Fault prediction method of in-vitro diagnostic equipment, computer equipment and computer readable storage medium

Abstract

The invention discloses a fault prediction method of in-vitro diagnostic equipment, computer equipment and a computer readable storage medium, and relates to the technical field of medical equipment, and the method comprises the steps: obtaining target data of the in-vitro diagnostic equipment collected in different equipment operation cycles; the target data comprises equipment calibration data and detection related data of the in-vitro diagnosis equipment; constructing corresponding equipment operation time sequence characteristics based on the target data; the equipment operation time sequence characteristics comprise a first time sequence characteristic used for reflecting the calibration quality of the in-vitro diagnostic equipment, a second time sequence characteristic used for reflecting the fluctuation degree of detection related data of the in-vitro diagnostic equipment, and a third time sequence characteristic used for reflecting the change trend of the accuracy of the detection related data of the in-vitro diagnostic equipment; and inputting the equipment operation time sequence characteristics into a pre-trained target linear regression model to obtain a fault prediction result of the in-vitro diagnosis equipment output by the target linear regression model. And the abnormal condition of the in-vitro diagnosis equipment in the operation process can be found in time.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to a method for predicting the failure of in vitro diagnostic equipment, a computer device, and a computer-readable storage medium. Background Technology

[0002] In vitro diagnostic equipment, such as testing instruments involving sample needles, reagent needles, and reaction cups, may experience issues like empty aspiration or needle blockage during the initial aspiration process due to operational problems, reagent placement order, temperature differences, etc., potentially leading to abnormal test results. Because the analysis process involves many influencing factors, large data volumes, and significant differences between different items, there is currently no consistent data for reference, and manual judgment of whether multiple devices are malfunctioning is impossible. Currently, information on equipment operation is mainly obtained through calibration test results, regular on-site maintenance, and feedback. This information is relatively delayed and cannot resolve issues promptly, relying heavily on timely checks or feedback from relevant personnel.

[0003] As can be seen from the above, how to promptly detect abnormalities in the operation of in vitro diagnostic equipment is an urgent problem to be solved. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a fault prediction method, a computer device, and a computer-readable storage medium for in vitro diagnostic equipment, capable of timely detecting abnormalities in the operation of in vitro diagnostic equipment. The specific solution is as follows:

[0005] In a first aspect, this application provides a method for predicting the failure of an in vitro diagnostic device, including:

[0006] Acquire target data from in vitro diagnostic equipment collected under different equipment operating cycles; the target data includes equipment calibration data and detection-related data of the in vitro diagnostic equipment.

[0007] Based on the target data, corresponding device runtime sequence features are constructed; the device runtime sequence features include a first time-series feature reflecting the calibration quality of the in vitro diagnostic device, a second time-series feature reflecting the fluctuation of the detection-related data of the in vitro diagnostic device, and a third time-series feature reflecting the changing trend of the accuracy of the detection-related data of the in vitro diagnostic device;

[0008] The device's runtime sequence features are input into a pre-trained target linear regression model to obtain the fault prediction results of the in vitro diagnostic device output by the target linear regression model.

[0009] Optionally, the device calibration data is the calibration deviation data of the in vitro diagnostic device; the detection-related data includes the qualitative detection result data obtained after the in vitro diagnostic device performs in vitro diagnostics.

[0010] The qualitative detection result data includes first data related to the first qualitative detection result and / or second data related to the second qualitative detection result.

[0011] Optionally, constructing the corresponding device runtime sequence features based on the target data includes:

[0012] The data items contained in the target data are preprocessed to obtain preprocessed data;

[0013] Based on the preprocessed data, construct the corresponding device runtime sequence characteristics;

[0014] The preprocessing includes a first preprocessing and / or a second preprocessing; the first preprocessing is to remove data items and / or duplicate data items from the target data whose missing value ratio is greater than a preset ratio threshold; the second preprocessing is to fill in missing values ​​for data items in the target data whose missing value ratio is not greater than the preset ratio threshold.

[0015] Optionally, before preprocessing the data items contained in the target data, the method further includes:

[0016] Determine the target data item type; the target data item type is a data item type that exists in both equipment calibration data and testing-related data.

[0017] Based on the target data item type and using a multi-table join query, the data table corresponding to the equipment calibration data and the data table corresponding to the detection-related data are joined together to obtain a corresponding integrated data table, so as to preprocess the data items contained in the target data recorded in the integrated data table.

