CN115169986A - A Quality Control System for Clinical Laboratory - Google Patents

Abstract

The invention discloses a data quality control system of a clinical laboratory, which comprises a data acquisition module, a data preposition module, a data filtering module, a data preprocessing module and a calculation module, wherein the data preprocessing module comprises: the device comprises a first sub-processing module and a second sub-processing module. According to the scheme, a PBRTQC system is established through an EWMA operation program to realize indoor quality monitoring, meanwhile, a medical technology sharing platform is established to promote data interconnection and intercommunication and result mutual recognition of PBRTQCs of large medical institutions, and reasonable supporting measures are taken to ensure that the platform runs well.

Description

A Quality Control System for Clinical Laboratory

technical field

The invention belongs to the field of medical technology, in particular to a quality control system of a clinical laboratory.

Background technique

The traditional internal quality control (IQC) is the core content of the clinical laboratory quality control system and an important method to evaluate the precision and accuracy of the detection system. But with the ever-increasing volume of testing and the rapid evolution of modern inspection models, today's labs are no longer the labs in the traditional sense. For example, in some large medical institutions, multiple devices in the laboratory are used to test the same test items, or different medical institutions test the same test items. Therefore, the quality control mode should also keep pace with the times. At present, more and more scholars believe that only using traditional indoor quality control cannot be accurate and fast in analyzing errors. The limitations of traditional quality control are mainly reflected in the following aspects: 1. Traditional quality control lacks interchangeability , on the one hand, it will lead to a decrease in the true positive rate of error identification or an increase in the false positive rate, and on the other hand, it will cause the IQC target value to be related to the instrument instead of reflecting the true level of the analyte. 2. Since IQC assumes that the error is continuous, it cannot be detected until the next quality control is performed, so it cannot effectively monitor short-term system drift. 3. IQC only monitors the process of analysis, and cannot detect changes in detection performance in real time. When the test results of quality control products are "out of control", it is often necessary to reserve samples for re-inspection of samples before the test time point of quality control products for traceability and Evaluate affected specimens prior to the "out-of-control" time point. Fourth, the quality control products have problems such as poor stability, narrow detection range and matrix effects, and the comparison function between systems is weak. Fifth, the cost of IQC (quality control materials and reagents) cannot be ignored.

In 1965, Professors Hoffman and Waid proposed the idea that the stability of the detection system can be assessed by monitoring the test results of patient samples, that is, monitoring the average value of normal (AoN) of patient data within the daily "normal range" for quality control. concept, thus laying the foundation for real-time quality control of patient data in laboratory medicine. However, due to the difficulty in obtaining data and the large amount of calculation, this method was not used in clinical practice. In 1974, under the continuous research and improvement of international scholars, a variant of the AoN method, the BULL method, appeared, and was widely used in the quality control of automatic blood analyzers. Over time, a class of derived algorithms based on patient data emerged, collectively referred to as Patient-Based Real-Time Quality Control (PBRTQC). The guidance document issued in 2020 by the Quality Committee under the Analysis of the International Federation of Clinical Chemistry and Laboratory Medicine pointed out: PBRTQC is a statistical and mathematical model based on the use of patient clinical specimen test results to monitor the analytical performance of the test process in real time and continuously. The most advanced quality control method is to evaluate the comparability of clinical test items, monitor pre-analytical errors (specimen collection, transportation and processing, etc.), monitor analytical errors (derived from reagents, instruments and calibration), and change the batch number of indoor quality control products during the period. The advantages of continuous monitoring of real performance changes of the detection system at all times, direct analysis of patient results without consumption, no matrix effects, real-time and whole-process comparison (longitudinal comparison of one instrument or horizontal comparison of multiple instruments) and low cost are the advantages of IQC. Supplementation, in a sense free quality control, is beneficial to the laboratory. PBRTQC includes a variety of operational procedures, including normal mean method (average ofnomals, AON), BULL method, moving median method (moving median, MA) and exponentially weighted moving average (exponentially weighted moving average, EWMA) and so on. Among them, the EWMA method was first proposed by Roberts in 1959. The EWMA method introduces a weight coefficient (λ), and calculates a new MA value for each new test result through the distribution of the weight of the two test results before and after, in order to achieve the best and fast bias detection, representing a more realistic and continuous moving average, with the advantage of early detection of small variations in inaccuracy or imprecision in the test analysis.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a quality control system of a clinical laboratory to solve the problems existing in the above-mentioned prior art.

