Multi-center collaborative execution project quality monitoring system based on database

Abstract

A multi-center collaborative execution project quality monitoring system based on a database relates to the technical field of project quality monitoring, including a monitoring center, wherein the monitoring center is communicatively connected with a data acquisition module, a quality detection module, a project chain module, a multi-center collaborative module, a timing estimation module, a comprehensive collaborative estimation module and a maintenance module; the characteristic data set of each project subsequence in the medical experiment process is collected; data quality detection is performed on the characteristic data set of the blockchain node; data chain operation is performed on the characteristic data set of each project subsequence; the progress influence coefficient of other project subsequences under the condition that each project subsequence has abnormal reporting conditions of various types of quality monitoring indicators; the comprehensive characteristic data set of each project subsequence is obtained; and detection and correction operations are performed according to the comprehensive characteristic data set of each project subsequence to improve the accuracy and real-time performance of project quality monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of project quality monitoring, in particular to a multi-center collaborative execution project quality monitoring system based on a database. Background Art

[0002] The prior art CN113487291A "Project Quality Automation Monitoring System" includes an RTC data processing unit, a data acquisition unit, a data cleaning unit, a data storage unit and a data display unit. The RTC data processing unit is used to provide project data to the data acquisition unit. The data acquisition unit stores key information in the project data in a local database. The data cleaning unit obtains information from the local database, cleans it into classified data, and stores it in the data storage unit. The data storage unit contains the collected data in the local database and the cleaned data. The data display unit accesses the site deployed on the server through a browser to display the dashboard of each project. The dashboard displays and analyzes historical data.

[0003] The quality monitoring indicators of the medical experiment process are key factors in ensuring that the test results of clinical laboratories are accurate, reliable and effective. These indicators cover all aspects of the experiment, from sample collection to result reporting, and aim to continuously improve the laboratory's workflow and performance. When the quality monitoring indicators of a certain link in the experiment are abnormal, it may cause obvious or potential experimental risks to other links, reducing the laboratory's workflow and performance. At the same time, multi-center collaborative execution of project quality monitoring based on databases is usually used to manage and monitor the quality of projects in multiple locations or centers. However, such systems may encounter a series of problems in actual applications, including multi-center data transmission and storage may face data leakage and security risks, especially when sensitive information is involved. At the same time, the quality of data input by each center is uneven, and there is a lack of a unified data verification and analysis mechanism, which affects the overall quality of the data and leads to incorrect quality assessment and decision-making. Summary of the invention

[0004] In order to solve the above technical problems, the object of the present invention is to provide a multi-center collaborative execution project quality monitoring system based on a database, including a monitoring center, wherein the monitoring center is communicatively connected with a data acquisition module, a quality detection module, a project uplink module, a multi-center collaborative module, a timing estimation module, a comprehensive collaborative estimation module and a maintenance module;

[0005] The data acquisition module is used to collect feature data sets of each item subsequence in the medical experiment process;

[0006] The quality detection module is used to perform data quality detection on the characteristic data set of the blockchain node and generate a quality detection report;

[0007] The project on-chain module is used to perform data on-chain operations on the feature data sets of each project subsequence, and update the feature data sets and quality inspection reports to the local copy databases of all blockchain nodes;

[0008] The multi-center collaboration module is used to obtain the progress impact coefficient of other project subsequences under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators;

[0009] The time series estimation module is used to obtain the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence;

[0010] The comprehensive collaborative estimation module is used to output a comprehensive feature data set of each project subsequence;

[0011] The inspection and repair module is used to perform detection and correction operations based on the comprehensive feature data set of each item subsequence.

[0012] Furthermore, the process of the data collection module collecting the feature data set of each item subsequence in the medical experiment process includes:

[0013] Obtain the project process information of medical experiments, preliminarily simplify the project process information, remove redundant project processes, build a project process sequence according to the simplified project process information, extract each experimental process in the project process sequence, use the experimental process as a segment node of the project process sequence, and divide the project process sequence into a number of project subsequences;

[0014] The experimental content of each project subsequence is extracted, and several types of corresponding quality monitoring indicators are matched according to the experimental content. An Internet of Things node is deployed in each project subsequence. The Internet of Things node is used to collect corresponding feature data sets according to several types of quality monitoring indicators of the project subsequence and mark the monitoring time and the monitoring period.

