Diagnostic system, automatic analyzer, and diagnostic method

Abstract

Provided is a diagnosis system for diagnosing a sensor that is provided to an automated analysis device and that outputs an analog electrical signal, the diagnosis system comprising: a memory for storing data of the electrical signals outputted by the sensor as well as the replacement history of the sensor; and a processing device for processing the data recorded in the memory. The processing device: reads out, from the memory, data of a set reference period, from among pieces of data of the electrical signal outputted by a sensor which was used in the past at the automated analysis device; computes a statistic for the data of the reference period; reads out, from the memory, data recorded in a set evaluation period, from among data pieces of the electrical signal outputted by a sensor to be diagnosed which is being used at the automated analysis device; computes a statistic for the data of the evaluation period; and assesses abnormality of the sensor to be diagnosed on the basis of the disparity found from the statistic for the reference period and the statistic for the evaluation period.

Description

[Technical field]

[0001] The present invention relates to a diagnostic system for diagnosing whether various sensors mounted in an automatic analyzer that measures trace components contained in biological samples such as blood and urine are in a normal or abnormal state, an automatic analyzer equipped with this diagnostic system, and a diagnostic method. [Background technology]

[0002] An automatic analyzer is generally made up of more than a thousand different parts. If one of the parts breaks down, it is necessary to repair or replace the broken part immediately in order to reduce the downtime of the automatic analyzer. In addition, the parts of an automatic analyzer include several types of regularly replaced parts that require regular replacement. It is also important to prevent downtime caused by breakdowns by replacing these regularly replaced parts at the appropriate time.

[0003] Patent Document 1 discloses a technology that generates a failure prediction algorithm from data regarding the occurrence of failures in an automatic analyzer, and uses this algorithm to predict failures such as damage to parts based on at least one of the calibration and quality control data of the automatic analyzer. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-536049 A Summary of the Invention [Problem to be solved by the invention]

[0005] In automated analyzers, since trace components of less than micromoles / liter (μmol / L) are generally measured, sensors are designed to be highly sensitive, and the response tendency of the sensor may vary depending on the situation. There is room for improvement in accuracy in diagnosing the status of various sensors involved in the operation control and status monitoring of each part of automated analyzers, which require high measurement accuracy, including sensors that measure such target components.

[0006] An object of the present invention is to provide a diagnostic system, an automatic analyzer, and a diagnostic method that can improve the diagnostic accuracy of a sensor in an automatic analyzer. [Means for solving the problem]

[0007] In order to achieve the above object, the present invention provides Measure the analyte contained in the sample as the measurement item. Equipped with an automatic analyzer The device is used to measure a plurality of measurement items, and a reaction product between the sample and the reagent is made to emit light by applying a predetermined voltage to the electrodes, and at least one of a current, a voltage, and a resistance generated in the electrodes by the application of the predetermined voltage is measured. A diagnostic system for diagnosing a sensor that outputs an analog electrical signal, the diagnostic system comprising: a memory for storing data of the electrical signal output by the sensor and a replacement history of the sensor; and a processing device for processing the data recorded in the memory, the processing device being configured to detect whether a sensor used in the past in the automatic analyzer has been replaced. When measuring a plurality of measurement items selected from the plurality of measurement items Of the output electrical signal data 、 reading data for a set reference period from the memory; The past output of the sensor when the selected measurement items were measured Data for the reference period The above are statistically processed together regardless of the measurement item and used as a single diagnostic index. Calculate statistical values, and the diagnostic target sensor being used in the automatic analyzer is When the selected multiple measurement items are measured Among the data of the outputted electrical signals, data recorded during a set evaluation period is read from the memory; As with the statistics for the reference period A diagnostic system is provided which calculates statistical values ​​of data for the evaluation period and judges an abnormality in the sensor to be diagnosed based on a difference between the statistical values ​​for the reference period and the statistical values ​​for the evaluation period. Effect of the Invention

[0008] According to the present invention, the diagnostic accuracy of a sensor in an automatic analyzer can be improved. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a plan view showing a schematic configuration example of an automatic analyzer to which a diagnostic system according to a first embodiment of the present invention is applied; [Diagram 2] Schematic diagram of the sensor used to measure samples in the automated analyzer shown in Figure 1. [Diagram 3] Block diagram showing the data processing flow in the automatic analyzer shown in Figure 1. [Figure 4] FIG. 1 is a diagram showing an example of time-series data of luminescence intensity measured during sample measurement. [Diagram 5] FIG. 1 is a diagram illustrating an example of time-series data of voltage values ​​measured during sample measurement. [Figure 6] FIG. 1 is a diagram illustrating an example of time-series data of current values ​​measured during sample measurement. [Figure 7] FIG. 1 is a diagram showing an example of time-series data of resistance values ​​measured during sample measurement. [Figure 8] Block diagram showing the sensor diagnosis process flow in the server [Figure 9] A flowchart showing the detailed procedure of the determination calculation process and the replacement necessity determination process in the flow of FIG. [Figure 10] A flowchart showing the detailed procedure for extracting basic data groups in the flow of FIG. 9. [Figure 11] Conceptual diagram of setting the reference period [Figure 12] FIG. 13 is a diagram illustrating an example of a setting screen for determining conditions for a sensor to be diagnosed. [Figure 13] FIG. 11 is a block diagram showing a data processing flow in a diagnostic system according to a second embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] 1. Definitions of Key Terms "Sensor to be diagnosed" In this specification, a sensor currently in use in an automatic analyzer that is to be diagnosed for abnormal or normal conditions is referred to as a diagnosis target sensor. The diagnosis target sensor includes all sensors mounted on the automatic analyzer that have electrical behavior. In other words, in addition to sensors that measure a measurement item (e.g., the concentration of a target component), sensors related to the operation control of the automatic analyzer and sensors used to monitor the status of each part of the automatic analyzer are also examples of the diagnosis target sensor. Therefore, for example, a pressure sensor that is installed in a fluid flow path and used to check the flow path status by voltage also falls under the diagnosis target sensor. Other examples of the diagnosis target sensor include a pressure sensor used to convert the liquid delivery pressure of a liquid delivery pump into an electrical signal and check it, and an impedance meter used to check the stirring strength of a stirring mechanism by impedance.

[0011] "Sensors of the past" A used sensor that has been used in an automatic analyzer and then removed from the automatic analyzer due to replacement is referred to as a past sensor. Needless to say, a past sensor may be a sensor that has been used in the automatic analyzer in which the diagnostic target sensor is used, but it may also include a sensor that has been used in an automatic analyzer other than the automatic analyzer in which the diagnostic target sensor is used. In other words, sensor data from other automatic analyzers may also be the basis for indicators for fault diagnosis.

[0012] "Reference period" A set period during which it is estimated that the sensor was in a normal state in the past is referred to as a reference period.

[0013] "Reference sample" Samples such as standard samples (calibration samples), QC samples (quality control samples), and dummy samples, which are measured by an automatic analyzer prior to the measurement of patient specimens, are referred to as reference samples. When referred to as reference samples, they shall include at least one type (one or more types) of standard samples, QC samples, and dummy samples. A standard sample is a prepared sample that is measured to create a calibration curve during calibration. A QC sample is a prepared sample that is measured during QC (quality control). A dummy sample is a predetermined sample that is measured as a preparatory operation before measuring patient specimens. Standard samples, QC samples, dummy samples, etc. are all produced in lots, and multiple samples belonging to the same lot can be sequentially measured by the same sensor of the automatic analyzer.

[0014] 2. Embodiments of the Invention In this embodiment, data of analog electrical signals output from sensors that have been used in the past by an automatic analyzer are statistically processed as indices to diagnose the currently used sensors. The sensors to be diagnosed are various sensors used for measuring analysis items of biological samples such as biochemical analyzers, immunoassays, blood coagulation time measurement devices, and ISE measurement devices, and are, for example, flow cell type sensors equipped with ion-selective electrodes. The analog electrical signals output by the sensors include not only the measured values of the analysis items of the samples (concentrations of the components to be analyzed), but also response values such as resistance values and current values generated in the electrodes when a control voltage is applied to the electrodes of the sensors. When the control target is current or resistance instead of voltage, the voltage values measured in response to the current or resistance also correspond to the electrical signals output by the sensors.

[0015] The above diagnosis can be performed artificially with manual calculations based on the data output from the sensors, but it can also be realized as a function of a computer equipped with a memory (storage media such as RAM, ROM, HHD, SSD, etc.) and a processing device (CPU, etc.). This function can also be provided to a computer incorporated in the automatic analyzer or directly connected by a cable. In addition, the same function can also be provided to a computer (server, etc.) communicably connected to the computer of the automatic analyzer via a global network or a local area network.

