signal data during the experiment is recorded as the final data in the data table of the cloud processing platform.

2. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 1, characterized in that: An edge node is set up on the immunoassay analyzer to store sample information and environmental data on the edge, and detection configuration parameters, experimental signal data, and quality control data on the cloud. By constructing a key-value pair about SampleID, data association between corresponding data is achieved, including the following steps: An edge node is set up at the immunoanalyzer end to obtain sample information of biological samples detected by the edge end, collect detection configuration parameters of the immunoanalyzer, and receive experimental signal data, quality control data, and environmental data of the immunoanalyzer in parallel through the multimodal sensor interface during the biological sample detection process; The collected data is timestamped and the signal baseline is corrected based on the noise filtering technology of wavelet transform. The detection configuration parameters, experimental signal data and quality control data are encrypted and sent to the cloud processing platform to generate the corresponding data table, with SampleID as the foreign key. According to the sample information and environmental data stored on the edge, the sample and environmental data table uses SampleID as the foreign key, and the storage address of the sample and environmental data table corresponding to SampleID is sent to the cloud processing platform; A reagent card management list and a device management table are established in the database of the cloud processing platform. The reagent cards are classified and numbered for each classification. The device affiliation of the reagent card is recorded. The SampleID of the reagent card test sample is added to the reagent card management list. The SampleID list is formed by counting the SampleIDs corresponding to the device test samples, and the corresponding device ID is used as the foreign key of the SampleID list.

3. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 2, characterized in that: Categorize the reagent cards, including: According to the reaction mechanism of the reagent card, it can be classified into: enzyme-linked immunosorbent assay reagent card, fluorescent immunoassay reagent card and chemiluminescent immunoassay reagent card; A reagent card management list is set for each category of reagent cards, and the target type of the corresponding reagent card is marked in the reagent card management list, including: bacterial target, viral target, toxin target and parasite target.

4. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 1, characterized in that: The device ID, device type, device location and usage frequency are recorded in the device management table. According to the usage frequencies of the devices being within different threshold ranges, the immunoassay analyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group, including the following steps: Record the device ID, device type, device location and usage frequency in the device management table, set the detection time period, and count the usage frequency of the devices in the detection time period; Obtain the usage frequency data of the device in several past detection time periods, calculate the average value to obtain the average frequency data, and perform weighted average of the average frequency data and the usage frequency data of the device in the latest detection time period to obtain the usage frequency data for device grouping; Collect historical usage frequency data of the equipment and rank them, taking the top 20%, top 40%, and top 60% usage frequency data nodes as the corresponding usage frequency thresholds; According to the usage frequencies of the obtained devices being within different threshold ranges, the immunoanalyzers are grouped into a high-frequency usage group, a sub-high-frequency usage group, a medium-frequency usage group and a low-frequency usage group, and the groupings are updated regularly.

5. The data management method for a multi-target biohazard factor immunoassay analyzer according to claim 1, characterized in that: Detecting the data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient of the edge node of each immune analyzer group includes the following steps: For the immunoanalyzers in the high-frequency usage group and the second-high-frequency usage group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are tested in each testing time period; For the immunoanalyzers in the medium frequency use group, the memory usage data, CPU usage data and operation abnormality data of the immunoanalyzers are tested at intervals of several test time periods; For the immunoanalyzers in the low-frequency use group, the memory usage data, CPU usage data, and operation abnormality data of the immunoanalyzers are detected with the receipt of error messages more than three times during the detection period as the trigger condition; The edge node data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient are calculated respectively according to the memory usage data, CPU usage data and operation abnormality data.

6. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 5, characterized in that: According to the memory usage data, CPU usage data and operation abnormality data, the edge node data storage saturation coefficient, operation saturation coefficient and operation abnormality coefficient are calculated respectively, including the following steps: The data storage saturation coefficient of the edge node is calculated using the following formula: ； Among them, SSI is the data storage saturation coefficient, Ct is the currently occupied storage space, C is the total storage capacity of the node, St is the hourly increment of node data, and α and β are the corresponding weights; The operating saturation coefficient of the edge node is calculated using the following formula: ； Among them, OSI is the ‌operation saturation coefficient, CPUt / CPU is the current CPU occupancy rate, Ct / C is the current memory occupancy rate, N is the maximum concurrent task capacity, Nt is the number of tasks in the task queue to be processed, and γ is the queue pressure weight; The occurrence data of different types of abnormal events in different detection time periods are counted, and the operation abnormality coefficient of the edge node is calculated using the following formula: ； Among them, OAI is the operation abnormality coefficient, Eti is the number of occurrences of the i-th abnormal event in the detection time period, Ebasei is the benchmark frequency of similar abnormal events, which is set to the historical mean of similar abnormal events, t0 is the length of the detection time period, Wi is the event severity weight, and μ is the time attenuation factor.

7. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 1, characterized in that: Analyze the comprehensive operating pressure, including the following steps: For the immunoassay analyzers in the high-frequency use group and the sub-high-frequency use group, the comprehensive operation pressure is analyzed by the following formula based on the edge node data storage saturation coefficient, operation saturation coefficient, and operation abnormality coefficient: ； Among them, Health1 is the comprehensive operation pressure analysis value of the immunoassay analyzer in the high-frequency use group and the sub-high-frequency use group, SSI is the data storage saturation coefficient, OSI is the operation saturation coefficient, OAI is the operation abnormality coefficient, σ is the node health correction factor, σ=1-(Fhardware / Fmax), Fhardware is the current hardware aging score, and Fmax is the initial value of the node hardware aging score; For the immunoassay analyzers in the medium frequency group, the comprehensive operating pressure is analyzed by the following formula: ； Among them, Health2 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the medium frequency use group; For the immunoassay analyzers in the low-frequency use group, the comprehensive operating pressure is analyzed by the following formula: ； Among them, Health3 is the comprehensive operating pressure analysis value of the immunoassay analyzer in the low-frequency use group.

8. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 7, characterized in that: According to the comprehensive operation pressure analysis results, the data storage and computing tasks of the edge nodes are allocated, including the following steps: According to the results of the comprehensive operation pressure analysis, if the comprehensive operation pressure analysis value is greater than the first threshold, the nearest node is selected from the nodes whose comprehensive operation pressure analysis value is less than the second threshold, 10% of the historical storage data is transferred to the selected node, and the last 10% of the tasks in the pending task queue are transferred to the cloud processing platform until the comprehensive operation pressure analysis value of the corresponding node is no greater than the first threshold.

9. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 1, characterized in that: The cloud processing platform performs real-time analysis of the test configuration parameters, experimental signal data, and quality control data of different types of reagent cards during experiments, and performs reliability analysis on the experimental data, including the following steps: The cloud processing platform obtains reagent cards of different categories, detection configuration parameters for different target types, experimental signal data, historical data of normal detection of quality control data, and historical data when detection failures occur, and adds normal detection and fault detection labels to the historical data; Establish a multimodal deep confidence assessment model, and use different classifications of reagent cards to perform target type detection on bacterial targets, viral targets, toxin targets, and parasite targets to train the corresponding model; The cloud processing platform obtains the detection configuration parameters, experimental signal data and quality control data of different categories of reagent cards in real time; Through the trained model, the reliability analysis of the experimental data is performed to identify the data of the corresponding target type detection performed by the corresponding classified reagent card, which is normal detection or fault detection data. If fault detection data appears, the data during the experiment is unreliable and is fault detection data. Otherwise, the data during the experiment is reliable.

10. A data management method for a multi-target biohazard factor immunoassay analyzer according to claim 9, characterized in that: Establishing a multimodal deep confidence assessment model includes the following steps: Construct a multimodal deep belief network and set static parameter branch, dynamic signal branch and quality control data branch in the input layer; The static parameter branch is used to input the detection configuration parameters and is set as a fully connected layer. The dynamic signal branch is used to input the experimental signal data and is set as an LSTM network layer. The LSTM network layer processes the time series signal of the experimental signal data, and the sliding window length is set to 30 sampling cycles. The quality control data branch is used to input the quality control data and is set as a time series convolution layer. The cumulative effect of quality control deviation is captured by constructing a time series convolution network, and the convolution kernel width matches the calibration cycle. The fusion layer sets up a cross-modal attention mechanism to weightedly fuse the features of each branch; The output layer uses the Sigmoid activation function, and the output data is the classification of normal detection and fault detection.

Images