To achieve the above object, the present invention provides: a data quality control system of a clinical laboratory, comprising:

Data acquisition module, used to obtain laboratory-related data information;

Data front module, used to remove part of the data;

The data filtering module is used to group the data, and establish a PBRTQC model based on the grouping results;

A data preprocessing module, the data preprocessing module includes: a first sub-processing module for truncating the grouped data; a second sub-processing module for transforming the truncated data through Box-Cox normalization transformation is a more normal distribution;

The calculation module is used to calculate the mean value of the preprocessed data;

The parameter setting module is used to calculate the control limit based on the preprocessed data and judge whether the control limit is exceeded.

Preferably, calculating the control limit includes: using a weighted moving average method, respectively giving different weights to the observed values, and obtaining the moving average according to the different weights.

Preferably, calculating the control limits comprises: calculating using the EWMA method.

Out low=Mean-4*SD

Error low=Mean-3*SD’

Warning low=Mean-2*SD’

Warning high=Mean+2*SD’

Error high=Mean+3*SD’

Out high=Mean+4*SD

SD’=SD*(λ/(2-λ))1/2

Among them, Mean represents the mean value, SD represents the standard deviation, SD' represents the standard deviation', λ represents the weighting coefficient, Outlow represents the lower limit of data removal, Out high represents the upper limit of data removal, Error low represents the lower limit of error, Error high represents the upper limit of error, Warning low represents the lower warning limit, and Warning high represents the upper warning limit.

Preferably, the data acquisition module is connected with a laboratory information management system for data acquired from the laboratory information management system, and the data acquisition module is connected with the data filtering module to transmit the acquired data to the data The filtering module allows real-time extraction to be saved during data collection.

Preferably, the collected data includes a training data set and a validation data set, the training data set is used for parameter setting and program establishment of the PBRTQC method, and the validation data set is used for performance verification and implementation effect evaluation of the PBRTQC real-time quality control program.

Preferably, the training data set and the validation data set are generally distributed identically.

Preferably, if the control limit is exceeded, an alarm is performed, and the set alarm conditions include:

控制限，判断为失控；a.1 _3s : 1 sample measured value exceeds

The control limit is judged to be out of control;

10个连续的样本测定值同时落在均值一侧，提示系统误差；b.

The measured values of 10 consecutive samples fall on the side of the mean at the same time, indicating a systematic error;

或

控制限，提示系统误差N＝NPed。cN _2s : The measured values of N consecutive samples exceed the

or

Control limit, prompting system error N=NPed.

Preferably, the control system is connected with a medical technology sharing platform system, and the medical technology sharing platform system includes:

The data transmission module is used to access the medical technology information system of each hospital in real time, and the test data is collected actively and uploaded to the platform repository;

Data sharing module, through which the patient's previous inspection data is obtained from the platform's repository;

The platform management module is connected with the data transmission module for receiving the data actively collected by the data transmission module, and the platform management module is connected with the data sharing module for maintaining the data in the repository, and is also used for the medical technology sharing platform Carry out daily operation and maintenance monitoring.

The technical effects of the present invention are as follows: the present application establishes a PBRTQC system through the EWMA operation program to realize indoor quality monitoring of serum biochemical items and gradually popularizes PBRTQC; Mutual recognition and rationalization of supporting measures ensure the good operation of the platform to promote the acceptance of PBRTQC in domestic clinical laboratories and achieve wide application.

Description of drawings

The accompanying drawings constituting a part of the present application are used to provide further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation of the present application. In the attached image:

1 is a schematic structural diagram in an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of a medical technology sharing platform system in an embodiment of the present invention.