[0016] When using a computer, data obtained by A / D converting an analog electrical signal output from a sensor of an automatic analyzer and data on the replacement history of the sensor are recorded in a memory (e.g., HHD or SSD). The data recorded in the memory is accumulated, and both data obtained from the sensor to be diagnosed and data obtained from past sensors are recorded. Also, a diagnostic program is stored in a memory (e.g., ROM, etc.), and a processing device (e.g., CPU) processes the data recorded in the memory so as to diagnose the state of the sensor. The processes executed by the processing device at that time are roughly classified into the following first to third processes.

[0017] The first process is a process of reading out from the memory data of an electrical signal output from a past sensor during a set reference period. The second process is a process of calculating a statistical value of the data during the reference period as an index for diagnosing the sensor. The third process is a process of diagnosing the state of the sensor to be diagnosed based on the index calculated in the second process. In this third process, among the data of the electrical signal output from the sensor to be diagnosed, data recorded during the set evaluation period is read out from the memory, and a statistical value of the data during the evaluation period is calculated. Then, based on the difference obtained from the statistical value during the reference period and the statistical value during the evaluation period, it is determined whether the sensor to be diagnosed is normal or abnormal. Preferred forms of each of the first to third processes will be exemplarily described below.

[0018] - First Process - · Regarding the setting of the reference period In the first process, how to specify the reference period is important. Since a diagnostic index for the sensor to be diagnosed is generated based on data acquired by the sensor in the past during the reference period, it is essential to set the period during which the sensor was in good condition in the past as the reference period. For example, if the period during which the sensor actually operated without problems is clear from data on the inspection history by a service technician, that period can be set as the reference period, but one suitable example is to set a period that excludes a specified period before the sensor was replaced in the past as the reference period. The specified period as an excluded period to be excluded from the reference period and the length of the reference period (or the start and end periods) are set according to the settings.

[0019] The specified period as the above-mentioned exclusion period can be set arbitrarily within a period excluding periods when the sensor is clearly operating normally based on the inspection history data, but a reasonable example is to set the specified period as a certain period immediately before the old sensor is replaced with a new sensor. This is because, when a sensor is replaced due to deterioration or failure, the sensor may already be malfunctioning for a certain period immediately prior to replacement. If the length of the specified period is set appropriately, a certain degree of validity can be ensured for the identified reference period as a period during which the sensor was operating normally.

[0020] When computer processing is taken into consideration, for example, if the length of the predetermined period and the reference period are given by setting, the processing device can calculate the time going back a predetermined period from the time of the previous sensor replacement based on the past sensor replacement history, and specify the set period ending at that time as the reference period. For example, if the length of the predetermined period is 30 days (1 month) and the length of the reference period is 1 week, the 7 days from 37 days before to 31 days before the end of the past sensor usage period (time of replacement) are automatically specified as the reference period.

[0021] When collecting past sensor data from multiple automated analyzers, the timing and intervals for sensor replacement may differ depending on the automated analyzer, so the reference period may differ for each automated analyzer. In addition, in an automated analyzer that has had its sensors replaced multiple times, multiple reference periods may be set according to the number of sensors that have been used in the past.

[0022] Data Extraction The data extracted from the specified reference period becomes the basic data group of diagnostic indicators for the diagnostic target sensor. All data during the reference period can be the basic data group, or a set number of data extracted randomly or according to a set rule (e.g., at regular time intervals) can be the basic data group. If the past sensor and diagnostic target sensor are used to measure multiple measurement items (TSH, CEA, etc.), a set number of data extracted for each of the selected two or more measurement items can also be the basic data group.

[0023] In addition, the inventors of the present application and others have found that in an automatic analyzer in which a wide variety of combinations of reagents and samples are used, it is preferable to extract data obtained under conditions similar to those for measuring patient samples as a basic data group in order to improve the diagnostic accuracy of the sensor state. Specifically, an automatic analyzer generally analyzes several tens of measurement items, and therefore a large number of analytical reagent kits are set according to the measurement items. The analytical reagent kits set vary in type for each automatic analyzer. When analyzing response data from a sensor in a measurement using a reagent, the sensor response varies due to differences in the components of the analytical reagent kit. In the study of the inventors of the present application and others, there were cases in which the diagnostic accuracy of the sensor state was reduced due to this variation. In particular, it was found that the influence is significant when diagnosing the judgment of the sensor from the electrical response generated in the sensor when a command is given to the sensor, rather than the measurement data of the measurement item (measured concentration of the target component). In addition, in the operation of an automatic analyzer, a constant voltage is applied to the electrode to measure the internal standard solution in order to check the stability of the potential of the electrode of the sensor when the device is started up, but the sensor response at that time does not clearly reflect the sensor state compared to when a patient sample is measured.

[0024] Therefore, as the basic data group extracted from the reference period, data previously obtained by a sensor during measurement of at least one of a patient sample and a reference sample can be preferably used.

[0025] In addition, from the viewpoint of improving the diagnostic accuracy of the sensor condition, it may be advantageous to use data obtained under steady conditions as basic data. From this viewpoint, data obtained by a sensor in the past during QC measurement is particularly preferable as a basic data group, compared to the measurement of a patient sample in which the concentrations of coexisting components other than the measurement item vary. This is because the QC sample measured in QC is a ready-made product with stable components, and QC measurement is often performed at least once a day for each automatic analyzer, so there is no bias in the time of data generation and a sufficient number of data can be secured. In addition, in the study by the present inventors, when the response data of the sensor during the measurement of a patient sample was analyzed, cases were found in which the electrical conductivity was affected by the amount of lipids contained in the patient sample.

[0026] In addition, from the viewpoint of extracting a basic data group having higher validity as normal data obtained from a past sensor in a normal state, data from the measurement of a reference sample during a predetermined period as the above-mentioned exclusion period can be used to confirm suitability as basic data. For example, this predetermined period is set as a confirmation period, and if it can be denied from data related to the measurement of a reference sample during the predetermined period that the past sensor was in a normal state during the reference period, the data of the past sensor is not extracted as basic data. Typically, if the history of calibration, QC measurement, or dummy measurement performed during the predetermined period includes failures or abnormalities, there is a possibility that the calibration or QC failure or the abnormality of the dummy measurement occurred due to a malfunction of the sensor.

[0027] Considering computer processing, first, the processing device records data on the success or failure of the reference sample measurement in memory every time a reference sample is measured using a past sensor. Then, when extracting a basic data group, the processing device reads out from memory the history of the reference sample measurements performed during a specified period, and allows extraction of a basic data group for past sensors if all of the measurements of the reference sample are successful or normal. In this way, for example, a reference period is set only for past sensors that did not have errors in QC, etc. performed during a specified period, and data for the reference period obtained with the sensors is extracted as a basic data group (or a part thereof). Data from sensors that had errors in QC, etc. is filtered and excluded from the basic data because it is doubtful whether the sensor was in a normal state.

[0028] -Second process- In the second process, when the sensor to be diagnosed is used to measure multiple measurement items, it is desirable to extract a set number of data for each of the two or more selected measurement items from the data of the past sensor reference period, and compile all the extracted data into statistics. The multiple items to be selected can be all measurement items, or a portion of the measurement items can be selected in multiples. The statistical value calculated based on the data obtained by measuring the measurement items selected in this way becomes an index for the diagnosis of the sensor to be diagnosed. Examples of the statistical value calculated here include at least one of the average value, moving average value, median value, and standard deviation, or a value based on at least one of these values ​​(Z score, etc.).

[0029] In this case, differences in the manufacturing lot and manufacturer of the analytical reagent kit installed in each automatic analyzer may affect the magnitude of the data (raw data) of the electrical signal obtained from the sensor. In order to suppress this effect, when integrating data from multiple measurement items to perform statistics, it is preferable to normalize the data for each measurement item in advance. For example, when taking data of the electrical resistance value measured when a voltage is applied to the sensor, the measured electrical resistance value is not directly statistically processed, but the electrical resistance value for a set period is first averaged for each measurement item, and then each individual electrical resistance value is converted into a deviation (difference from the average value) and normalized. A set number of this normalized data is extracted for each measurement item, and the extracted data is statistically processed together regardless of the measurement item, to define a single index for diagnosis of the sensor to be diagnosed.

[0030] -Third Process- Data extraction and statistics In the third process, data of the sensor to be diagnosed for the set evaluation period (hereinafter, abbreviated as target data) is statistically collected in the same manner as in the second process. The statistical value of the target data is a value that reflects the current state of the sensor to be diagnosed, and the current state of the sensor to be diagnosed is diagnosed by comparing this with the statistical value of the basic data calculated in the second process. The evaluation period is a set period that ends at the present time, for example, the most recent week or month.