[0015] Furthermore, the quality detection module performs data quality detection on the characteristic data set of the blockchain node, and the process of generating a quality detection report includes:

[0016] Obtain standard numerical values ​​corresponding to several types of quality monitoring indicators of the project subsequence to which the Internet of Things node that is communicatively connected to the blockchain node belongs, extract the numerical time series sequence corresponding to each type of quality monitoring indicator in the feature data set, compare the numerical time series sequence with the corresponding standard numerical value, obtain the quality deviation rate of each type of quality monitoring indicator in the feature data set, preset a quality deviation rate threshold, and if there is a quality deviation rate of a certain type of quality monitoring indicator greater than the quality deviation rate threshold, generate an abnormal report of the certain type of quality monitoring indicator of the project subsequence; if the quality deviation rates of each type of quality monitoring indicator are less than or equal to the quality deviation rate threshold, generate a normal report of the project subsequence.

[0017] Furthermore, the project on-chain module performs a data on-chain operation on the feature data set of each project subsequence, and the process of updating the feature data set and the quality inspection report to the local copy database of all blockchain nodes includes:

[0018] A project chain module is constructed based on blockchain technology. A number of blockchain nodes are arranged in the project chain module, and the blockchain nodes are interconnected to form a blockchain network. The blockchain nodes and the project subsequences are communicated and connected through the Internet of Things nodes.

[0019] The IoT node encrypts the collected feature data set and uploads it to the blockchain node. After the blockchain node decrypts the encrypted feature data set, the quality inspection module performs data quality inspection on the feature data set of the blockchain node, generates a quality inspection report, packages the feature data set and the quality inspection report into a new block and broadcasts it to the entire blockchain network. Other blockchain nodes in the blockchain network verify the correctness of the block data of the new block.

[0020] After the new block is verified, it is added to the end of the blockchain, and the feature data set and quality inspection report in the new block are updated to the local copy database of all blockchain nodes.

[0021] Furthermore, the process of the multi-center collaboration module obtaining the progress impact coefficients of other project subsequences under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators includes:

[0022] When the local replica database of the blockchain node receives the feature data set and quality inspection report of the project subsequence of other blockchain nodes, if there is an abnormal report of a certain type of quality monitoring indicator in the quality inspection report, the historical monitoring period corresponding to the abnormal report of the certain type of quality monitoring indicator in the project subsequence of the other blockchain node stored in the local replica database and the feature data set in the historical monitoring period are extracted, the historical monitoring period is marked as a historical collaboration period, the feature data set is marked as a collaborative feature data set, and the feature similarity between each collaborative feature data set and the feature data set of the project subsequence of other blockchain nodes at the current moment is obtained. The weight coefficient of the historical collaboration period corresponding to each collaborative feature data set is obtained according to the feature similarity, and at the same time, the quality deviation rate of each type of quality monitoring indicator of the blockchain node in each historical collaboration period is extracted. According to the quality deviation rate of the blockchain node in each historical collaboration period and the weight coefficient of each historical collaboration period, the progress impact coefficient of the project subsequence corresponding to the blockchain node is obtained under the condition that the project subsequence of the other blockchain node has an abnormal report of the certain type of quality monitoring indicator. Similarly, the progress impact coefficient of other project subsequences under the condition that each project subsequence has an abnormal report of each type of quality monitoring indicator at the current moment is obtained.

[0023] Furthermore, the process of the time series estimation module obtaining the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence includes:

[0024] Build a feature data set prediction model based on deep learning, and use the feature data sets in the local copy database of each blockchain node as test sets and training sets;

[0025] Input the training set into the feature data set prediction model for training until the loss function training is stable, and save the model parameters, test the feature data set prediction model with the test set until it meets the preset requirements, and output the feature data set prediction model;

[0026] The feature data set of each project subsequence collected during the current monitoring cycle is obtained and input into the feature data set prediction model. According to the output layer of the feature data set prediction model, the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence is obtained.

[0027] Furthermore, the process of the comprehensive collaborative estimation module outputting the comprehensive feature data set of each project subsequence includes:

[0028] The blockchain nodes connected to the IoT nodes of each project subsequence are used as nodes, and the connection relationship between each blockchain node is used as the connection relationship between nodes to build a project topology diagram;

[0029] Perform data quality inspection on the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence, generate a quality inspection report for each project subsequence, if there is an abnormal report of a certain type of quality monitoring indicator of a certain project subsequence in the quality inspection report, obtain the progress impact coefficient of the certain project subsequence on other project subsequences at the current moment, and set weight labels between each node according to the progress impact coefficient of the certain project subsequence on other project subsequences, and input the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence into the corresponding node in the project topology map;

[0030] A comprehensive feature dataset prediction model is constructed based on the graph attention network to learn and represent the project topology graph and output the comprehensive feature dataset of each project subsequence.