[0018] Optionally, the target data item type includes any one or a combination of several of the following: device identifier of in vitro diagnostic equipment, device operating time cycle, and in vitro diagnostic item identifier.

[0019] Optionally, the step of filling missing values ​​in data items in the target data where the proportion of missing values ​​is not greater than the preset proportion threshold includes:

[0020] Data items collected within a preset time period whose types are consistent with the type of data items whose missing value ratio is not greater than the preset ratio threshold, but whose types are inconsistent with the target data item type, and which do not have missing values;

[0021] Calculate the mean of the acquired data items;

[0022] The mean of the data is filled into the missing value positions of data items whose missing value ratio is not greater than the preset ratio threshold, to obtain the filled data items.

[0023] Optionally, constructing the corresponding device runtime sequence features based on the target data includes:

[0024] Based on any type of target data, determine the corresponding discrete target features; each discrete target feature is a feature corresponding to any type of target data collected by the same in vitro diagnostic equipment under different equipment operation time periods for the same in vitro diagnostic project.

[0025] The target features are sorted according to the time sequence of the equipment operation time cycle to obtain the equipment operation time sequence feature corresponding to any one of the target data.

[0026] Optionally, the feature corresponding to the equipment calibration data is the average value of the equipment calibration data.

[0027] Optionally, the feature corresponding to the detection-related data is the number of abnormal data in the target control chart; the target control chart is the control chart of the exponentially weighted moving average corresponding to the detection-related data, and the abnormal data is the data in the target control chart whose exponentially weighted moving average exceeds a preset threshold range.

[0028] Optionally, the feature corresponding to the detection-related data is the difference between the detection-related data and the average value of historically collected detection-related data.

[0029] Optionally, the difference between the detection-related data and the average value of historically collected detection-related data is the product of the difference between the detection-related data and the average value of historically collected detection-related data and the confidence level of the detection-related data.

[0030] The reliability of the detection-related data is the ratio between the number of tests conducted within the equipment's operating time period and the average of the corresponding historical test counts; the number of tests is the number of times the same in vitro diagnostic equipment performs tests on the same in vitro diagnostic item.

[0031] Optionally, the fault prediction method for the in vitro diagnostic device further includes:

[0032] Acquire target historical data collected under different equipment operating cycles; the target historical data is historical data collected during the operation of an in vitro diagnostic device in a faulty state, and the target historical data includes historical equipment calibration data and historical test-related data;

[0033] Assign corresponding label values ​​to the target historical data; the label values ​​represent the probability that the in vitro diagnostic equipment malfunctioned at the time the target historical data was collected;

[0034] The target linear regression model is trained based on the target historical data and the corresponding label values ​​to obtain the target linear regression model.

[0035] Optionally, assigning corresponding tag values ​​to the target historical data includes:

[0036] Determine the time difference between the time the target historical data was collected and the time the fault occurred;

[0037] Based on the time difference, assign corresponding label values ​​to the target historical data;

[0038] Specifically, the target historical data is assigned a corresponding label value based on the negative correlation between the label value and the time difference.

[0039] Optionally, the process of training the linear regression model based on the target historical data and the corresponding label values ​​further includes:

[0040] The linear regression model is optimized by cross-validation and using the objective loss function.

[0041] The target loss function is a loss function determined based on minimizing the mean squared error, and the parameter optimization operation includes optimization operations on the feature weight coefficients and bias terms of the linear regression model.

[0042] Optionally, the fault prediction method for the in vitro diagnostic device further includes:

[0043] Based on the SHAP analysis method, the marginal contribution of each device's runtime sequence characteristics to the fault prediction result output by the target linear regression model is determined, and the runtime sequence characteristics of each device and the corresponding marginal contribution are visualized.

[0044] Optionally, the fault prediction method for the in vitro diagnostic device further includes:

[0045] Using the simple moving average method and the target Z-score set, the corresponding simple moving average line is generated;

[0046] Based on the degree of deviation between the Z-score corresponding to the target data and the simple moving average, the in vitro diagnostic equipment is predicted to be faulty, and the target fault prediction result is obtained.

[0047] By comparing the target fault prediction results with the fault prediction results output by the target linear regression model, it is determined whether the target linear regression model should be optimized.