[0031] The method of extracting and statisticizing the target data can be the same as in the first and second processes, except that the target period for data extraction is the evaluation period, in other words, the sensor that is the output source of the extracted data is the diagnostic target sensor. For example, it is preferable to convert the individual data of the diagnostic target sensor into deviation data before statistics, as with past sensor data, and use it. In addition, all data during the evaluation period can be statistically collected as the target data group, or a set number of data extracted randomly or according to a set rule (for example, at a fixed time interval) can be statistically collected as the target data group. If the diagnostic target sensor is used to measure multiple measurement items (TSH, CEA, etc.), a set number of data can be extracted as the target data group for each of the selected two or more measurement items. In this case, it is preferable to select the measurement item selected as the basic data for the index in the first process. In addition, as in the first process, data obtained by the diagnostic target sensor during measurement of at least one of the patient sample and the reference sample can be adopted as candidates for data to be extracted as the target data group, and among them, data during QC measurement is preferable. As in the second process, examples of statistical values ​​based on the target data include at least one of the average value, moving average value, median value, and standard deviation, or a value based on at least one of these values ​​(Z score, etc.).

[0032] Diagnosis of the sensor to be diagnosed As described above, the statistical value calculated in the second process is an index obtained by statistically analyzing data (i.e., basic data) from a reference period in which past sensor data is assumed to be normal. Therefore, if the difference (the magnitude of the absolute value of the difference) between the statistical value of the target data calculated in this process and the statistical value of the basic data is large, an abnormality is suspected in the sensor to be diagnosed. Considering computer processing, for example, a first judgment value for abnormality judgment is set for the difference between the statistical value of the basic data and the statistical value of the target data. This allows the processing device to calculate the difference between the statistical value of the basic data and the statistical value of the target data, and to judge that there is an abnormality in the sensor to be diagnosed if the difference is greater than the first judgment value.

[0033] In this case, as a diagnostic algorithm by the processing device, an algorithm that simply compares the difference between the statistical values ​​with the first judgment value can be applied. In addition, a diagnostic algorithm that judges whether the difference between the data of the evaluation period and the data of the reference period is greater than the first judgment value, that is, whether the sensor to be diagnosed is abnormal, can be applied by a significant difference test based on the statistical value of the basic data and the statistical value of the target data. As a specific example, for example, the average value and standard deviation of the basic data are calculated as the statistical value of the data of the reference period. Then, for each data of the evaluation period, the deviation from the average value of the basic data is divided by the standard deviation of the basic data to calculate a Z score, and whether the sensor to be diagnosed is abnormal is judged based on a comparison between the average value of the calculated Z score and the first judgment value. In this case, the first judgment value is a value at which the hypothesis that the sensor to be diagnosed is abnormal is rejected at a significant difference level (for example, 5% probability) if the magnitude (absolute value) of the average value of the Z score is equal to or less than this value. In other words, if the magnitude of the average value of the Z score is greater than the first judgment value, an abnormality of the sensor to be diagnosed is estimated. When the magnitude of the average value of the Z scores is equal to or less than the first determination position, it is estimated that the sensor to be diagnosed is not abnormal.

[0034] The diagnosis results of the sensor to be diagnosed (including a diagnosis of normal or a diagnosis of neither abnormal nor normal, which will be described later) are displayed and output by the processing device on a UI (user interface) such as a monitor connected to a computer, and are notified to a user, etc. It is also conceivable that the UI displays a setting screen on which judgment conditions for the sensor to be diagnosed can be set, and the judgment conditions can be changed from the preset values ​​on the setting screen to adjust the diagnostic sensitivity of the sensor to be diagnosed by the processing device. The judgment conditions set on the setting screen include, for example, at least one of the above-mentioned reference period or the length of the predetermined period, the measurement item (TSH, CEA, etc.), the first judgment value, the second judgment value (described later), the data type (voltage, current, resistance, etc.), and the measurement type (measurement of a patient sample, calibration, QC measurement, etc.). For example, a user who takes a long time to procure parts due to traffic conditions, etc., may need to set stricter judgment conditions to increase the diagnostic sensitivity and order a new sensor early. By making the judgment conditions adjustable, it is possible to flexibly respond to such user circumstances.

[0035] In addition to the judgment to estimate the sensor to be diagnosed, a judgment to estimate the normality of the sensor to be diagnosed can also be executed. In this case, for example, a second judgment value having an absolute value smaller than the first judgment value is set. This allows the processing device to judge that the sensor to be diagnosed is normal when the difference between the basic data group and the target data group is smaller than the second judgment value based on the statistical value of the basic data and the statistical value of the target data. The judgment to estimate the normality of the sensor to be diagnosed can also apply a significant difference test in the same way as the judgment to estimate an abnormality. For example, as described above, the basic data is taken into account and each target data is converted into a Z score, and whether the sensor to be diagnosed is normal is judged based on a comparison between the average value of the Z scores and the second judgment value. In this case, the second judgment value is a value at which the hypothesis that the sensor to be diagnosed is normal is rejected at a significant difference level (for example, 5% probability) if the magnitude (absolute value) of the average value of the Z scores is equal to or greater than this value. In other words, if the magnitude of the average value of the Z scores is smaller than the second judgment value, the sensor to be diagnosed is estimated to be normal. If it can be concluded that the sensor to be diagnosed is normal and does not need to be replaced, this is useful for sensor maintenance planning and inventory management.

[0036] Furthermore, as a result of the above judgment, there may be cases where the sensor to be diagnosed is not judged to be abnormal or normal, that is, the data of the basic data group and the data of the target data group deviate to such an extent that they cannot be said to be abnormal or normal. In such cases, the handling can be flexibly set according to the operational philosophy of the automatic analyzer, such as judging that replacement is not necessary (normal) from the viewpoint of avoiding stoppage of the automatic analyzer for maintenance as much as possible, or judging that replacement is necessary (abnormal) from the viewpoint of prioritizing measurement accuracy.

[0037] Furthermore, when a deviation occurs between the data of the basic data group and the data of the target data group that cannot be said to be abnormal or normal, the cause may be an abnormality of the sensor to be diagnosed, or a defect in a component of the automatic analyzer other than the sensor to be diagnosed. Therefore, it is possible to notify the user or serviceman of the possibility of a defect in various components related to the measurement, including the sensor to be diagnosed, without concluding that the sensor to be diagnosed is abnormal or normal, and to encourage the user or serviceman to make a decision to perform an inspection to identify the cause of the malfunction or to wait and see. Considering computer processing, when the difference between the statistical value of the basic data and the statistical value of the target data is equal to or less than the first judgment value and equal to or more than the second judgment value, the sensor to be diagnosed is judged by the processing device taking into account the factors of the components other than the sensor to be diagnosed. As a specific example, when the difference between the statistical value of the basic data and the statistical value of the target data is equal to or less than the first judgment value and equal to or more than the second judgment value, a form can be adopted in which the processing device notifies the user, etc. through the UI when the difference between the statistical value of the basic data and the statistical value of the target data is equal to or less than the first judgment value and equal to or more than the second judgment value. The notification at that time may simply be in the form of a text display of the judgment result that the difference between the statistical value of the basic data and the statistical value of the target data is equal to or less than the first judgment value and equal to or more than the second judgment value. In addition, it may be in the form of a text display of a comment encouraging inspection of components related to the measurement, including the sensor to be diagnosed, to notify the user of specific recommended measures.

[0038] In the above example, the abnormality and normality of the sensor to be diagnosed are determined by a one-tailed t-test, but a two-tailed t-test may also be applied. In any case, the significance test by comparison with the basic data as described above is useful as an algorithm for determining whether the sensor of the automatic analyzer is abnormal or normal. In general, an automatic analyzer used for measuring patient samples is not operated while leaving the sensor in an abnormal state, so it is difficult to collect data on the sensor in an abnormal state, and it is difficult to define the abnormal state based on the abnormal data. In response to this, the state of the sensor to be diagnosed can be suitably diagnosed by defining a reference period in which a normal state is estimated for the past sensor, and performing a significance test using basic data sampled from the reference period.

[0039] 3. First Example The present invention can be applied to an automatic analyzer. Examples of the analysis unit mounted on the automatic analyzer include a biochemical analyzer and an immunoanalyzer. However, this is only an example, and the present invention is not limited to the embodiment described below, and can be widely applied to an automatic analyzer equipped with an analysis unit that analyzes a sample based on the reaction result with a reagent. For example, an automatic analyzer equipped with a mass spectrometer used in clinical testing and a coagulation analyzer that measures blood coagulation time may also be included in the scope of application. In addition, the present invention can be applied to a composite automatic analyzer equipped with a plurality of these various analysis units, and an automatic analysis system including at least one automatic analyzer. A specific embodiment in which the present invention is applied to the diagnosis of a sensor for measuring a measurement item of an automatic analyzer will be described below with reference to the drawings.

[0040] -Automatic analyzer- Fig. 1 is a plan view showing a schematic configuration example of an automatic analyzer to which a diagnostic system according to a first embodiment of the present invention is applied. The automatic analyzer 1 shown in Fig. 1 includes a rack transport line 2, an incubator disk 3, a first transport mechanism 4, a holding member 5, a sample dispensing nozzle 6, a reagent disk 7, a reagent dispensing nozzle 8, a second transport mechanism 9, an analysis unit 10, and a control device 20.