[0031] Furthermore, the process of the inspection module performing the inspection and correction operation according to the comprehensive feature data set of each item subsequence includes:

[0032] Performing data quality detection on the comprehensive feature data set of each project subsequence, obtaining a quality detection report of each project subsequence, obtaining several project subsequences with abnormal reports of a certain type of quality monitoring indicator, marking the several project subsequences as abnormal project subsequences, extracting the quality deviation rate of each type of quality monitoring indicator in the comprehensive feature data set of each abnormal project subsequence, and the progress impact coefficient of each abnormal project subsequence on other project subsequences, and obtaining the comprehensive abnormal coefficient of each abnormal project subsequence according to the quality deviation rate of each type of quality monitoring indicator and the progress impact coefficient of each abnormal project subsequence;

[0033] The abnormal item subsequences are sorted according to the comprehensive abnormal coefficient, and the abnormal item subsequences are selected in turn for detection and correction operations according to the sorting results.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. Upload the feature data sets of each project subsequence to the blockchain, and update the feature data sets and quality inspection reports to the local copy databases of all blockchain nodes. By taking advantage of the immutability and decentralization of blockchain, the data security and trustworthiness of multi-center collaborative project quality monitoring can be increased.

[0036] 2. Under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators, the progress impact coefficients of other project subsequences and the estimated numerical time series sequence of various types of quality monitoring indicators in the feature data set of each project subsequence are obtained. According to the estimated numerical time series sequence and the progress impact coefficient, the comprehensive feature data set of each project subsequence is output, and the impact trend and impact range of each project subsequence when an abnormality occurs are analyzed to provide a basis for subsequent detection and correction operations, thereby improving the accuracy and real-time performance of project quality monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a schematic diagram of a database-based multi-center collaborative execution project quality monitoring system according to an embodiment of the present application. DETAILED DESCRIPTION

[0038] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0039] like Figure 1 As shown, a multi-center collaborative execution project quality monitoring system based on a database includes a monitoring center, and the monitoring center is communicatively connected with a data acquisition module, a quality detection module, a project uplink module, a multi-center collaborative module, a timing estimation module, a comprehensive collaborative estimation module and a maintenance module;

[0040] The data acquisition module is used to collect feature data sets of each item subsequence in the medical experiment process;

[0041] The quality detection module is used to perform data quality detection on the characteristic data set of the blockchain node and generate a quality detection report;

[0042] The project on-chain module is used to perform data on-chain operations on the feature data sets of each project subsequence, and update the feature data sets and quality inspection reports to the local copy databases of all blockchain nodes;

[0043] The multi-center collaboration module is used to obtain the progress impact coefficient of other project subsequences under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators;

[0044] The time series estimation module is used to obtain the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence;

[0045] The comprehensive collaborative estimation module is used to output a comprehensive feature data set of each project subsequence;

[0046] The inspection and repair module is used to perform detection and correction operations based on the comprehensive feature data set of each item subsequence.

[0047] It should be further explained that, in the specific implementation process, the process of the data acquisition module acquiring the feature data set of each item subsequence in the medical experiment process includes:

[0048] Obtain the project process information of the medical experiment, preliminarily simplify the project process information, remove redundant project processes, and construct a project process sequence based on the simplified project process information. The project process sequence includes:

[0049] In the planning and design phase, goal setting: define the purpose of the trial, research questions and hypotheses; experimental design: determine the type of trial (such as a randomized controlled trial), sample size, participant inclusion and exclusion criteria, control group setting, etc.; plan formulation: write a detailed experimental plan, including data collection methods, statistical analysis plan, etc.;

[0050] Regulatory and ethical approval, submission for approval: submit the trial plan and related documents to the ethics committee (IRB or EC) to ensure that the trial design complies with ethical standards; Regulatory filing: apply for clinical trial permission from the health regulatory authorities of relevant countries or regions;

[0051] In the preparation stage, select the trial center: select qualified medical institutions as trial sites; train researchers: ensure that all participants understand the trial plan and operation procedures; prepare equipment: ensure the availability and compliance of medical devices used in the trial;

[0052] Participant recruitment and screening, publicity and recruitment: finding eligible participants and evaluating volunteers according to inclusion and exclusion criteria; trial conduct, use and monitoring: using medical devices according to the trial protocol, regularly collecting data and monitoring effects;