Detailed ways

It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

As shown in Figure 1, this embodiment provides a quality control system for a clinical laboratory. The system is designed based on the Remisol Advance system. The quality control system includes:

Data acquisition module, used to obtain laboratory-related data information;

The data pre-processing module is connected with the data filtering module, and is used to eliminate some data. The data is the data of the unavoidable factors that have a greater impact on the data distribution, including proficiency testing sample data, quality control sample data, research sample data, Data such as dialysis sample data, non-numerical results and non-serum sample types; the data pre-module.

The data filtering module is used to group data; for example, it can be grouped according to biological variation, specifically, select WST403-2012 "Analysis Quality Index of Routine Items in Clinical Biochemical Testing", the National Centers for Environmental Cholesterol Education Program (National Centers for Environmental Prediction, NECP) and the biological variability database as sources of quality control criteria for different projects. CVs that meet or are close to the standard are selected to count the smallest statistical indicators; it can also be grouped according to outpatient clinics, wards, and physical examination groups; it can also be grouped according to age and gender; it can also be grouped according to instruments or modules; data to establish the PBRTQC model respectively;

Data preprocessing modules include:

The first sub-processing module is used to truncate the grouped data. Specifically, for the distributed data with skewed distribution and extreme values, the data outside the truncation value range will be shortened or removed. ;

The second sub-processing module transforms the truncated data into a relatively normal distribution through Box-Cox normalization transformation.

The calculation module is used to calculate the mean value of the preprocessed data; the Remisol Advance system is used to calculate the mean value. Specifically, the weighted moving average method is used to give different weights to the observed values, and the moving average is obtained according to different weights. , and based on the last moving average, determine a forecast value that reflects the trend of recent changes in the data.

The parameter setting module is used to calculate the control limit based on the preprocessed data. When the control limit is exceeded, an alarm will be issued. In this embodiment, the selected parameters are: target false positive alarm rate (desirable false alarm rate, DFAR), specifically:

Control limits were determined based on the Remisol Advance system manual. Remisol Advance system: including data collection, data storage, data extraction and analysis, statistical analysis, visual quality control chart, performance verification capability of test algorithm, real-time operation function, early warning and out-of-control correction and audit records, system log tracking, multi-dimensional visualization tools , data normal distribution statistics, data filter and integrated IQC and other basic functions. The specific method of EWMA is as follows:

Out low=Mean-4*SD

Error low=Mean-3*SD’

Warning low=Mean-2*SD’

Warning high=Mean+2*SD’

Error high=Mean+3*SD’

Out high=Mean+4*SD

SD’=SD*(λ/(2-λ))1/2

Among them, Mean represents the mean value, SD represents the standard deviation, SD' represents the standard deviation', λ represents the weighting coefficient, Outlow represents the lower limit of data removal, Out high represents the upper limit of data removal, Error low represents the lower limit of error, Error high represents the upper limit of error, Warning low represents the lower warning limit, and Warning high represents the upper warning limit.

In an optional embodiment, the collected clinical sample data is collected from October 2021 to November 2023, and the detection results of serum biochemical items of outpatients and inpatients in the hospital. Among them, the test results of the specimens from October 2021 to October 2022 were used as the training data set for parameter setting and program establishment of the PBRTQC method; the test results of the specimens from November 2022 to November 2023 were used as the verification data set, and For the performance verification and implementation effect evaluation of the PBRTQC real-time quality control program. The training dataset and the validation dataset have the same population distribution.

Further, the data acquisition module is connected with a laboratory information management system, which is used for acquiring data from the laboratory information management system and transmitting the acquired data to the data filtering module. In this embodiment, when data is collected, it is allowed to extract data in real time for storage. When transferring patient data to the data acquisition module, care should be taken to cancel the identification information of the patient's name in the data set. The data includes: sample testing date and time (hour, minute, second), doctor's order number, sample type, billing department, clinical diagnosis information, patient age, gender, hospital number, test item name, test result and unit, etc. Basic information; instrument information includes reagent batch number, bottle number, indoor quality control, instrument alarm information, serum index and other information.

In an optional embodiment, optimizing the PBRTQC model includes: optimizing a variety of PBRTQC algorithms through a grid search method similar to the exhaustive method. Calculate the combination of hyperparameters such as upper alarm limit, lower alarm limit, upper warning limit, and lower warning limit to obtain the PBRTQC model with the best performance.

