CN-122022563-A - LIMS-based full-class detection data integration and quality traceability management system

CN122022563ACN 122022563 ACN122022563 ACN 122022563ACN-122022563-A

Abstract

The invention relates to the technical field of laboratory information management systems, and discloses a LIMS-based full-class detection data integration and quality traceability management system, which comprises a data modeling module, a data management module and a data management module, wherein the data modeling module automatically generates domain ontology extensions by adopting a method of combining ontology modeling and natural language processing; the system comprises a data processing module, a block chain tracing module, a quality prediction early warning module, a cross-system integration module and a data management platform, wherein the data processing module acquires original data of multi-source heterogeneous instrument data, performs format recognition and semantic mapping on the original data by adopting a unified model of all-class detection data, the block chain tracing module stores tracing standardized detection data in a classified mode, establishes a layered architecture, constructs a quality influence factor library, generates a prediction result based on the quality influence factor library, and adopts a containerization technology and an event tracing architecture to obtain the cross-system integrated data management platform.

Inventors

WANG YANBIN
DU YUANFENG
WANG YINFENG
YANG HONGFU
SHENG QINGQUAN
ZHANG QINGWEN
CUI SHUQIANG
Xin Siliang
SUN FANGFANG

Assignees

山东滨农科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260114

Claims (10)

1. LIMS-based full-class detection data integration and quality traceability management system is characterized by comprising: the data modeling module is used for automatically generating domain ontology expansion by adopting a method combining ontology modeling and natural language processing based on the detection standard specification document to obtain a unified model of all-class detection data; the data processing module is used for acquiring the original data of the multi-source heterogeneous instrument data, and carrying out format identification and semantic mapping on the original data by adopting a unified model of all-class detection data to acquire standardized detection data; The block chain tracing module is used for classifying and storing the tracing standardized detection data by adopting data importance evaluation based on the standardized detection data, establishing hierarchical architecture balance performance and security, and obtaining a quality tracing data chain; The quality prediction early warning module is used for extracting historical detection data based on a quality traceability data chain, constructing a quality influence factor library, and generating a prediction result based on the quality influence factor library to obtain quality early warning information and improvement suggestions; And the cross-system integration module is used for obtaining a cross-system integrated data management console by adopting a containerization technology and an event tracing architecture based on the unified model of the whole class detection data, the standardized detection data, the quality tracing data chain and the quality early warning information.
2. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein the method of combining ontology modeling with natural language processing comprises: identifying technical terms in the detection field from the business requirement document by adopting a named entity identification algorithm, wherein the technical terms comprise sample types, detection projects, instrument equipment and quality indexes; Extracting semantic relations between entities by adopting a dependency syntax analysis technology, and identifying detection relations between samples and detection items, adopting relations between detection items and detection methods and using relations between detection methods and equipment; Based on the identified entities and relationships, domain ontology extension definitions are automatically generated to form an ontology model containing a concept hierarchy and attribute constraints.
3. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein said format identification comprises: Reading header information of an instrument original data file, and extracting a file magic number, a version identification and a coding format; Analyzing the data organization structure of the file, identifying byte order and data type of the binary format, and identifying separator and field order of the text format; And analyzing key field names in the file, and calculating by using word frequency inverse document frequency weights to obtain semantic feature vectors.
4. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 3, wherein said format identification further comprises: Carrying out cosine similarity calculation on the format feature vector of the file to be identified and the known format feature vector in the data format feature library; selecting a format with highest similarity and exceeding a set threshold as a matching result; and calling a corresponding data analyzer according to the matched format type to extract the detection data.
5. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein said data importance assessment comprises: Defining data importance evaluation criteria, including legal evidence value dimension, quality influence degree dimension and query frequency dimension; carrying out importance scoring on each type of data by adopting a weighted scoring model; The data is classified into two categories, namely, on-chain key data and off-chain auxiliary data according to importance scores.
6. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein the establishing of the hierarchical architecture comprises: the local fast chain is deployed inside the detection mechanism, and a consensus algorithm based on a leader election mechanism is adopted to support the writing of high-frequency traceability data; The alliance main chain is jointly maintained by a plurality of detection institutions as alliance nodes, and a Bayesian fault-tolerant consensus algorithm is adopted; the local fast chain submits the block abstract to the alliance main chain periodically to perform cross-chain anchoring, so that notarization of local traceable data is realized.
7. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein the quality-affecting factor library comprises five dimensions, and specifically comprises: The equipment dimension factors comprise equipment running time, calibration state scores and fault times; environmental dimension factors include laboratory temperature, relative humidity, and cleanliness levels; The personnel dimension factors comprise operator qualification level, detection task number and historical accuracy; the reagent dimension factors comprise reagent batch numbers, unsealing days and storage condition compliance; Sample dimension factors include sample matrix complexity and desired concentration levels.
8. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein the generating the prediction result based on the quality-affecting factor library comprises: Extracting a history detection record from a traceable data chain, and constructing a training data set containing each dimension factor value and quality index; Training a quality prediction model by adopting a gradient lifting decision tree algorithm, and constructing a plurality of decision tree fitting prediction residues through iteration; Calculating the contribution value of each factor to the predicted result by adopting a tree model shape algorithm, and expressing the predicted result as the sum of the baseline predicted value and the contribution value of each factor; and generating a visual interpretation graph to show the contribution size and direction of each factor, so as to obtain a prediction result.
9. The LIMS-based all-class inspection data integration and quality traceability management system according to claim 1, wherein the generating of the prediction result comprises: calculating the contribution value of each factor to the prediction result by adopting a tree model shape value algorithm; representing the predicted result as the sum of the baseline predicted value and the contribution value of each factor; And generating a visual interpretation graph to show the contribution size and direction of each factor, thereby providing basis for quality improvement.
10. A host computer readable storage medium storing computer readable instructions which, when read by a computer, are capable of running the LIMS-based all-class inspection data integration and quality traceability management system according to any one of claims 1-9.