[0053] Data collection and management, data recording: accurately record all data generated during the test;

[0054] Extract the experimental processes in the project process sequence. It should be noted that the experimental processes of medical experiments include:

[0055] The doctor issues a laboratory test request form. The doctor issues a laboratory test request form based on the patient's symptoms, medical history and clinical manifestations, indicating the items that need to be tested;

[0056] Appointment and preparation: patients make appointments for test times according to the test application form. Some tests (such as blood sugar and liver function) require patients to be performed under specific conditions (such as fasting);

[0057] Sample collection: trained medical personnel collect samples according to standard procedures. Common samples include blood, urine, saliva, feces, sputum, cerebrospinal fluid or tissue samples;

[0058] Sample processing: In the laboratory, samples are pre-processed, such as centrifugation to separate serum or plasma, dilution, filtration or preparation into slices; samples are assigned to the corresponding laboratory departments, such as biochemistry, immunology, microbiology or molecular biology departments;

[0059] Inspection and analysis: laboratory technicians use specialized equipment and reagents to conduct tests. Depending on the test items, they may use a variety of methods such as chemical analysis, immunoassay, microbial culture, PCR (polymerase chain reaction), etc.

[0060] Data recording, the test results are recorded, the technician or laboratory doctor interprets the test results, compares them with the reference range, determines whether there are any abnormalities, and the results are compiled into a report format, including the test items, values, units, reference ranges, and any necessary comments or suggestions;

[0061] Send the results, and the report is sent to the ordering physician electronically or in paper form or directly to the patient, depending on the hospital policy and the patient's consent; use the experimental process as a segment node of the project process sequence, and divide the project process sequence into several project subsequences;

[0062] Extract the experimental content of each project subsequence, match the corresponding several types of quality monitoring indicators according to the experimental content, and deploy an Internet of Things node in each project subsequence. The Internet of Things node is used to collect the corresponding feature data set according to the several types of quality monitoring indicators of the project subsequence and mark the monitoring time and the monitoring period;

[0063] It should be further explained that the quality monitoring indicators of the medical experiment process are key factors in ensuring that the test results of clinical laboratories are accurate, reliable and effective. These indicators cover all aspects of the experiment, from sample collection to result reporting, and aim to continuously improve the laboratory's workflow and performance. The following are some of the main quality monitoring indicators of the medical experiment process:

[0064] Sample type error rate: the rate at which the sample type does not match the required test;

[0065] Sample container error rate: how often the wrong type of container was used to collect the sample;

[0066] Sample collection error rate: the rate at which insufficient or excessive sample size is collected;

[0067] Specimen processing time: the time from collection to the start of processing to ensure sample stability;

[0068] In-house quality control: Use control samples of known concentration to monitor the stability of the analytical process;

[0069] External quality assessment: Comparison of laboratory results with other laboratories to assess inter-laboratory comparability and accuracy;

[0070] Analyze the number of equipment failures: monitor the normal operation time and failure frequency of equipment;

[0071] Imprecision: The variability of a measurement, including repeatability and reproducibility;

[0072] Bias: The difference between a test result and the true value;

[0073] Indoor quality control out-of-control rate: the frequency of indoor parameter control out-of-control;

[0074] Indoor quality control out-of-control handling rate: the rate at which out-of-control situations are handled promptly and correctly;

[0075] Result reporting time: the time from the completion of sample analysis to the formal reporting of the results to the doctor;

[0076] Result review rate: how often results need to be reviewed or retested;

[0077] Result error rate: the rate at which errors are reported in the results;

[0078] Laboratory safety and environment: monitor whether the laboratory's working environment meets safety standards;

[0079] Equipment performance: regular calibration and operating data of equipment;

[0080] Human capacity: training and qualification of staff;

[0081] Document control system: effectiveness of updating, distribution and archiving of documents;

[0082] Blood culture contamination rate: the frequency of occurrence of non-target microorganisms in blood cultures;

[0083] Anticoagulation specimen agglutination rate: the coagulation rate of anticoagulated blood specimens, etc.