In an optional embodiment, set the alarm rules of the PBRTQC system:

控制限，判断为失控。a.1 _3S : 1 sample measured value exceeds

The control limit is judged to be out of control.

10个连续的样本测定值同时落在均值一侧，提示系统误差。b.

The measured values of 10 consecutive samples fall on one side of the mean at the same time, indicating a systematic error.

或

控制限，提示系统误差(N＝NPed)。cN _2S : N consecutive sample measured values exceed at the same time

or

Control limit, indicating systematic error (N=NPed).

d. Analysis of the cause of the alarm: aging of electrodes, deviation of calibrator packaging, replacement of reagent batch numbers, micro-blocking of pipelines, replacement of reagents with new batch numbers, failure to perform calibration or project calibration problems, etc.

In an optional embodiment, the performance of the PBRTQC model is verified by simulating calculation by drawing a bias detection curve graph on the verification data set. It mainly includes the following aspects: false positive alarm rate (FAR) and the number of samples required for error detection (NPed). Currently, the evaluation of specificity mainly uses false positive alarm rate (FAR), and its calculation formula is:

FAR=false positive alarms/total samples or FAR=1-specificity;

The number of samples (NPed) required for error detection is used to evaluate the sensitivity. When evaluating the sensitivity of the model, it is usually necessary to divide the data (construct a virtual day) to obtain multiple NPeds, and calculate the average NPed (ANPed), median Models were fully evaluated by counting NPed (MNPed) or 95% site NPed (95NPed).

In an optional embodiment, it also includes the construction of a BRTQC medical technology sharing platform system: the PBRTQC sharing platform is jointly negotiated by the medical inspection center and cooperative manufacturers, and operates under the supervision of the health quality control center after its establishment. As shown in Figure 2, the BRTQC medical technology sharing platform includes:

Data transmission module: It is used to access the medical technology information system of each hospital in real time. It adopts active collection of inspection data and uploads it to the platform repository, which helps to improve the quality control level of each hospital and effectively prevents data fraud and incomplete patient data. Specifically, the data transmission module actively collects data and returns it to the platform management module, without the need for additional intervention by the platform management module. Using this method can reduce the pressure on the platform in certain procedures.

Data sharing module: Physicians can use this module to obtain the patient's previous inspection data from the platform's repository. On the one hand, it can help solve the problems of patients' previous medical treatment data loss and forgetting to carry the diagnosis and treatment results. On the other hand, it can eliminate the problem of incomplete entity data. And cause the patient to repeat the examination phenomenon, to improve the efficiency of the doctor, reduce the burden on the patient.

The platform management module is connected with the data transmission module for receiving the data actively collected by the data transmission module, and the platform management module is connected with the data sharing module for maintaining the data in the repository. This module performs daily operation and maintenance monitoring, such as the doctor's permission setting, etc.

The monitoring process includes:

1) Quality management system: IQC and instrument maintenance are carried out in accordance with the standardized management requirements of ISO 15189 quality management system, and the clinical specimens of the day can be tested after IQC is in control.

2) Statistical method of patient data: The test results are connected from the laboratory information system (LIS) to the RemisolAdvance system platform, and the EWMA control chart is used for verification and real-time dynamic monitoring.

The general calculation formula for EWMA chart is: Y(r)=λX(r)+(1-λ)Y(r-1), where X(r) is the value of run batch r and Y(r) is the monitoring value (average), λ is the weighting coefficient, 0<λ≤1 (λ is usually selected in the range 0.2-0.5, the default value is 0.2).

3) Test analysis of normal distribution of patient data: Kolmogorov-Smirnov test was performed on the test results through the automatic statistical system of normal distribution of patient data on the Remisol Advance system platform to determine whether the range of cutoff values included in the overall patient population of different biochemical projects conforms to the positive state distribution.