[0041] The rack transport line 2 is a unit that transports the rack R, and transports the rack R to a sample dispensing position by a sample dispensing nozzle 6. A plurality of sample containers C1 for holding samples can be placed on the rack R. The example in Fig. 1 illustrates a configuration for transporting samples along a line, but there is also a case where a disk-shaped transport unit that rotates to transport the samples is provided.

[0042] The incubator disk 3 is a disk on which the reaction vessels C2 are placed, and multiple reaction vessels C2 can be placed in a ring shape. The incubator disk 3 is rotated by a drive device (not shown), and can move any reaction vessel C2 to multiple predetermined positions including the dispensing position by the sample dispensing nozzle 6.

[0043] The first transfer mechanism 4 is a unit that transfers the sample dispensing chip and the reaction vessel C2, is operable in the three axial directions of X, Y, and Z, and transfers the sample dispensing chip and the reaction vessel between a predetermined position of the stirrer mechanism M and the incubator disk 3, the waste hole D and the chip mounting position P, and the holding member 5. The stirrer mechanism M is a unit that stirs the sample contained in the reaction vessel C2. The waste hole D is a hole for discarding the used sample dispensing chip and reaction vessel. The chip mounting position P is a position where the sample dispensing chip is mounted on the sample dispensing nozzle 6.

[0044] The holding member 5 is a member that holds the sample dispensing chip and the reaction vessel C2, and a plurality of unused reaction vessels C2 and sample dispensing chips are installed. The unused reaction vessel C2 held by the holding member 5 is transferred by the first transfer mechanism 4 and installed at a predetermined position on the incubator disk 3. Similarly, the unused sample dispensing chip held by the holding member 5 is transferred by the first transfer mechanism 4 and installed at the chip mounting position P.

[0045] The sample dispensing nozzle 6 is a unit that aspirates and discharges the sample. The sample dispensing nozzle 6 is configured to be rotatable and vertically movable, rotates and moves above the chip mounting position P and descends, and presses and mounts the sample dispensing chip at the tip of the sample dispensing nozzle 6. The sample dispensing nozzle 6 with the sample dispensing chip mounted thereon moves above the sample container C1 placed on the rack R and descends, and aspirates a predetermined amount of the sample held in the sample container C1. The sample dispensing nozzle 6 that has aspirated the sample moves above the incubator disk 3 and descends, and discharges the sample into the unused reaction vessel C2 held on the incubator disk 3. When the sample discharge is completed, the sample dispensing nozzle 6 moves above the waste hole D and discards the used sample dispensing chip from the waste hole D.

[0046] The reagent disk 7 is a unit on a disk where a plurality of reagent containers C3 are installed. A disk cover 7a (the left part is partially broken in FIG. 1) is provided on the upper part of the reagent disk 7, and the inside of the reagent disk 7 is kept at a predetermined temperature. An opening 7b is provided in the disk cover 7a at a portion close to the incubator disk 3.

[0047] The reagent dispensing nozzle 8 is a unit that aspirates and discharges a reagent. Like the sample dispensing nozzle 6, the reagent dispensing nozzle 8 can rotate and move up and down, and rotates to above the opening 7b of the disk cover 7a and then descends, inserting the tip of the reagent dispensing nozzle 8 into a predetermined reagent container C3 and aspirating a predetermined amount of reagent. Next, the reagent dispensing nozzle 8 rises and rotates to above a predetermined position on the incubator disk 3, and discharges the reagent into the reaction container C2 containing the sample.

[0048] The reaction vessel C2 into which the sample and reagent have been discharged is moved to a predetermined position by the rotation of the incubator disk 3, and is transported to the stirring mechanism M by the first transport mechanism 4. The stirring mechanism M stirs and mixes the sample and reagent in the reaction vessel C2 by rotating the reaction vessel C2. After stirring, the reaction vessel C2 is returned to a predetermined position on the incubator disk 3 by the first transport mechanism 4.

[0049] The second transport mechanism 9 is a unit that transfers the reaction vessel C2 between the incubator disk 3 and the analysis unit 10, and is configured to be capable of rotation and vertical movement. This second transport mechanism 9 grasps the reaction vessel C2 that has been returned to the incubator disk 3 after mixing the sample and reagent and has passed a predetermined reaction time, moves up, and transfers it to the analysis unit 10 by rotation.

[0050] The analysis unit 10 is a unit that measures measurement items such as specific biological components and chemical substances contained in the reaction liquid in the reaction vessel C2. The diagnostic object sensor to be diagnosed in this embodiment is used in this analysis unit 10. The sensor will be described later.

[0051] The control device 20 is a computer including a memory 21 such as a RAM, a ROM, an HDD, an SSD, a processing device 22 such as a CPU, a timer, etc. This control device 20 controls each device mounted on the automatic analyzer 1, and records and processes data input from the analysis unit 10, etc. The control device 20 is housed, for example, in the exterior of the automatic analyzer, or is installed outside the exterior and directly connected to the main body of the automatic analyzer by wire or wirelessly.

[0052] The control device 20 is connected to the server 30 via the communication interface 23, the network NW, and the communication interface 33. The server 30 is also a computer configured to include a memory 31 such as a RAM, a ROM, an HDD, an SSD, a processing device 32 such as a CPU, a timer, etc. In this embodiment, a diagnostic function for a sensor to be diagnosed used in the automatic analyzer 1 is installed in the server 30. The server 30 records data acquired by the automatic analyzer 1 or a plurality of automatic analyzers including the automatic analyzer 1 in the memory 31, and processes the data recorded in the memory 31 with the processing device 32 to diagnose the sensor to be diagnosed of the automatic analyzer 1 (described later).

[0053] -Sensor- FIG. 2 is a schematic diagram of a sensor used for measuring a sample in the automatic analyzer shown in FIG. 1. The analysis unit 10 of the automatic analyzer 1 is equipped with a flow cell type sensor 11. The sensor 11 is configured to include a flow cell 12 and three electrodes (reference electrode 13, counter electrode 14, working electrode 15) provided inside the flow cell 12. The three electrodes are controlled to a target voltage by a potentiostat 16. When a reaction product RP between the sample and the reagent is collected on the reference electrode 13, the reaction product RP emits light when a specific voltage is applied between the reference electrode 13 and the working electrode 15 by the potentiostat 16. The luminescence intensity of the reaction product RP is detected by a photomultiplier tube 17 arranged on the opposite side of the reference electrode 13 across the flow cell 12 (upper side in FIG. 2). The luminescence intensity detected by the photomultiplier tube 17 is digitized by the A / D converter 18, and is recorded as raw data of the measured value of the measurement item in the memory 21 (or the memory of the sensor 11) together with the measurement date and time through the raw data recording process P1. At that time, the current value, voltage value, and resistance value generated between the counter electrode 14 and the working electrode 15 by applying a voltage between the reference electrode 13 and the working electrode 15 are measured by the potentiostat 16. In this embodiment, not only the output of the photomultiplier tube 17 but also the voltage value applied to the reference electrode 13 and the working electrode 15, and the current value, voltage value, and resistance value generated between the counter electrode 14 and the working electrode 15 are recorded in the raw data recording process P1.

[0054] -Operation of automatic analyzer- In order to perform highly accurate qualitative and quantitative analysis of the analyte components contained in unknown patient samples, calibration and QC measurements are performed at appropriate times. For example, in the case of quantitative analysis, the automatic analyzer 1 is operated daily according to the following work procedures (1)-(5).

[0055] (1) Equipment startup First, the power is turned on to start up the automatic analyzer 1. Then, the reagent container C3 is installed and initially filled with reagent, the temperature inside the reagent disk 7 is adjusted, a constant voltage is applied to the electrodes to continuously measure the internal standard solution, and it is checked whether the potential of the electrodes of the sensor 11 is stable, and maintenance is performed as necessary.

[0056] (2) Calibration A high-concentration standard sample and a low-concentration standard sample, in which the concentration of the measurement item (analyte component) is known, are measured. From these measurements, a relationship equation (calibration curve) between the concentration of the measurement item and the output of the sensor 11 (photomultiplier tube 17) is created. However, the frequency of calibration differs depending on the measurement item, and for example, calibration for each measurement item is performed periodically (for example, monthly) in turn.

[0057] (3) QC measurement Multiple QC samples with different concentration levels, each with a known range of possible concentrations of the measurement item, are measured, and the concentration of the measurement item in the QC sample is calculated using the calibration curve created during calibration. The calibration curve is confirmed to be appropriate by checking whether the calculated concentration is within the known concentration range of the QC sample. QC measurements are performed frequently, as they are positioned as a condition check to ensure the measurement results of patient samples. For example, QC measurements are performed 1-3 times a day in parallel for multiple measurement items.

[0058] (4) Measurement of patient samples A patient sample with an unknown concentration of a measurement item is measured, and the concentration of the measurement item is calculated using a calibration curve. Prior to this patient sample measurement, a so-called background measurement or dummy measurement may be performed to check the state of the automatic analyzer 1.