[0084] It should be further explained that, in the specific implementation process, the quality detection module performs data quality detection on the characteristic data set of the blockchain node, and the process of generating a quality detection report includes:

[0085] Obtain standard numerical values ​​corresponding to several types of quality monitoring indicators of the project subsequence to which the Internet of Things node that is communicatively connected to the blockchain node belongs, extract the numerical time series sequence corresponding to each type of quality monitoring indicator in the feature data set, compare the numerical time series sequence with the corresponding standard numerical value, obtain the quality deviation rate of each type of quality monitoring indicator in the feature data set, preset a quality deviation rate threshold, and if there is a quality deviation rate of a certain type of quality monitoring indicator greater than the quality deviation rate threshold, generate an abnormal report of the certain type of quality monitoring indicator of the project subsequence; if the quality deviation rates of each type of quality monitoring indicator are less than or equal to the quality deviation rate threshold, generate a normal report of the project subsequence.

[0086] It should be further explained that, in the specific implementation process, the calculation formula for obtaining the quality deviation rate of each type of quality monitoring indicator in the feature data set is:

[0087]

[0088] in, represents the quality deviation rate of the i-th type quality monitoring indicator, represents the value of the quality monitoring indicator of type i at the tth moment, The standard value of the quality monitoring indicator of type i, and n represents the total number of moments.

[0089] It should be further explained that, in the specific implementation process, the project on-chain module performs a data on-chain operation on the feature data set of each project subsequence, and the process of updating the feature data set and the quality inspection report to the local copy database of all blockchain nodes includes:

[0090] A project chain module is constructed based on blockchain technology. A number of blockchain nodes are arranged in the project chain module, and the blockchain nodes are interconnected to form a blockchain network. The blockchain nodes and the project subsequences are communicated and connected through the Internet of Things nodes.

[0091] The IoT node encrypts the collected feature data set and uploads it to the blockchain node. After the blockchain node decrypts the encrypted feature data set, the quality detection module performs data quality detection on the feature data set of the blockchain node and generates a quality detection report. The quality detection report includes a normal report of each type of quality monitoring indicator of each project subsequence and an abnormal report of each type of quality monitoring indicator. The feature data set and the quality detection report are packaged into a new block and broadcasted to the entire blockchain network. Other blockchain nodes in the blockchain network verify the correctness of the block data of the new block.

[0092] After the new block is verified, it is added to the end of the blockchain, and the feature data set and quality inspection report in the new block are updated to the local copy database of all blockchain nodes.

[0093] It should be further explained that, in the specific implementation process, the process of other blockchain nodes in the blockchain network verifying the correctness of the block data of the new block includes:

[0094] Obtain components of a block header in a new block, including a version number, a hash value of a previous block, a timestamp, a difficulty target, and a random number; obtain a hash value of the new block according to the components of the block header; obtain the current difficulty target of the blockchain network; if the obtained hash value is less than the current difficulty target of the blockchain network, the new block passes the block data correctness verification.

[0095] It should be further explained that, in the specific implementation process, the multi-center collaborative module obtains the progress impact coefficient of other project subsequences under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators, and includes:

[0096] When the local replica database of the blockchain node receives the feature data set and quality inspection report of the project subsequence of other blockchain nodes, if there is an abnormal report of a certain type of quality monitoring indicator in the quality inspection report, the historical monitoring period corresponding to the abnormal report of the certain type of quality monitoring indicator in the project subsequence of the other blockchain node stored in the local replica database and the feature data set in the historical monitoring period are extracted, the historical monitoring period is marked as a historical collaboration period, the feature data set is marked as a collaborative feature data set, and the feature similarity between each collaborative feature data set and the feature data set of the project subsequence of other blockchain nodes at the current moment is obtained. The weight coefficient of the historical collaboration period corresponding to each collaborative feature data set is obtained according to the feature similarity, and at the same time, the quality deviation rate of each type of quality monitoring indicator of the blockchain node in each historical collaboration period is extracted. According to the quality deviation rate of the blockchain node in each historical collaboration period and the weight coefficient of each historical collaboration period, the progress impact coefficient of the project subsequence corresponding to the blockchain node is obtained under the condition that the project subsequence of the other blockchain node has an abnormal report of the certain type of quality monitoring indicator. Similarly, the progress impact coefficient of other project subsequences under the condition that each project subsequence has an abnormal report of each type of quality monitoring indicator at the current moment is obtained.

[0097] It should be further explained that, in the specific implementation process, the calculation formula for obtaining the feature similarity between the collaborative feature data set and the feature data set is:

[0098]

[0099] in, represents the feature similarity between the collaborative feature dataset Z and the feature dataset R, represents the value of the j-th type quality monitoring indicator in the collaborative feature dataset Z at the t-th moment, represents the standard value of the j-th type of quality monitoring indicator in the collaborative feature dataset Z, represents the value of the k-th type of quality monitoring indicator in the feature data set R at the t-th moment, represents the standard value of the k-th type of quality monitoring indicator in the feature data set R, It represents the total number of types of quality monitoring indicators in the collaborative feature dataset Z and the feature dataset R, and n2 represents the total number of moments.