[0048] The target Z-score set includes the target data and the corresponding historical data, each with its own Z-score.

[0049] Secondly, this application provides a computer device, comprising:

[0050] Memory, used to store computer programs;

[0051] A processor is used to execute the computer program to implement the aforementioned fault prediction method for in vitro diagnostic devices.

[0052] Thirdly, this application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned fault prediction method for in vitro diagnostic devices.

[0053] This application acquires target data of an in vitro diagnostic device collected under different device operating cycles; the target data includes device calibration data and detection-related data of the in vitro diagnostic device; based on the target data, corresponding device operating sequence features are constructed; the device operating sequence features include a first time-series feature reflecting the calibration quality of the in vitro diagnostic device, a second time-series feature reflecting the fluctuation of the detection-related data of the in vitro diagnostic device, and a third time-series feature reflecting the changing trend of the accuracy of the detection-related data of the in vitro diagnostic device; the device operating sequence features are input into a pre-trained target linear regression model to obtain the fault prediction results of the in vitro diagnostic device output by the target linear regression model.

[0054] As can be seen from the above, this application first acquires equipment calibration data and testing-related data of in vitro diagnostic equipment collected under different equipment operating cycles to construct time-series features that reflect the trends in calibration quality, data fluctuation, and accuracy. Then, a pre-trained objective linear regression model is used to fit the linear relationship between the equipment operating time-series features and the failure probability, thereby achieving failure prediction of the in vitro diagnostic equipment. In this way, by capturing early abnormal signals of in vitro diagnostic equipment through time-series features, early failure prediction can be achieved, reducing the risk of diagnostic delays caused by equipment failure and improving the quality of equipment diagnosis. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0056] Figure 1 This is a flowchart of a fault prediction method for an in vitro diagnostic device disclosed in this application;

[0057] Figure 2 A schematic diagram of a data list provided for this application;

[0058] Figure 3 A schematic diagram of the device's operating sequence characteristics provided in this application;

[0059] Figure 4 A schematic diagram of SHAP values ​​corresponding to a device runtime sequence feature provided in this application;

[0060] Figure 5 This is a structural diagram of a computer device disclosed in this application. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Currently, information on the operational status of in vitro diagnostic equipment is mainly obtained based on calibration test results, regular on-site maintenance, and feedback. However, this information is relatively delayed and cannot resolve issues promptly, relying heavily on timely checks or feedback from relevant personnel. Therefore, this application provides a method for predicting the failure of in vitro diagnostic equipment.

[0063] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for predicting the failure of an in vitro diagnostic device, comprising:

[0064] Step S11: Obtain target data of the in vitro diagnostic equipment collected under different equipment operation cycles; the target data includes the equipment calibration data and detection-related data of the in vitro diagnostic equipment.

[0065] In this embodiment, initial data output by the in vitro diagnostic device is first acquired under different device operating cycles. If the in vitro diagnostic device is a molecular diagnostic device, including a nucleic acid purification and PCR (Polymerase Chain Reaction) fluorescence detector, the initial output data is the sample concentration value. Based on the initial data, the device calibration data and detection-related data of the in vitro diagnostic device are derived. Further, the device calibration data, including calibration deviation data, is determined using the deviation between the sample concentration value and the theoretical concentration value of a preset standard. The theoretical concentration value of the preset standard is a standard concentration value used to calibrate the accuracy of the detection results after calibration and verification. It is then determined whether the sample concentration value is greater than a first target threshold. If the sample concentration value is greater than the first target threshold, the corresponding sample detection result is positive. If the sample concentration value is less than the first target threshold but greater than a second target threshold, the corresponding sample detection result is weakly positive. The first target threshold and the second target threshold can be set according to actual conditions. Then, the number of positive and weakly positive samples is counted to obtain the corresponding positive rate data and weakly positive rate data; the positive rate data includes the positive rate and the corresponding number of positive rate samples; the weakly positive rate data includes the weakly positive rate and the corresponding number of weakly positive rate samples.