[0059] (5) Equipment shutdown If necessary, each part of the automatic analyzer 1 is cleaned or inspected, and the automatic analyzer 1 is shut down by turning off the power.

[0060] In addition, in the measurement of standard samples, QC samples, or dummy samples, including patient specimens, in the sensor 11, the processes of cleaning the flow cell 12, conditioning, and detecting the luminescence intensity are repeatedly performed. Since the appropriate values of the voltage applied to the electrodes are different in each of these processes, the voltage applied to the electrodes by the potentiostat 16 is complexly controlled in a single measurement. For example, in the cleaning process, a voltage of a specific pattern is applied to the electrodes for a certain period of time with the cleaning liquid being supplied to the flow cell 12 so that the reaction product RP from the previous measurement does not remain in the sensor 11. In the conditioning process, a voltage of a pattern different from that in the cleaning process is applied to the electrodes for a certain period of time with an auxiliary reagent being supplied to the flow cell 12 in order to bring the electrodes into a state suitable for measurement. In the detection process, the voltage required for the luminescence reaction of the reaction product RP captured by the reference electrode 13 is applied. Thus, in the sensor 11, a complex voltage pattern is continuously and precisely repeatedly applied to the electrodes along with the measurement of the sample.

[0061] -Data processing flow (automatic analyzer)- FIG. 3 is a block diagram showing the processing flow in the automatic analyzer shown in FIG. 1. Data from the sensor 11, the mechanism unit 91, the sample data reading device 92, the reagent data reading device 93, and the UI (user interface) 94 are input to the control device 20 of the automatic analyzer 1.

[0062] Data recorded in the raw data recording process P1 (FIG. 2) is input to the control device 20 from the sensor 11 at any time. The data input from the sensor 11 to the control device 20 includes not only the data at the time of measuring the patient specimen, but also the data for each measurement type, such as QC measurement and calibration, and further the data at the time of dummy measurement performed as a preparation operation immediately before the measurement of the patient specimen.

[0063] The mechanism unit 91 is a general term for each piece of hardware (such as the sample dispensing nozzle 6 and the incubator disk 3) mounted on the automatic analyzer 1. The data input from this mechanism unit 91 to the control device 20 is, for example, log data such as the operation timing, operation amount, and current value of each motor, the signals of the sensors used for controlling each motor, and the opening / closing timing and current value of the fluid valve.

[0064] The sample data reader 92 is a device that reads registered data of a sample (e.g., a barcode or RFID reader) and is provided in the automatic analyzer 1. A storage medium such as a barcode or RFID is attached to the sample container C1, and the sample data recorded on the storage medium is read by the sample data reader 92. The data read by the sample data reader 92 and input to the control device 20 is, for example, a sample ID.

[0065] The reagent data reader 93 is a device that reads registration data of a reagent (e.g., a barcode or RFID reader) and is provided in the automatic analyzer 1. A storage medium such as a barcode or RFID is attached to the reagent container C3, and the reagent data recorded on the storage medium is read by the reagent data reader 93. The data read by the reagent data reader 93 and input to the control device 20 includes, for example, the reagent ID, lot number, expiration date, etc.

[0066] The UI 94, which is comprised of a monitor and an input device and allows the user to view data and input data to the control device 20, is provided in the automatic analyzer 1. Various data are input through the UI 94, and for example, data relating to the reagent such as the ID of the reagent used in the sample measurement, the lot number, expiration date, on-board expiration date, and required remaining amount are input to the control device 20. In addition, data relating to the sample such as the sample ID, the measurement type for the sample ID (patient sample measurement, QC measurement, calibration, dummy measurement, etc.), the measurement items, etc. are input to the control device 20.

[0067] The various data input to the control device 20 as described above is processed in real time by the processing device 22 and transmitted as a log file to the server 30 via the communication interface 23. The processes executed by the processing device 22 include, for example, measurement data conversion process P2 and measurement data recording process P3 for measurement data. Other processes executed by the processing device 22 include operation log recording process P4, reagent data recording process P5, sample data recording process P6, calibration data recording process P7, quality control data recording process P8, maintenance history recording process P9, and log file generation process P10. Each process will be described in turn below.

[0068] Measurement data conversion processing In the measurement data conversion process P2, the processing device 22 converts the measurement values ​​(raw data) input from the sensor 11 into effective values. The measurement values ​​input from the sensor 11 are each raw data of emission intensity, current value, voltage value, and resistance value.

[0069] Here, FIG. 4 shows an example of time series data of luminescence intensity measured during sample measurement, FIG. 5 shows an example of time series data of voltage value, FIG. 6 shows an example of time series data of current value, and FIG. 7 shows an example of time series data of resistance value. The horizontal axis of each figure corresponds to discrete time, and the measured values ​​for each discrete time are plotted. During sample measurement, the control device 20 applies a voltage to the electrodes at a specific timing from the start of measurement to obtain data. In the examples shown in FIG. 4 to FIG. 7, a predetermined voltage is applied to the reference electrode 13 and the working electrode 15 at the 41st point for all data.

[0070] Under this control characteristic, in the measurement data conversion process P2, the processing device 22 converts the measurement value (raw data) input from the sensor 11 for each measurement into a valid value using the following two formulas pre-stored in the memory 21 (e.g., ROM).

[0071]

number

[0072]

number

[0073] Measurement data recording and processing In the measurement data recording process P3, the processing device 22 assigns a measurement ID to each measurement, and records the raw data and valid values ​​of the measurement values ​​in the memory 21 in association with the measurement ID.

[0074] - Operation log recording process In an operation log recording process P4, the processing device 22 records the operation logs input from the mechanism unit 91 and the sensor 11 in the memory 21. The operation logs input from the mechanism unit 91 etc. include, for example, the operation timing and operation amount of each motor, the current value of the motor, the signal of the sensor for controlling the operation of the motor, the opening and closing timing and current value of the fluid valve, etc.

[0075] Reagent data recording and processing In the reagent data recording process P5, the processing device 22 compares the reagent data input from the reagent data reader 93 with condition data previously recorded in the memory 21, and records the reagent in the memory 21 as a usable reagent if it matches the condition data. The condition data with which the reagent data is compared includes the reagent ID, lot number, expiration date, on-board expiration date, required remaining amount, etc., and is input by the UI 94 or another computer, input to the control device 20 via the communication interface 23, and recorded in the memory 21. Also, in the reagent data recording process P5, the processing device 22 records the history of the reagent used for each measurement ID in the memory 21. As a result, the raw data and valid values ​​of the measurement values ​​are linked to the data of the reagent used in the measurement via the measurement ID.

[0076] -Sample data recording and processing In sample data recording process P6, the processing device 22 compares the sample data input from the sample data reader 92 with condition data previously recorded in memory 21, and if the condition data matches, records the sample in memory 21 as a measurable sample and performs measurement at the appropriate time. The condition data with which the sample data is compared includes the sample ID, measurement type (QC measurement, measurement of a patient sample, etc.), measurement items, etc., and is input by the UI 94 or another computer, input to the control device 20 via the communication interface 23, and recorded in memory 21.

[0077] Calibration data recording and processing In the calibration data recording process P7, each time calibration is performed, the processing device 22 records the calibration execution history for calculating the concentration of the measurement item of the patient sample in the memory 21. Specifically, the calculated calibration curve, the date and time of the calibration, the measurement item, the success or failure of the calibration, etc. are recorded in the memory 21 as the calibration execution history.

[0078] -Accuracy control data recording and processing In quality control data recording processing P8, the processing device 22 stores the implementation history of the QC measurement in the memory 21. Specifically, the measurement items, the QC judgment results (success or failure), etc. are recorded in the memory 21 as the implementation history of the QC measurement.

[0079] · Maintenance history record processing In a maintenance history recording process P9, the processor 22 records the maintenance history in the memory 21. Specifically, the history of inspections by a user, sensor replacements by a service technician, and the like are recorded in the memory 21 as the maintenance history.

[0080] Log file generation process In the log file generation process P10, the processing device 22 aggregates data necessary for fault diagnosis of the sensor 11 for each measurement from among the data stored in the memory 21 to generate a log file. The processing device 22 also transmits the log file to the server 30 via the communication interface 23 and the network NW. Since a log file is created for each measurement and uploaded to the server 30 successively, the server 30 accumulates not only data of measurements made by the sensor 11 currently being used in the automatic analyzer 1, but also data of measurements made by the sensors 11 that were used in the automatic analyzer 1 in the past.