[0100] It should be further explained that, in the specific implementation process, according to the quality deviation rate of the blockchain node in each historical collaboration cycle and the weight coefficient of each historical collaboration cycle, the calculation formula for the progress impact coefficient of the project subsequence corresponding to the blockchain node under the condition that the project subsequence of the other blockchain node has an abnormal report of the certain type of quality monitoring indicator is obtained:

[0101]

[0102] in, represents the progress impact coefficient, represents the quality deviation rate in the mth historical collaboration cycle, It should be further explained that, in the specific implementation process, the process of the time series estimation module obtaining the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence includes:

[0103] Build a feature data set prediction model based on deep learning, and use the feature data sets in the local copy database of each blockchain node as test sets and training sets;

[0104] Input the training set into the feature data set prediction model for training until the loss function training is stable, and save the model parameters, test the feature data set prediction model with the test set until it meets the preset requirements, and output the feature data set prediction model;

[0105] The feature data set of each project subsequence collected during the current monitoring cycle is obtained and input into the feature data set prediction model. According to the output layer of the feature data set prediction model, the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence is obtained.

[0106] It should be further explained that, in a specific implementation process, the process of the comprehensive collaborative estimation module outputting the comprehensive feature data set of each project subsequence includes:

[0107] The blockchain nodes connected to the IoT nodes of each project subsequence are used as nodes, and the connection relationship between each blockchain node is used as the connection relationship between nodes to build a project topology diagram;

[0108] Perform data quality inspection on the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence, generate a quality inspection report for each project subsequence, if there is an abnormal report of a certain type of quality monitoring indicator of a certain project subsequence in the quality inspection report, obtain the progress impact coefficient of the certain project subsequence on other project subsequences at the current moment, and set weight labels between each node according to the progress impact coefficient of the certain project subsequence on other project subsequences, and input the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence into the corresponding node in the project topology map;

[0109] A comprehensive feature dataset prediction model is constructed based on the graph attention network to learn and represent the project topology graph and output the comprehensive feature dataset of each project subsequence.

[0110] It should be further explained that in the specific implementation process, the comprehensive feature dataset prediction model is constructed based on the graph attention network, the project topology graph is learned and represented, and the specific process of outputting the comprehensive feature dataset of each project subsequence is as follows:

[0111] The project topology graph is learned through the graph attention network, and the attention weights of other nodes on the target node are obtained through the attention mechanism. The attention weights are distributed to each adjacent node, and the aggregated representation of features is generated by neighbor aggregation, and the comprehensive feature data set of each project subsequence is output. It should be noted that the process of obtaining the attention weight is: ;

[0112] in, represents the attention weight matrix corresponding to node O, Represents the progress influence coefficient matrix of node O on other nodes, = , n4 Represents the fully connected layer of the graph attention network, and the superscript T indicates the transposition of the matrix;

[0113] It should be noted that the process of using neighbor aggregation to generate the aggregate representation of features is:

[0114] Get a pair of nodes (c, h) in the project topology graph, get the initial representation graph of nodes c and h, and the fully connected layer of the attention network. The superscript T indicates that the matrix is ​​transposed.

[0115] It should be noted that the process of using neighbor aggregation to generate the aggregate representation of features is:

[0116] Get a pair of nodes (c, h) in the project topology graph and get the initial values ​​of nodes c and h. ;

[0117] in, represents the activation function, G represents the parameter matrix of feature transformation, represents the attention weight of node c to node h,

[0118] According to the above formula, the vector representation of node c is generated to represent the attention weight of node c to node h,

[0119] According to the above formula, the vector representation of node c is generated Then the next pair of nodes is represented by vectors, and so on. After all nodes in the project topology graph are represented, the inner product of the node vectors is calculated to obtain the comprehensive feature data set of each project subsequence.