[0066] It is understood that the target data includes not only the equipment calibration data and testing-related data of the in vitro diagnostic equipment, but also the equipment identification, equipment operating time cycle, and in vitro diagnostic item identification of the in vitro diagnostic equipment. The equipment identification can be the equipment serial number; the equipment operating time cycle can be divided into time units such as days and weeks; the in vitro diagnostic item identification can be the name of the in vitro diagnostic item, the in vitro diagnostic item number, or other information that can characterize the in vitro diagnostic item. Specifically, the equipment calibration data is the calibration deviation data of the in vitro diagnostic equipment; the testing-related data includes the qualitative testing result data obtained after the in vitro diagnostic equipment performs in vitro diagnostics; wherein, the qualitative testing result data includes first data related to the first qualitative testing result and / or second data related to the second qualitative testing result; the first qualitative testing result is positive, and the second qualitative testing result is weakly positive.

[0067] Figure 2 This is a schematic diagram of a data list provided in this embodiment. The Z-score of the detection result can be the result of the initial data output by the in vitro diagnostic device after standardization by the Z-score formula. The calibration deviation is the calibration deviation data. The data corresponding to the positive rate analysis and weak positive rate analysis are the results obtained by statistically analyzing the number of positive and weak positive samples respectively. The detection result is the initial data output by the in vitro diagnostic device, such as sample concentration value, luminescence value, etc.

[0068] Step S12: Construct corresponding device runtime sequence features based on the target data; the device runtime sequence features include a first time-series feature reflecting the calibration quality of the in vitro diagnostic device, a second time-series feature reflecting the fluctuation of the detection-related data of the in vitro diagnostic device, and a third time-series feature reflecting the changing trend of the accuracy of the detection-related data of the in vitro diagnostic device.

[0069] In this embodiment, after obtaining the target data, data cleaning is performed on the target data, which includes various data items, i.e., each data item in the target data. Data items in the target data with a missing value ratio greater than a preset threshold are removed, and duplicate data items in the target data are also removed to obtain cleaned target data, ensuring the uniqueness of the target data and avoiding deviations caused by repeated calculations. Then, missing value filling is performed on data items with a missing value ratio not greater than the preset threshold to obtain preprocessed data. In this embodiment, the preset threshold is set to 70%, but it can also be adjusted according to actual conditions.

[0070] Specifically, constructing corresponding device runtime sequence features based on the target data includes: preprocessing the data items contained in the target data to obtain preprocessed data; constructing corresponding device runtime sequence features based on the preprocessed data; wherein, the preprocessing includes a first preprocessing and / or a second preprocessing; the first preprocessing is to remove data items and / or duplicate data items from the target data whose missing value ratio is greater than a preset ratio threshold; the second preprocessing is to fill in missing values ​​for data items in the target data whose missing value ratio is not greater than the preset ratio threshold.

[0071] It is understood that when filling missing values ​​in the target data, the mean value of the complete data items corresponding to the missing values, excluding the device identifier, device operating time cycle, and in vitro diagnostic item identifier of the in vitro diagnostic equipment, can be used to fill the missing values. A complete data item refers to a data item without missing values. Specifically, filling missing values ​​in data items where the proportion of missing values ​​in the target data is not greater than the preset proportion threshold includes: obtaining data items collected within a preset time period whose data item type is consistent with the data item type where the proportion of missing values ​​is not greater than the preset proportion threshold but is inconsistent with the target data item type, and which do not have missing values; calculating the mean value of the obtained data items; and filling the missing values ​​in the data items where the proportion of missing values ​​is not greater than the preset proportion threshold to obtain the filled data items.

[0072] It is worth mentioning that before preprocessing the target data, it is necessary to integrate the target data into a main table. Through multi-table join queries, the data table corresponding to the equipment calibration data and the data table corresponding to the detection-related data are joined and integrated into a single table, resulting in an integrated data table. Specifically, the data item types present in both the equipment calibration data and the detection-related data are defined as the field names of the integrated data table, such as detection time, equipment serial number, detection item name, positive rate, weak positive rate, etc. Specifically, before preprocessing the data items contained in the target data, the process includes: determining the target data item type; the target data item type being a data item type present in both the equipment calibration data and the detection-related data; based on the target data item type and using multi-table join queries, joining the data table corresponding to the equipment calibration data and the data table corresponding to the detection-related data to integrate them into a corresponding integrated data table, so that the data items contained in the target data recorded in the integrated data table can be preprocessed. The target data item type includes any one or a combination of several of the following: device identifier of in vitro diagnostic equipment, device operating time cycle, and in vitro diagnostic item identifier.