[0081] -Sensor diagnosis processing flow (server)- FIG. 8 is a block diagram showing a sensor diagnosis process flow by the processing device of the server. In this embodiment, the diagnostic function of the sensor 11 of the automatic analyzer 1 is executed by the server 30, and the server 30 constitutes a diagnostic system. When the log file is received from the automatic analyzer 1, the processing device 32 records the log file in the memory 31 in a log file storage process P21. In the subsequent determination data extraction process P22, the processing device 32 extracts data necessary for the diagnosis of the sensor 11 from the log file stored in the memory 31, such as the effective value of the measurement value, the QC result, the calibration result, and the replacement history of the sensor 11. In the subsequent determination data storage process P23, the processing device 32 stores the extracted data necessary for the diagnosis in the memory 31. Thereafter, when performing the diagnosis, the processing device 32 reads out the necessary data from the memory 31 again in a determination calculation process P24. Then, based on the calculation result of the determination calculation process P24, the processing device 32 diagnoses the sensor 11 in a replacement necessity determination process P25 to determine whether or not replacement is necessary. The judgment result is transmitted to the automatic analyzer 1 via the communication interface 33 and the network NW, and is notified to a user or the like via the display of the UI 94. It is also possible to display the diagnosis result on the UI (user interface) 34 of the server 30. The UI 34 of the server 30 is similar to the UI 94 of the control device 20 of the automatic analyzer 1.

[0082] -Judgment calculation process- Fig. 9 is a flowchart showing the detailed procedures of the determination calculation process and the replacement necessity determination process in the flow of Fig. 8, and Fig. 10 is a flowchart showing the detailed procedures of the basic data group extraction in the flow of Fig. 9. Here, an example will be described in which the state of the electrodes of the flow cell type sensor 11 is diagnosed by the processing device 32. The determination calculation process corresponds to the procedures of steps 100-130 in the flow of Fig. 9, and also corresponds to the first and second processes explained in the embodiment of the invention.

[0083] Step 100 9 starts, the processing device 32 first reads out from the memory 31 the determination data obtained by the sensor 11 in the past, and extracts from the determination data a basic data group that will be used as the basis for the diagnostic index of the sensor 11 currently being used by the automatic analyzer 1 (step 100). The basic data group is determined by setting a reference period during which the automatic analyzer 1 estimates that the sensor 11 operated normally in the past under a predetermined algorithm, and extracting data obtained by the sensor 11 in the reference period in the past as the basic data group.

[0084] Step 100 will be described in detail with reference to Fig. 10. Here, an example will be described in which the previous replacement date of sensor 11 is set as day 0, and the "predetermined period" described in the embodiment of the invention is set to 30 days from -30 to 0, and the "reference period" is set to one week from -37 to -31.

[0085] Step 101 In the procedure of step 100, the processing device 32 first reads out from the memory 31 the determination data for 37 days immediately before the sensor 11 was replaced, such as the QC result, the calibration result, and the effective value of the measurement value (step 101). For convenience of explanation, this 37 days' worth of data on the sensor 11 is called a "data set." At this time, if the server 30 also collects data on other automatic analyzers other than the automatic analyzer 1, the processing device 32 also reads out the determination data of the sensor 11 in the past acquired by the other automatic analyzer. In addition, depending on the automatic analyzer, the sensor 11 may have been replaced multiple times. In that case, multiple data sets (i.e., data sets for the number of sensors 11 that have been replaced in the past) are read out by one automatic analyzer (see FIG. 11).

[0086] Step 102 Next, the data of the QC measurements for 30 days (-30th to 0th days) immediately prior to replacement, which are included in the data set, is referenced, and it is confirmed that the data does not include any data that has failed QC (step 102). The success or failure of the QC measurements is determined from data recorded in the process of collecting log files in the automatic analyzer 1. When determining the success or failure of the QC in the server 30, for example, a Z score (described later) is calculated from the measurement data of the QC measurements, and a QC with a Z score outside the range of ±3 can be determined to have failed. In step 102, the processing device 32 excludes data sets that have failed QC at least once in the 30 days prior to replacement from the source of extraction of basic data. This prevents a basic data group from being extracted from a data set of a past sensor 11 that has failed QC in the 30 days prior to replacement.

[0087] Step 103 Furthermore, the calibration data for the 30 days (-30th to 0th days) immediately prior to replacement included in the data set is referenced to confirm that no data resulting from calibration failure is included (step 103). The success or failure of calibration is also determined from data recorded in the process of collecting log files in the automatic analysis device 1. When determining the success or failure of the configuration in the server 30, for example, it can be performed using an algorithm similar to that used to determine the success or failure of QC. In step 103, the processing device 32 excludes data sets that have failed calibration at least once in the 30 days prior to replacement from the source of extraction of basic data. A basic data group will also not be extracted from a data set of a past sensor 11 that has failed calibration in the 30 days prior to replacement.

[0088] In this embodiment, the basic data group is filtered based on the success or failure of QC and calibration, but it is also possible to filter the basic data group based on the results of dummy measurements. In this case, the condition for the basic data group can be, for example, that the valid value or variation of the measurement value of the dummy measurement falls within a specified range.

[0089] Step 104 As described above, filtering is performed based on the success or failure of QC and calibration in the 30 days prior to replacement, and only data sets with no history of failure in QC or calibration are extracted, and a basic data group is extracted from the data from days -37 to -31 of the extracted data sets (step 104). The data extracted as the basic data group is a set number (e.g., 50) of data randomly sampled from the data in the period from days -37 to -31. The measurement type of the sampled data is at least one type selected by setting from among QC measurement, calibration, patient sample measurement, and dummy measurement. If there are multiple measurement items, data for multiple measurement items can be mixed and selected.

[0090] Step 110 9, the processing device 32 extracts effective values ​​of the basic data group extracted in step 100 (step 110). Here, for example, it is assumed that the effective values ​​of the resistance values ​​between the counter electrode 14 and the working electrode 15 measured at each measurement of the basic data group are extracted.

[0091] Step 120 Next, the processor 32 uses all the data (Xall) extracted in step 110 to calculate the average value (XallAve) and the standard deviation (XallSD) (step 120).

[0092] Step 130 After calculating the average value and standard deviation, the processing device 32 calculates (step 130) a Z-score (ZscoreSTD) for each data item (Xstd) of all data items (Xall) extracted in step 110. The Z-score (ZscoreSTD) is calculated by subtracting the average value (XallAve) calculated in step 120 from each data item (Xstd) and dividing the result by the standard deviation (XallSD), as shown in the following formula.

[0093]

number

[0094] The processing of the basic data group is completed at step 130. The procedure up to step 130 corresponds to the processing contents of the determination calculation process P24.

[0095] - Replacement necessity determination process - After step 140, a replacement necessity determination process P25 is performed based on data from the sensor to be diagnosed, that is, the sensor 11 currently being used in the automatic analyzer 1. This replacement necessity determination process P25 corresponds to the third process described in the embodiment of the invention.

[0096] Step 140 The processing device 32 extracts a target data group to be compared with the basic data group for diagnosing the sensor to be diagnosed from the data of the sensor to be diagnosed (step 140). The data extracted as the target data group is a set number of data (e.g., 50 or more of the basic data group) randomly sampled from the data of the most recent set period (e.g., one week up to the present). The data type to be sampled is data corresponding to the basic data group, and in this embodiment, is the effective resistance value. The measurement type of the data to be sampled is at least one type selected from QC measurement, calibration, patient sample measurement, and dummy measurement in accordance with the extraction conditions of the basic data group. When there are multiple measurement items, data for multiple measurement items can be mixed and selected in accordance with the extraction of the basic data group.

[0097] Step 150 Next, for each piece of data (XJDG) in the extracted target data group, a Z-score (ZscoreJDG) is calculated using the following formula (step 150). The following formula is an example of calculating a Z-score (ZscoreJDG) for each piece of data (XJDG) in the target data group by subtracting the average value (XallAve) of the basic data group and dividing by the standard deviation (XallSD) of the basic data group.

[0098]

number

[0099] Step 160 After calculating the Z-score (ZscoreJDG) for each data (XJDG) of the target data group, the processing device 32 executes a first significant difference test (step 160). In this embodiment, a one-tailed t-test is exemplified as the first significant difference test. In the first significant difference test, for example, the average value of the Z-score (ZscoreJDG) calculated in step 150 is calculated, and if the average value is smaller than the first judgment value, the test is performed at a significance level of 5% based on the hypothesis that the sensor 11 needs to be replaced. In other words, if the average value of the Z-score (ZscoreJDG) is smaller than the first judgment value and there is a significant difference, it means that the state of the sensor 11 is worthy of replacement.

[0100] In this embodiment, a negative value is set to the first judgment value on the assumption that the values ​​of the target data group decrease with respect to the values ​​of the basic data group according to the degree of deterioration of the sensor 11. Therefore, when the average value of the Z-score (ZscoreJDG) is smaller than the first judgment value, it means that the difference between the data of the target data group and the data of the basic data group is large.