Description

LIMS-based full-class detection data integration and quality traceability management system Technical Field The invention relates to the technical field of laboratory information management systems, in particular to a LIMS-based full-class detection data integration and quality traceability management system. Background With the rapid development of the detection industry and the continuous expansion of the service range, the comprehensive detection mechanism bears detection tasks in multiple fields such as food safety detection, environment monitoring, medical detection, industrial product quality detection and the like. The detection business in different fields has certain difference in aspects of sample characteristics, detection standards, data formats, quality requirements and the like, and higher requirements are provided for a laboratory information management system. The traditional LIMS system is mainly designed aiming at the detection service of single field or limited product, is difficult to adapt to the service requirements of all product types and cross-field, and faces a plurality of technical challenges in practical application. The existing LIMS system has the problem of insufficient universality of data modeling in the aspect of overall quality detection data management; the data in different detection fields have certain information such as recording sampling points, meteorological conditions and the like, the medicine detection needs to record parameters such as medicine batch numbers, validity periods and the like, and the specific data attributes in the fields are difficult to describe by using a unified data model; the existing system lacks a unified data modeling method which can describe cross-domain common elements and flexibly expand to support special requirements of various fields, so that data islanding phenomenon is serious, cross-domain data is difficult to integrate and transversely compare, in the aspect of multi-source heterogeneous data integration, instruments and equipment provided by a detection mechanism come from different manufacturers, the instrument output data formats of each manufacturer are different, for example, data files of analysis instruments such as a mass spectrometer, a chromatograph, a spectrometer and the like comprise proprietary binary formats, XML formats, CSV formats and the like, the structure is complex and the analysis documents are lacked, the existing LIMS system has limited integration capability of multi-source heterogeneous instrument data, the problem of data acquisition is mainly solved by developing special data interfaces for each instrument, the development cost is high, the period is long, the intelligent identification capability of the data formats is lacked, automatic mapping and conversion of heterogeneous data cannot be realized, in the aspect of quality traceability management, the accuracy and the reliability of detection results are directly related to product quality evaluation and public security, the traceability quality is crucial, the traditional quality traceability is mainly dependent on paper records and databases, when the detection results need to be detected, the system is more time-consuming and has more important requirements on the data, and the data can be tampered by a certain system is difficult to finish based on the data, the database manager or the user with advanced authority can modify the historical record without leaving marks, the evidence effectiveness of the traceable data is questioned in quality accident investigation or legal litigation, the prior art lacks an effective data tamper-proof mechanism, a reliable traceable system with legal effectiveness cannot be established, the prior quality control mainly adopts a post-inspection mode, the quality problem is found by analyzing the detection result of a quality control sample, active monitoring and predictive analysis on quality influence factors are lacking, in fact, the detection quality is comprehensively influenced by a plurality of dimension factors such as instrument equipment state, environmental conditions, reagent batch, operator proficiency and the like, but the factors are often independently managed and cannot form a comprehensive analysis system, and the prior system lacks intelligent analysis capability of learning influence rules of the factors and carrying out quality prediction based on the historical data, so that the advanced identification and active prevention of quality risks cannot be realized. Therefore, an innovative technical scheme is needed, unified modeling and intelligent integration of all-class detection data can be achieved, a tamper-proof quality traceability system is built, quality prediction and intelligent early warning based on multi-dimensional factor fusion are achieved, and therefore data management capacity and quality guarantee level of a detection mechanism are comprehensively improved. Disclosure of