[0073] In this embodiment, corresponding discrete target features are constructed based on the target data, and the target features are sorted according to the time sequence of the device operation time cycle to obtain the corresponding device operation time sequence features. The device operation time sequence features can be in vector form. By constructing the device operation time sequence features, the changing trend of features over time can be accurately captured. Specifically, constructing the corresponding device operation time sequence features based on the target data includes: determining corresponding discrete target features based on any type of target data; each discrete target feature is a feature corresponding to any type of target data collected by the same in vitro diagnostic device for the same in vitro diagnostic project under different device operation time cycles; and sorting the target features according to the time sequence of the device operation time cycle to obtain the device operation time sequence features corresponding to the arbitrary type of target data.

[0074] It is understood that the target features corresponding to the target data include the features corresponding to the device calibration data and the features corresponding to the detection-related data. Figure 3 This embodiment provides a schematic diagram of device runtime sequence characteristics, wherein the features corresponding to the device calibration data are equivalent to Figure 3The calibration deviation mean characteristic is calculated as the average value of the equipment calibration data, specifically the average value of equipment calibration data grouped according to the equipment identifier, equipment operating time cycle, and in vitro diagnostic item identifier. This value reflects the accuracy and stability of the in vitro diagnostic equipment during the calibration process. In one specific embodiment, the characteristic corresponding to the detection-related data is... Figure 3 The out-of-limit characteristic is calculated as the number of abnormal data in the target control chart. The target control chart is a control chart of the exponentially weighted moving average corresponding to the detection-related data, and the abnormal data is data in the target control chart whose exponentially weighted moving average exceeds a preset threshold range. Specifically, firstly, data exceeding four standard deviations are excluded from the detection-related data to obtain the target detection-related data, thus avoiding interference from extreme outliers. Then, the exponentially weighted moving average corresponding to the target detection-related data is determined, and a target control chart is constructed based on each of the exponentially weighted moving averages. Finally, the preset threshold range in the target control chart is determined, and the corresponding formula is as follows:

[0075] ;

[0076] in, The mean value corresponding to the target detection related data; The standard deviation of the target detection-related data; The weighting factor is used as the weighting factor; the resulting threshold is the upper and lower limits corresponding to the preset threshold range, and the exponentially weighted moving average value exceeding the preset threshold range in the target control chart is identified as abnormal data; the more abnormal data there are, the greater the fluctuation of the measured value and the higher the risk of equipment failure. It is worth mentioning that the weighting factor can be adjusted according to actual needs, and no specific limitation is made here.

[0077] In another specific implementation, the feature corresponding to the detection-related data is equivalent to Figure 3The calculation principles for the positive rate difference feature and the weak positive rate difference feature are the same. For example, the calculation method for the positive rate difference feature is that the feature corresponding to the test-related data is the difference between the test-related data and the average value of the historically collected test-related data; the difference between the test-related data and the average value of the historically collected test-related data is the product of the difference between the test-related data and the average value of the historically collected test-related data and the reliability of the test-related data; wherein, the reliability of the test-related data is the ratio between the number of tests within the equipment operating time period and the average value of the corresponding historical test number; the number of tests is the number of times the same in vitro diagnostic equipment performs the same in vitro diagnostic project; the historical test number includes the number of tests within the equipment operating time period. Specifically, the positive rate difference feature is the product of the difference between the positive rate data in the test-related data and the average value of the positive rate data in the historically collected test-related data and the reliability of the test-related data, and the corresponding formula is as follows:

[0078] ;

[0079] in, The indicator function is typically set to 0 or 1. When the positive rate is greater than 0, the indicator function is 1; otherwise, it is 0. The test reliability is the reliability of the detection-related data. Similarly, the positive rate difference feature is the product of the difference between the weak positive rate data in the detection-related data and the average of the weak positive rate data in historically collected detection-related data, and the reliability of the detection-related data. It is worth noting that an online learning strategy can be used to update the statistics corresponding to the target data in real time, allowing the performance of the in vitro diagnostic device to change over time.

[0080] Step S13: Input the device's runtime sequence features into the pre-trained target linear regression model to obtain the fault prediction results of the in vitro diagnostic device output by the target linear regression model.