4) EWMA quality control parameter setting and program establishment: parameters include quality target setting, quality control rule setting, automatic data extraction, parameter setting, intelligent calculation, performance verification, optimal PBRTQC method selection, performance evaluation, implementation and operation, etc. The estimated value of EWMA for each test result included in the patient population, the cumulative coefficient of variation (CV) of patient data was calculated, and compared with the precision quality standard; the EWMA quality control chart and the external quality control L-J quality control chart were used for Trend correlation analysis, through on-site quality records and instrument information verification, to confirm the cause, change trend and time node of the warning, and screen out the most suitable EWMA quality control parameters.

5) Implementation of the EWMA quality control program: Use the most suitable EWMA quality control parameters to record the warning situation of the EWMA quality control program in the project inspection and the corresponding situation of the IQC. If an early warning occurs, observe the IQC during the start-up period, and redo the quality control. By viewing the biochemical analyzer records and on-site quality records, analyze the reasons that may affect the performance of the detection system to determine true/false alarms, and deal with the true alarms in time.

The above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited to this. Substitutions should be covered within the protection scope of this application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims (8)

1. a data quality control system of a clinical laboratory, is characterized in that, comprises: Data acquisition module, used to obtain laboratory-related data information; Data front module, used to remove part of the data; The data filtering module is used to group the data, and establish a PBRTQC model based on the grouping results; A data preprocessing module, the data preprocessing module includes: a first sub-processing module for truncating the grouped data; a second sub-processing module for transforming the truncated data through Box-Cox normalization transformation is a more normal distribution; The calculation module is used to calculate the mean value of the preprocessed data; The parameter setting module is used to calculate the control limit based on the preprocessed data and judge whether the control limit is exceeded. 2. the data quality control system of clinical laboratory according to claim 1, is characterized in that, calculating the control limit comprises: using weighted moving average method, give different weights to observation value respectively, obtain according to different weights moving average. 3. the data quality control system of clinical laboratory according to claim 1, is characterized in that, calculating control limit comprises: use EWMA method to calculate; Out low=Mean-4*SD Error low=Mean-3*SD’ Warning low=Mean-2*SD’ Warning high=Mean+2*SD’ Error high=Mean+3*SD’ Out high=Mean+4*SD SD’=SD*(λ/(2-λ))1/2 Among them, Mean represents the mean value, SD represents the standard deviation, SD' represents the standard deviation', λ represents the weighting coefficient, Out low represents the lower limit of data removal, Out high represents the upper limit of data removal, Error low represents the lower limit of error, Error high represents the upper limit of error, Warning low represents the lower warning limit, and Warning high represents the upper warning limit. 4. the data quality control system of clinical laboratory according to claim 1, is characterized in that, described data acquisition module is connected with laboratory information management system, for the data that will be acquired from laboratory information management system, so The data acquisition module is connected with the data filtering module, and transmits the acquired data to the data filtering module, and allows real-time extraction to be stored during data acquisition. 5. the data quality control system of clinical laboratory according to claim 4, is characterized in that, the described data of acquisition comprises practice data set, verification data set, and practice data set is used for the parameter setting of PBRTQC method and program establishment, The validation dataset is used for the performance verification and implementation effect evaluation of the PBRTQC real-time quality control program. 6 . The data quality control system of a clinical laboratory according to claim 5 , wherein the overall distribution of the training data set and the validation data set is consistent. 7 . 7. the data quality control system of clinical laboratory according to claim 1, is characterized in that, if exceed control limit, then carry out an alarm, and the alarm condition of setting comprises: a.1 _3s : 1 sample measured value exceeds

The control limit is judged to be out of control; b.

The measured values of 10 consecutive samples fall on the side of the mean at the same time, indicating a systematic error; cN _2s : The measured values of N consecutive samples exceed the

or

Control limit, prompting system error N=NPed. 8. The data quality control system of the clinical laboratory according to claim 1, wherein the control system is connected with a medical technology sharing platform system, and the medical technology sharing platform system comprises: The data transmission module is used to access the medical technology information system of each hospital in real time, and the test data is collected actively and uploaded to the platform repository; Data sharing module, through which the patient's previous inspection data is obtained from the platform's repository; The platform management module is connected with the data transmission module for receiving the data actively collected by the data transmission module, and the platform management module is connected with the data sharing module for maintaining the data in the repository, and is also used for the medical technology sharing platform Carry out daily operation and maintenance monitoring.

Images