[0101] Step 170 After completing the first significant difference test, the processing device 32 executes a second significant difference test (step 170). In this embodiment, a one-tailed t-test is exemplified as the second significant difference test. In the second significant difference test, for example, a test is performed at a significance level of 5% based on the hypothesis that if the average value of the Z score (ZscoreJDG) is greater than the second judgment value, the sensor 11 does not need to be replaced. In other words, if the average value of the Z score (ZscoreJDG) is greater than the second judgment value and there is a significant difference, it means that the state of the sensor 11 is such that replacement is not necessary.

[0102] In this embodiment, it is assumed that the value of the target data group decreases with respect to the value of the basic data group according to the degree of deterioration of the sensor 11, and a negative value is also set for the second judgment value. However, the second judgment value has an absolute value smaller than that of the first judgment value and is closer to 0. Therefore, the average value of the Z score (ZscoreJDG) being larger than the second judgment value means that the difference between the data of the target data group and the data of the basic data group is small.

[0103] Step 180 After completing the second significant difference test, the processing device 32 determines whether or not the sensor 11 needs to be replaced based on the results of the first significant difference test and the second significant difference test (step 180). Specifically, if the result of the first significant difference test is significant and the result of the second significant difference test is not significant, the processing device 32 determines that the sensor 11 needs to be replaced. If the result of the first significant difference test is not significant and the result of the second significant difference test is significant, the processing device 32 determines that the sensor 11 does not need to be replaced. If the results of both the first significant difference test and the second significant difference test are not significant, the processing device 32 determines that there is an abnormality in some part of the automatic analysis device 1, including the sensor 11. If the results of both the first significant difference test and the second significant difference test are significant and contradictory, the processing device 32 does not determine whether or not the sensor 11 needs to be replaced.

[0104] Step 190 Finally, the processing device 32 outputs the judgment result via the communication interface 33, and notifies the user, etc., via the UI 94 of the automatic analyzer 1, for example, to end the diagnosis process of FIG. 9 (step 190). The notification destination of the judgment result is not limited to the UI 94 of the automatic analyzer 1, but may be other UIs, for example, the UI 34 of the server 30 or the UI of a computer of another service center. For example, if the first significant difference test shows a significant difference and the second significant difference test shows no significant difference, a text message is displayed on the UI 94 indicating that the sensor 11 needs to be replaced, and the user, etc. is prompted to take action to replace the sensor. If the first significant difference test shows no significant difference and the second significant difference test shows a significant difference, for example, a text message is displayed on the UI 94 indicating that the sensor 11 does not need to be replaced, and the user, etc. is notified. If both the first significant difference test and the second significant difference test show no significant difference, for example, a text message is displayed on the UI 94 indicating that some part, including the sensor 11, is suspected to be abnormal, and the user, etc. is prompted to take action. If both the first and second significant difference tests indicate a significant difference, for example, an error is displayed on the UI 94 as a diagnostic result.

[0105] -Judgment condition setting screen- Fig. 12 is a diagram showing an example of a setting screen for judgment conditions for a sensor to be diagnosed. The setting screen in this figure is displayed on the UI 34 (Fig. 8) of the server 30. It is also possible to configure it to be displayed on the UI of a computer that can access the server 30, such as the UI 94 of the control device 20 of the automatic analyzer 1. The example in Fig. 12 is just one example, and it is also possible to set condition items other than the items exemplified in this figure. Furthermore, the setting screen in Fig. 12 can be shared by all automatic analyzers connected to the server 30, or can be prepared for each ID of an automatic analyzer.

[0106] In the setting screen shown in FIG. 12, the measurement items can be set in the area at the top of the screen labeled "Target Measurement Items." The basic data group and target data group described above are extracted from the data acquired by the measurement of the measurement items set here. The figure shows an example of a screen configured to select from among preset 1, preset 2, dummy measurement, and manual. For example, when preset 1 is selected, standard prescribed measurement items are automatically set by the processing device 32, when preset 2 is selected, predefined measurement items with strict judgment are automatically set, and when dummy measurement is selected, the measurement of a dummy sample is automatically set. The prescribed measurement items of presets 1 and 2 can be default items or can be set by the user. When manual is selected, the user can arbitrarily set at least one measurement item from the prepared options such as TSH and CEA.

[0107] Additionally, in the area at the bottom of the screen labeled "Target data period," parameters related to the extraction of basic data groups can be set. In the field labeled "QC / Calibration failure confirmation period," the period of time for the creation date and time of data for checking the success or failure of QC or calibration can be specified by entering a numerical value. Additionally, in the field labeled "Zscore calculation period," the period of time for the creation date and time of data used to calculate Z scores (ZscoreSTD, ZscoreJDG) can be specified by entering a numerical value. The numerical values ​​for the period displayed in FIG. 12 are, for example, the currently set values ​​(or default values), and can be changed as desired and the settings saved.

[0108] -effect- (1) As described above, in this embodiment, data from a reference period during which the sensor 11 used in the past in the automatic analyzer 1 is estimated to be operating normally is extracted as a basic data group, and the sensor 11 currently being used in the automatic analyzer 1 is diagnosed by comparing with the basic data group. In this way, by comparing with data that is statistically estimated to be in a normal state, the sensor to be diagnosed by the automatic analyzer 1 can be diagnosed with high accuracy. This makes it possible to preventively reduce downtime of the automatic analyzer 1 due to deterioration or failure of the sensor 11.

[0109] Furthermore, the basic data group is successively accumulated and updated as the automatic analyzer 1 operates, and changes in the operation of the automatic analyzer 1 and the lots of consumables are also reflected in the basic data group. Therefore, the sensor 11 can be diagnosed in a flexible manner in response to the conditions of the automatic analyzer 1 during operation.

[0110] (2) By setting the reference period excluding the specified period before the sensor 11 was replaced in the past (for example, immediately before replacement), the validity of the data in the reference period being normal data is ensured. Even when the sensor is replaced due to deterioration or failure, taking into account the number of days from when deterioration or failure was suspected to when it was replaced, it is highly likely that the sensor was operating normally for a period going back sufficiently from the date of replacement. This goes without saying when the reason for replacing the sensor is periodic replacement. Therefore, by appropriately setting the specified period to be excluded from the reference period in this way, data acquired by a sensor in a normal state can be rationally extracted as a basic data group, and high diagnostic accuracy of the sensor can be ensured.

[0111] (3) By extracting basic data groups and target data groups from multiple measurement items, it is possible to ensure a sufficient amount of data and perform an appropriate diagnosis even in an automatic analyzer that performs a small number of measurements. In addition, the data acquired by the sensor 11 during measurement may vary in value or the degree of variation may differ depending on the measurement item. In this regard, too, by extracting basic data groups and target data groups from multiple measurement items and running statistics on them, it is possible to standardize the effect of data variation due to measurement items.

[0112] (4) By extracting the basic data group by filtering based on the success or failure of QC or calibration, the validity that the basic data group is data acquired from the sensor 11 in a normal state is further increased compared to the case where the basic data group is extracted based only on the setting of the reference period. In this case, the reliability of the diagnosis result is further increased.

[0113] (5) The QC sample, standard sample (calibration sample), and dummy sample are all measured under the same or similar conditions as the patient sample. Therefore, the state of the sensor 11 can be diagnosed with high accuracy by making a diagnosis using data for at least one measurement type among the patient sample, the QC sample, the standard sample, and the dummy sample.

[0114] QC samples and standard samples have the advantage of having stable components, few impurities, and excellent constancy, and QC samples in particular have the advantage of being measured frequently and ensuring a large amount of data. Measurement of patient samples is also advantageous in that a large amount of data can be ensured. In the case of dummy measurements, the same type of sample and reagent are used each time, making them suitable for monitoring the state of sensor 11 over time under the same conditions. The advantages for monitoring the state of sensor 11 over time are also shared by the measurements of QC samples and standard samples with excellent constancy.

[0115] (6) By using a significant difference test to diagnose the sensor 11, even in an automatic analyzer 1 where it is difficult to collect data obtained by an abnormal sensor 11 due to operational reasons, the state of the sensor 11 can be properly diagnosed based on data that is presumed to be normal. Note that, although the present embodiment illustrates an example in which a one-tailed t-test is used as the significant difference test, other tests such as a two-tailed t-test or a chi-square test can also be used.

[0116] (7) In addition to cases where the sensor 11 is estimated to be in an abnormal state, notifying that the sensor 11 is estimated to be in a normal state helps users, etc., to understand the state of the automatic analyzer 1 and to plan maintenance. Also, notifying that it cannot be determined that the sensor 11 is in an abnormal or normal state can encourage users, etc., to take flexible measures.

[0117] (8) Depending on the measurement item, the measurement value may exhibit special behavior regardless of the state of the sensor 11, which may affect the diagnostic accuracy of the sensor 11. Therefore, by allowing the user to arbitrarily set the measurement items in the UI, it is possible to adjust the diagnostic accuracy by taking into account the influence of the measurement items. In addition, by presetting the default measurement items, it is possible to reduce the effort required to set the measurement conditions.