[0081] In this embodiment, the pre-trained target linear regression model is obtained by training target historical data and corresponding label values. Specifically, target historical data collected under different equipment operating cycles is acquired; the target historical data is historical data collected during operation of an in vitro diagnostic device in a faulty state, and the target historical data includes historical device calibration data and historical detection-related data; corresponding label values ​​are assigned to the target historical data; the label values ​​represent the probability that the in vitro diagnostic device malfunctions at the time of collection of the target historical data; the linear regression model is trained based on the target historical data and the corresponding label values ​​to obtain the target linear regression model. Specifically, assigning corresponding label values ​​to the target historical data includes: determining the time difference between the time of collection of the target historical data and the time of fault occurrence; assigning corresponding label values ​​to the target historical data based on the time difference; wherein, the label values ​​are assigned to the target historical data according to the negative correlation between the label values ​​and the time difference. In one specific implementation, a series of labels from 0 to 10 can be designed to finely describe the fault development process. The larger the time difference between the time of collection of the target historical data and the time of the fault occurrence, the larger the label value; the smaller the time difference, the smaller the label value. For example, the label value corresponding to the day before the fault is 10, the label value is 9 two days before the fault, and so on. When the time difference exceeds 10 days, the label value is 0. This allows the model to capture the characteristic change trend when the fault is approaching, rather than just judging whether a fault exists. The target historical data and the corresponding label values ​​are input into a linear regression model to learn the mapping relationship between features and fault probability.

[0082] It is understood that during the training of the linear regression model, the feature weight coefficients and bias terms of the linear regression model are optimized through cross-validation and minimizing the mean squared error. Specifically, the process of training the linear regression model based on the target historical data and the corresponding label values ​​also includes: performing parameter optimization operations on the linear regression model through cross-validation and using a target loss function; wherein, the target loss function is a loss function determined based on minimizing the mean squared error, and the parameter optimization operations include optimization operations on the feature weight coefficients and bias terms of the linear regression model.

[0083] Furthermore, based on SHAP (SHapley Additive exPlanation, a machine learning model interpretation method based on Shapley values ​​in game theory), the marginal contribution of each device's runtime sequence characteristics to the fault prediction result output by the target linear regression model is determined, and the runtime sequence characteristics of each device and the corresponding marginal contribution are visualized. Figure 4This embodiment provides a schematic diagram of the SHAP value corresponding to the device runtime sequence feature. The vertical axis represents the feature name corresponding to the device runtime sequence feature, and the horizontal axis represents the SHAP value (i.e., marginal contribution) corresponding to the device runtime sequence feature. This is used to characterize the average influence of the device runtime sequence feature on the model output. It can be seen that the feature that exceeds the limit has the greatest impact on the fault, while the feature that calibrates the deviation has the least impact on the fault.

[0084] In this embodiment, after outputting the fault prediction result of the in vitro diagnostic equipment using the target linear regression model, the fault prediction result can be verified and optimized using the simple moving average method. Specifically, a corresponding simple moving average line is generated using the simple moving average method and the target Z-score set; based on the deviation between the Z-score corresponding to the target data and the simple moving average line, the fault of the in vitro diagnostic equipment is predicted, thus obtaining the target fault prediction result; by comparing the target fault prediction result with the fault prediction result output by the target linear regression model, it is determined whether the target linear regression model needs to be optimized; wherein, the target Z-score set includes the Z-scores corresponding to the target data and the corresponding historical data.

[0085] In one specific implementation, data exceeding three standard deviations from the target data are first excluded to improve the accuracy of subsequent Z-score calculations. Then, a target Z-score set is constructed using the Z-scores of the excluded target data and historical data within 30 days. A simple moving average is generated using the target Z-score set and a simple moving average line is generated. A threshold is set to indicate whether the deviation between the Z-score corresponding to the target data and the simple moving average line exceeds a target deviation threshold. This yields the target fault prediction result for the in vitro diagnostic equipment. The target deviation threshold can be adjusted according to actual conditions and is not specifically limited here. At this point, it is determined whether the fault prediction result output by the target linear regression model indicates that the in vitro diagnostic equipment has malfunctioned. If so, the fault prediction result is directly output. If the target fault prediction result is inconsistent with the fault prediction result output by the model, the target linear regression model is optimized to obtain a new target linear regression model. Then, the process jumps to the step of inputting the device's runtime sequence features into the pre-trained target linear regression model until the target fault prediction result is consistent with the fault prediction result output by the model. Finally, the fault prediction result of the in vitro diagnostic equipment is output. The fault prediction result outputs a fault occurrence probability value and a corresponding label value. The larger the label value, the closer the fault occurrence time.