[0120] It should be further explained that, in the specific implementation process, the process of the inspection module performing the detection and correction operation according to the comprehensive feature data set of each item subsequence includes:

[0121] Performing data quality detection on the comprehensive feature data set of each project subsequence, obtaining the quality detection report of each project subsequence, obtaining several project subsequences with abnormal reports of a certain type of quality monitoring indicator, it should be noted that the abnormal report of a certain type of quality monitoring indicator of the project subsequence includes the abnormal report of a certain type of quality monitoring indicator in the feature data set of the project subsequence and the abnormal reports of several types of quality monitoring indicators in the feature data set of the project subsequence, marking the several project subsequences as abnormal project subsequences, extracting the quality deviation rate of each type of quality monitoring indicator in the comprehensive feature data set of each abnormal project subsequence, and the progress impact coefficient of each abnormal project subsequence on other project subsequences, the project subsequences include abnormal project subsequences, according to the quality deviation rate of each type of quality monitoring indicator and the progress impact coefficient of each abnormal project subsequence, obtaining the comprehensive abnormal coefficient of each abnormal project subsequence;

[0122] The abnormal item subsequences are sorted according to the comprehensive abnormality coefficient, and the abnormal item subsequences are selected in turn according to the sorting results for detection and correction operations. The abnormal item subsequence with the largest comprehensive abnormality coefficient is first subjected to detection and correction operations. The detection and correction operations include detecting the cause of the abnormality of the detection item and increasing the detection frequency of the Internet of Things node.

[0123] It should be further explained that, in the specific implementation process, according to the quality deviation rate and progress impact coefficient of each type of quality monitoring indicator, the calculation formula for obtaining the comprehensive abnormality coefficient is:

[0124] ;

[0125] Among them, zx represents the comprehensive abnormal coefficient, represents the conversion factor, , Represents the weight factor.

[0126] 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 multi-center collaborative project quality monitoring system based on a database, characterized by: It includes a monitoring center, which is communicatively connected with a data acquisition module, a quality detection module, a project uplink module, a multi-center collaboration module, a time sequence estimation module, a comprehensive collaborative estimation module and a maintenance module; The data acquisition module is used to collect feature data sets of each item subsequence in the medical experiment process; The quality detection module is used to perform data quality detection on the characteristic data set of the blockchain node and generate a quality detection report; The project on-chain module is used to perform data on-chain operations on the feature data sets of each project subsequence, and update the feature data sets and quality inspection reports to the local copy databases of all blockchain nodes: A project chain module is constructed based on blockchain technology. A number of blockchain nodes are arranged in the project chain module, and the blockchain nodes are interconnected to form a blockchain network. The blockchain nodes and the project subsequences are communicated and connected through the Internet of Things nodes. The IoT node encrypts the collected feature data set and uploads it to the blockchain node. After the blockchain node decrypts the encrypted feature data set, the quality inspection module performs data quality inspection on the feature data set of the blockchain node, generates a quality inspection report, packages the feature data set and the quality inspection report into a new block and broadcasts it to the entire blockchain network. Other blockchain nodes in the blockchain network verify the correctness of the block data of the new block. After the new block is verified, the new block is added to the end of the blockchain, and the feature data set and quality inspection report in the new block are updated to the local copy database of all blockchain nodes; The multi-center collaboration module is used to obtain the progress impact coefficient of other project subsequences under the condition that each project subsequence has abnormal reports of various types of quality monitoring indicators: When the local replica database of the blockchain node receives the feature data set and quality inspection report of the project subsequence of other blockchain nodes, if there is an abnormal report of a certain type of quality monitoring indicator in the quality inspection report, the historical monitoring period corresponding to the abnormal report of the certain type of quality monitoring indicator in the project subsequence of the other blockchain node stored in the local replica database and the feature data set in the historical monitoring period are extracted, the historical monitoring period is marked as a historical collaboration period, the feature data set is marked as a collaborative feature data set, and the feature similarity between each collaborative feature data set and the feature data set of the project subsequence of other blockchain nodes at the current moment is obtained, and the weight coefficient of the historical collaboration period corresponding to each collaborative feature data set is obtained according to the feature similarity, and at the same time, the quality deviation rate of each type of quality monitoring indicator of the blockchain node in each historical collaboration period is extracted, and according to the quality deviation rate of the blockchain node in each historical collaboration period and the weight coefficient of each historical collaboration period, the progress impact coefficient of the project subsequence corresponding to the blockchain node under the condition that the project subsequence of the other blockchain node has an abnormal report of the certain type of quality monitoring indicator is obtained, and so on, the progress impact coefficient of other project subsequences under the condition that each project subsequence has an abnormal report of each type of quality monitoring indicator at the current moment is obtained; The time series estimation module is used to obtain the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence: Build a feature data set prediction model based on deep learning, and use the feature data sets in the local copy database of each blockchain node as test sets and training sets; Input the training set into the feature data set prediction model for training until the loss function training is stable, and save the model parameters, test the feature data set prediction model with the test set until it meets the preset requirements, and output the feature data set prediction model; Obtain the feature data set of each project subsequence collected in the current monitoring cycle and input it into the feature data set prediction model, and obtain the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence according to the output layer of the feature data set prediction model; The comprehensive collaborative estimation module is used to output a comprehensive feature data set of each project subsequence; The inspection and repair module is used to perform detection and correction operations based on the comprehensive feature data set of each item subsequence.