[0118] (9) In addition, by allowing the user to arbitrarily set the timing and length of the period for extracting the basic data group and the target data group in the UI, the data extraction period can be adjusted based on past diagnostic results, etc., allowing the diagnostic accuracy to be optimized for each automatic analyzer. By preparing default values, the effort required for setting measurement conditions can also be reduced.

[0119] 4. Second Example 13 is a block diagram showing a data processing flow in a diagnostic system according to a second embodiment of the present invention, in which elements that are the same as or correspond to those in the first embodiment are given the same reference numerals as those in the previously mentioned drawings, and the description thereof will be omitted.

[0120] This embodiment differs from the first embodiment in that the automatic analyzer 1 is provided with a diagnostic system for the sensor 11. The processes displayed in diagnostic function section F in the figure are a series of processes related to the diagnostic function that was handled by the server 30 (FIG. 8) in the first embodiment. Data and processes related to the diagnostic function that were allocated to the memory 31 and processing device 32 in the first embodiment are allocated to, for example, the memory 21 and processing device 22 of the control device 20 in this embodiment. The diagnostic algorithm for the state of the sensor 11 is the same as in the first embodiment, and the diagnostic result is notified to the user, etc., via the UI 94.

[0121] Although not shown in the figures, the automatic analyzer 1 of this embodiment can also be configured to be able to communicate with the server 30 or other automatic analyzers via the network NW. In that case, similar to the first embodiment, past sensor data of the other automatic analyzers can be taken into account in the diagnosis of the sensor 11. Furthermore, in cases where the automatic analyzer 1 is a stand-alone type, for example, it can also be configured to diagnose the sensor 11 based on its own data without taking into account data acquired by other automatic analyzers.

[0122] In other respects, this embodiment is similar to the first embodiment, and can provide the same effects as the first embodiment. [Explanation of symbols]

[0123] 1...automatic analyzer, 11...sensor, 21...memory, 22...processing device, 30...server (diagnostic system), 31...memory, 32...processing device, 34...user interface, 94...user interface, F...diagnostic function unit (diagnostic system), ZscoreSTD, ZscoreJDG...Z score (statistical value)

Claims

1. A diagnostic system for diagnosing a sensor that is provided in an automatic analyzer that measures an analyte component contained in a sample as a measurement item, is used to measure a plurality of measurement items, and causes a reaction product between the sample and a reagent to emit light by applying a predetermined voltage to an electrode, and outputs an analog electrical signal that is at least one of a current, a voltage, and a resistance generated in the electrode by the application of the predetermined voltage, comprising: a memory that stores data of an electrical signal output by the sensor and a replacement history of the sensor; a processing device for processing the data recorded in the memory, The processing device includes: reads out from the memory data of a set reference period among data of electrical signals outputted when a sensor used in the automated analyzer in the past measured a plurality of measurement items selected from the plurality of measurement items; Statistically processing the data of the reference period output by the past sensors when measuring the selected plurality of measurement items, regardless of the measurement items, and calculating a statistical value as a single index for diagnosis; reads out from the memory data recorded during a set evaluation period among data of electrical signals output by the diagnostic subject sensor being used in the automated analyzer when measuring the selected plurality of measurement items; calculating statistics for the evaluation period data in the same manner as the reference period statistics; An abnormality in the sensor to be diagnosed is determined based on a difference between the statistical value in the reference period and the statistical value in the evaluation period. Diagnostic system.

2. The diagnostic system of claim 1, A diagnostic system in which the reference period is a period set excluding a predetermined period before the past sensor replacement.

3. The diagnostic system of claim 1, The processing device calculates a point in time a predetermined period back from the time of the previous sensor replacement based on the history of the previous sensor replacement, and identifies a set period ending at that point in time as the reference period.

4. The diagnostic system of claim 2, The processing device includes: Prior to the measurement of the patient sample, a reference sample that can be measured multiple times from the same lot is measured using the previous sensor, and data on the success or failure of the measurement of the reference sample is recorded in the memory each time the previous sensor is used to measure the reference sample; reading from the memory a history of measurements of the reference sample performed during the predetermined period; Setting the reference period for the past sensors if all of the measurements of the reference samples performed during the predetermined period are successful Diagnostic system.

5. The diagnostic system of claim 4, A diagnostic system in which the reference sample is at least one of a quality control sample, a standard sample used for calibration, and a dummy sample.

6. 2. The diagnostic system of claim 1, The processing device calculates the statistical values ​​of the data for the reference period and the statistical values ​​of the data for the evaluation period by collecting and calculating statistics on all of the data extracted for each of the selected plurality of measurement items by a set number. Diagnostic system.

7. 2. The diagnostic system of claim 1, A diagnostic system in which the statistical value of the data for the reference period and the statistical value of the data for the evaluation period are at least one of an average value, a moving average value, a median value, and a standard deviation, or a value based on at least one of the above values.

8. 2. The diagnostic system of claim 1, The processing device determines that there is an abnormality in the sensor to be diagnosed when a difference between the data in the reference period and the data in the evaluation period is greater than a first determination value.

9. The diagnostic system of claim 8, The processing device is a diagnostic system that determines whether the difference between the data of the reference period and the data of the evaluation period is greater than the first judgment value by a significant difference test based on a statistical value of the reference period and a statistical value of the evaluation period.

10. The diagnostic system of claim 9, The processing device includes: Calculating the average value and standard deviation of the data for the reference period as statistical values ​​of the data for the reference period; For each data item during the evaluation period, a Z-score is calculated by dividing the deviation from the average value by the standard deviation, and whether or not the sensor to be diagnosed has an abnormality is determined based on a comparison between the average value of the calculated Z-scores and the first determination value. Diagnostic system.

11. The diagnostic system of claim 8, The processing device determines, based on the statistical value of the reference period and the statistical value of the evaluation period, that the sensor to be diagnosed is normal when a difference between the data of the reference period and the data of the evaluation period is smaller than a second determination value set smaller than the first determination value. Diagnostic system.

12. The diagnostic system of claim 8, When a difference between the data of the evaluation period and the data of the reference period is equal to or less than the first determination value and equal to or more than a second determination value set smaller than the first determination value, the processing device determines the sensor to be diagnosed while taking into account factors of components other than the sensor to be diagnosed, based on the statistical value of the reference period and the statistical value of the evaluation period. Diagnostic system.

13. 2. The diagnostic system of claim 1, A diagnostic system including a user interface configured to enable setting of a judgment condition for the sensor to be diagnosed.

14. The diagnostic system of claim 6, a user interface configured to allow setting of a judgment condition for the diagnostic target sensor; The user interface is configured to allow setting of at least one of the reference period and the measurement item. Diagnostic system.

15. 2. The diagnostic system of claim 1, A diagnostic system in which the sensor is of a flow cell type.

16. An automatic analyzer for diagnosing the sensor, the automatic analyzer comprising: a sensor used for measuring a plurality of measurement items, with an analyte component contained in a sample being the measurement item, the sensor causing a reaction product between the sample and a reagent to emit light by applying a predetermined voltage to an electrode, and outputting an analog electrical signal which is at least one of a current, a voltage, and a resistance generated in the electrode by application of the predetermined voltage; a memory for storing data on the electrical signal output by the sensor and a replacement history of the sensor; and a processing device for processing the data recorded in the memory, The processing device includes: reading out from the memory data of a set reference period among data of electrical signals outputted by a previously used sensor when measuring a plurality of measurement items selected from the plurality of measurement items; Statistically processing the data of the reference period output by the past sensors when measuring the selected plurality of measurement items, regardless of the measurement items, and calculating a statistical value as a single index for diagnosis; Among the data of electrical signals output by the diagnostic sensor in use when measuring the selected plurality of measurement items, data recorded during a set evaluation period is read from the memory; calculating statistics for the evaluation period data in the same manner as the reference period statistics; An abnormality in the sensor to be diagnosed is determined based on a difference between the statistical value in the reference period and the statistical value in the evaluation period. Automatic analyzer.

17. A diagnostic method for diagnosing a sensor that is provided in an automatic analyzer that measures an analyte component contained in a sample as a measurement item, is used to measure a plurality of measurement items, and causes a reaction product between the sample and a reagent to emit light by applying a predetermined voltage to an electrode, and outputs an analog electrical signal that is at least one of a current, a voltage, and a resistance generated in the electrode by the application of the predetermined voltage, comprising: Among the data of electrical signals output by the past sensors used in the automatic analyzer when measuring a plurality of measurement items selected from the plurality of measurement items, data for a set reference period is statistically processed collectively regardless of the measurement item, and a statistical value is calculated as a single index for diagnosis; Calculating statistical values ​​of data recorded during a set evaluation period among data of electrical signals output by the diagnostic subject sensor being used in the automated analyzer when measuring the selected plurality of measurement items, in the same manner as the statistical values ​​during the reference period; An abnormality in the sensor to be diagnosed is determined based on a difference between the statistical value in the reference period and the statistical value in the evaluation period. Diagnostic methods.

Images