[0086] As can be seen from the above, this application first acquires equipment calibration data and testing-related data of in vitro diagnostic equipment collected under different equipment operating cycles to construct time-series features that reflect the trends in calibration quality, data fluctuation, and accuracy. Then, a pre-trained objective linear regression model is used to fit the linear relationship between the equipment operating time-series features and the failure probability, thereby achieving failure prediction of the in vitro diagnostic equipment. In this way, by capturing early abnormal signals of in vitro diagnostic equipment through time-series features, early failure prediction can be achieved, reducing the risk of diagnostic delays caused by equipment failure and improving the quality of equipment diagnosis.

[0087] Furthermore, embodiments of this application also disclose a computer device. Figure 5 This is a structural diagram of a computer device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. The computer device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the fault prediction method for the in vitro diagnostic device disclosed in any of the foregoing embodiments.

[0088] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the computer device 20; the communication interface 24 can create a data transmission channel between the computer device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0089] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0090] The operating system 221 is used to manage and control the various hardware devices on the computer device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the fault prediction method of the in vitro diagnostic device executed by the computer device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0091] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned fault prediction method for in vitro diagnostic devices. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0092] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0093] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0094] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0095] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0096] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for predicting faults in an in vitro diagnostic device, characterized in that, include: Acquire target data from in vitro diagnostic equipment collected under different equipment operating cycles; The target data includes the equipment calibration data and testing-related data of the in vitro diagnostic equipment; Construct corresponding device runtime sequence characteristics based on the target data; The device operation sequence characteristics include a first time-series characteristic reflecting the calibration quality of the in vitro diagnostic device, a second time-series characteristic reflecting the fluctuation of the detection-related data of the in vitro diagnostic device, and a third time-series characteristic reflecting the changing trend of the accuracy of the detection-related data of the in vitro diagnostic device. The device's runtime sequence features are input into a pre-trained target linear regression model to obtain the fault prediction results of the in vitro diagnostic device output by the target linear regression model.

2. The fault prediction method for in vitro diagnostic equipment according to claim 1, characterized in that, The device calibration data refers to the calibration deviation data of the in vitro diagnostic device; the detection-related data includes the qualitative detection results data obtained after the in vitro diagnostic device performs in vitro diagnostics. The qualitative detection result data includes first data related to the first qualitative detection result and / or second data related to the second qualitative detection result.

3. The fault prediction method for in vitro diagnostic equipment according to claim 1, characterized in that, The construction of corresponding device runtime sequence features based on the target data includes: The data items contained in the target data are preprocessed to obtain preprocessed data; Based on the preprocessed data, construct the corresponding device runtime sequence characteristics; The preprocessing includes a first preprocessing and / or a second preprocessing; the first preprocessing is to remove data items and / or duplicate data items from the target data whose missing value ratio is greater than a preset ratio threshold; the second preprocessing is to fill in missing values ​​for data items in the target data whose missing value ratio is not greater than the preset ratio threshold.

4. The fault prediction method for in vitro diagnostic equipment according to claim 3, characterized in that, Before preprocessing the data items contained in the target data, the process further includes: Determine the target data item type; the target data item type is a data item type that exists in both equipment calibration data and testing-related data. Based on the target data item type and using a multi-table join query, the data table corresponding to the equipment calibration data and the data table corresponding to the detection-related data are joined together to obtain a corresponding integrated data table, so as to preprocess the data items contained in the target data recorded in the integrated data table.

5. The fault prediction method for in vitro diagnostic equipment according to claim 4, characterized in that, The target data item type includes any one or a combination of several of the following: device identifier of in vitro diagnostic equipment, device operating time cycle, and in vitro diagnostic item identifier.

6. The fault prediction method for in vitro diagnostic equipment according to claim 4, characterized in that, The step of filling missing values ​​in data items in the target data where the proportion of missing values ​​is not greater than the preset proportion threshold includes: Data items collected within a preset time period whose types are consistent with the type of data items whose missing value ratio is not greater than the preset ratio threshold, but whose types are inconsistent with the target data item type, and which do not have missing values; Calculate the mean of the acquired data items; The mean of the data is filled into the missing value positions of data items whose missing value ratio is not greater than the preset ratio threshold, to obtain the filled data items.