2. The database-based multi-center collaborative execution project quality monitoring system according to claim 1 is characterized in that: The process of the data collection module collecting the feature data set of each item subsequence in the medical experiment process includes: Obtain the project process information of medical experiments, preliminarily simplify the project process information, remove redundant project processes, construct a project process sequence according to the simplified project process information, extract each experimental process in the project process sequence, use the experimental process as a segment node of the project process sequence, and divide the project process sequence into a number of project subsequences; The experimental content of each project subsequence is extracted, and several types of corresponding quality monitoring indicators are matched according to the experimental content. An Internet of Things node is deployed in each project subsequence. The Internet of Things node is used to collect corresponding feature data sets according to several types of quality monitoring indicators of the project subsequence and mark the monitoring time and the monitoring period.

3. The database-based multi-center collaborative execution project quality monitoring system according to claim 2 is characterized in that: The quality detection module performs data quality detection on the characteristic data set of the blockchain node, and the process of generating a quality detection report includes: Obtain standard numerical values ​​corresponding to several types of quality monitoring indicators of the project subsequence to which the Internet of Things node that is communicatively connected to the blockchain node belongs, extract the numerical time series sequence corresponding to each type of quality monitoring indicator in the feature data set, compare the numerical time series sequence with the corresponding standard numerical value, obtain the quality deviation rate of each type of quality monitoring indicator in the feature data set, preset a quality deviation rate threshold, and if there is a quality deviation rate of a certain type of quality monitoring indicator greater than the quality deviation rate threshold, generate an abnormal report of the certain type of quality monitoring indicator of the project subsequence; if the quality deviation rates of each type of quality monitoring indicator are less than or equal to the quality deviation rate threshold, generate a normal report of the project subsequence.

4. The database-based multi-center collaborative execution project quality monitoring system according to claim 3 is characterized in that: The process of the comprehensive collaborative estimation module outputting the comprehensive feature data set of each project subsequence includes: The blockchain nodes connected to the IoT nodes of each project subsequence are used as nodes, and the connection relationship between each blockchain node is used as the connection relationship between nodes to build a project topology diagram; Perform data quality inspection on the estimated numerical time series of each type of quality monitoring indicator in the feature data set of each project subsequence, generate a quality inspection report for each project subsequence, if there is an abnormal report of a certain type of quality monitoring indicator of a certain project subsequence in the quality inspection report, obtain the progress impact coefficient of the certain project subsequence on other project subsequences at the current moment, and set weight labels between each node according to the progress impact coefficient of the certain project subsequence on other project subsequences, and input the estimated numerical time series of each type of quality monitoring indicator in the feature data set of the remaining time period of the current monitoring cycle of each project subsequence into the corresponding node in the project topology map; A comprehensive feature dataset prediction model is constructed based on the graph attention network to learn and represent the project topology graph and output the comprehensive feature dataset of each project subsequence.

5. The database-based multi-center collaborative execution project quality monitoring system according to claim 4 is characterized in that: The process of the inspection and correction module performing the inspection and correction operation according to the comprehensive feature data set of each item subsequence includes: Performing data quality detection on the comprehensive feature data set of each project subsequence, obtaining a quality detection report of each project subsequence, obtaining several project subsequences with abnormal reports of a certain type of quality monitoring indicator, marking the several project subsequences as abnormal project subsequences, extracting the quality deviation rate of each type of quality monitoring indicator in the comprehensive feature data set of each abnormal project subsequence, and the progress impact coefficient of each abnormal project subsequence on other project subsequences, and obtaining the comprehensive abnormal coefficient of each abnormal project subsequence according to the quality deviation rate of each type of quality monitoring indicator and the progress impact coefficient of each abnormal project subsequence; The abnormal item subsequences are sorted according to the comprehensive abnormal coefficient, and the abnormal item subsequences are selected in turn for detection and correction operations according to the sorting results.

Images