7. The fault prediction method for in vitro diagnostic equipment according to any one of claims 1 to 6, characterized in that, The construction of corresponding device runtime sequence features based on the target data includes: Based on any type of target data, determine the corresponding discrete target features; each discrete target feature is a feature corresponding to any type of target data collected by the same in vitro diagnostic equipment under different equipment operation time periods for the same in vitro diagnostic project. The target features are sorted according to the time sequence of the equipment operation time cycle to obtain the equipment operation time sequence feature corresponding to any one of the target data.

8. The fault prediction method for in vitro diagnostic equipment according to claim 7, characterized in that, The characteristic corresponding to the equipment calibration data is the average value of the equipment calibration data.

9. The fault prediction method for in vitro diagnostic equipment according to claim 7, characterized in that, The feature corresponding to the detection-related data is the number of abnormal data in the target control chart; the target control chart is the control chart of the exponentially weighted moving average corresponding to the detection-related data, and the abnormal data is the data in the target control chart whose exponentially weighted moving average exceeds a preset threshold range.

10. The fault prediction method for in vitro diagnostic equipment according to claim 7, characterized in that, The feature corresponding to the detection-related data is the difference between the detection-related data and the average value of historically collected detection-related data.

11. The fault prediction method for in vitro diagnostic equipment according to claim 10, characterized in that, The difference between the detection-related data and the average value of the historically collected detection-related data is the product of the difference between the detection-related data and the average value of the historically collected detection-related data and the reliability of the detection-related data. The reliability of the detection-related data is the ratio between the number of tests conducted within the equipment's operating time period and the average of the corresponding historical test counts; the number of tests is the number of times the same in vitro diagnostic equipment performs tests on the same in vitro diagnostic item.

12. The fault prediction method for in vitro diagnostic equipment according to any one of claims 1 to 6, characterized in that, Also includes: Acquire target historical data collected under different equipment operating cycles; the target historical data is historical data collected during operation of an in vitro diagnostic device in a faulty state, and the target historical data includes historical equipment calibration data and historical test-related data; Assign corresponding tag values ​​to the target historical data; The tag value represents the probability that the in vitro diagnostic equipment malfunctioned at the time the target historical data was collected; The target linear regression model is trained based on the target historical data and the corresponding label values ​​to obtain the target linear regression model.

13. The fault prediction method for in vitro diagnostic equipment according to claim 12, characterized in that, Assigning corresponding tag values ​​to the target historical data includes: Determine the time difference between the time the target historical data was collected and the time the fault occurred; Based on the time difference, assign corresponding label values ​​to the target historical data; Specifically, the target historical data is assigned a corresponding label value based on the negative correlation between the label value and the time difference.

14. The fault prediction method for in vitro diagnostic equipment according to claim 12, characterized in that, The process of training the linear regression model based on the target historical data and the corresponding label values ​​also includes: The linear regression model is optimized by cross-validation and using the objective loss function. The target loss function is a loss function determined based on minimizing the mean squared error, and the parameter optimization operation includes optimization operations on the feature weight coefficients and bias terms of the linear regression model.

15. The fault prediction method for in vitro diagnostic equipment according to any one of claims 1 to 6, characterized in that, Also includes: Based on the SHAP analysis method, the marginal contribution of each device's runtime sequence characteristics to the fault prediction result output by the target linear regression model is determined, and the runtime sequence characteristics of each device and the corresponding marginal contribution are visualized.

16. The fault prediction method for in vitro diagnostic equipment according to any one of claims 1 to 6, characterized in that, Also includes: Using the simple moving average method and the target Z-score set, the corresponding simple moving average line is generated; Based on the degree of deviation between the Z-score corresponding to the target data and the simple moving average, the in vitro diagnostic equipment is predicted to be faulty, and the target fault prediction result is obtained. By comparing the target fault prediction results with the fault prediction results output by the target linear regression model, it is determined whether the target linear regression model should be optimized. The target Z-score set includes the target data and the corresponding historical data, each with its own Z-score.

17. A computer device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the fault prediction method for an in vitro diagnostic device as described in any one of claims 1 to 16.

18. A computer-readable storage medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements the fault prediction method for an in vitro diagnostic device as described in any one of claims 1 to 16.

Images