阅读说明：本技术 用于检查分析仪状态的技术 (Techniques for checking analyzer status ) 是由 M·卡明斯基 S·金特 B·蒂曼 于 2020-08-25 设计创作，主要内容包括：在一个总体方面中,本发明涉及一种监测自动化分析仪的液相色谱(LC)流的状态的方法。所述方法包括：自动监测所述液相色谱流的注射组件的系统压力以生成系统压力的时间序列；将所述时间序列分类为指示所述LC流的不同状态的两个或更多个预定类别中的一者；以及基于分类结果来触发响应。(In one general aspect, the present invention relates to a method of monitoring the status of a Liquid Chromatography (LC) flow of an automated analyzer. The method comprises the following steps: automatically monitoring a system pressure of an injection assembly of the liquid chromatography flow to generate a time series of system pressures; classifying the time series into one of two or more predetermined categories indicative of different states of the LC flow; and triggering a response based on the classification result.)

1. A method of monitoring the status of a Liquid Chromatography (LC) flow of an automated analyzer, the method comprising automatically performing the steps of:

monitoring a system pressure of an injection assembly (10) of the liquid chromatography stream to generate a time series (31; 40a-d) of system pressures;

classifying the time series (31; 40a-d) into one of two or more predetermined classes (54) indicative of different states of the LC flow; and

a response (55) is triggered based on the classification result.

2. The method of claim 1, wherein the monitoring occurs during a sample injection process.

3. The method of claim 2, wherein the time series spans at least a portion of an injection process of the sample into an LC column (5) of the LC flow.

4. The method according to any one of claims 1 to 3, wherein classifying the time series (31; 40a-d) comprises using a classifier trained by a machine learning algorithm, optionally wherein the classifier is trained in terms of historical stress time series or simulated stress time series.

5. The method according to any one of claims 1 to 4, wherein classifying the time series (31; 40a-d) comprises a feature analysis of one or more predetermined features of the time series.

6. The method of any preceding claim 1 to 5, wherein the two or more predetermined categories (54) comprise at least one category indicative of injection of gas into the injection assembly, optionally wherein the two or more categories (54) comprise at least two categories indicative of injection of different amounts of gas.

7. The method according to any one of the preceding claims 1 to 6, wherein the two or more categories (54) comprise at least one category indicating injections below a nominal sample volume and/or indicating injections above a normal sample volume.

8. The method according to any one of the preceding claims 1 to 7, wherein the two or more classes (54) comprise at least one class indicative of abnormal sample composition, optionally wherein the abnormal sample composition is a composition comprising at least one unexpected substance, optionally an unexpected organic substance, or a composition in which an expected substance is absent.

9. The method of any of the preceding claims 1-8, wherein the two or more categories (54) include at least one category indicative of normal operation of the injection assembly (10).

10. The method according to any of the preceding claims 1 to 9, wherein the response (55) comprises starting or scheduling an automatic maintenance operation.

11. The method according to any of the preceding claims 1 to 10, wherein the response (55) comprises requiring an operator to perform a predetermined inspection or maintenance operation, optionally comprising providing instructions regarding the respective inspection or maintenance operation.

12. The method of claim 11, wherein the inspection or maintenance operation is one or more of inspecting the syringe assembly (10) for air bubbles or inspecting a sample dilution process.

13. The method according to any of the preceding claims 1 to 12, wherein the response (55) comprises notifying a service provider.

14. A computer system configured to perform the steps of any one of the methods of claims 1-13.

15. A computer-readable medium having instructions stored thereon, which when executed by a computer system, prompt the computer system to perform the steps of any one of the methods of claims 1-13.

Technical Field

The present disclosure relates to automated methods for monitoring the status of a Liquid Chromatography (LC) flow of an automated analyzer.

Background

Automated analyzers (e.g., in vitro analyzers) are common in today's laboratory and hospital environments. These devices tend to become more and more complex due to increased functionality and increased throughput and the need to perform analysis tasks in an automated fashion. As a result, errors and malfunctions may occur in many components, which may result in a decrease in the productivity of the analyzer or a decrease in the reliability of the measurement results. In some instances, an external service person may be required to discover and repair the error, which may take hours or even days during which the analyzer or a portion of the analyzer may not be available.

Disclosure of Invention

In one general aspect, the present invention relates to a method of monitoring the status of a Liquid Chromatography (LC) flow of an automated analyzer. The method comprises the following steps: automatically monitoring a system pressure of an injection assembly of the liquid chromatography flow to generate a time series of system pressures; classifying the time series into one of two or more predetermined categories indicative of different states of the LC flow; and triggering a response based on the classification result.

In a second general aspect, the present invention is directed to a computer system configured to perform the steps of the techniques of the first general aspect.

The techniques of the first and second general aspects may have advantageous technical effects.

First, in some examples, monitoring techniques can be seamlessly integrated into existing analyzer workflows. For example, the detection and/or monitoring techniques may be performed as part of an analyzer initialization workflow. In some examples, the monitoring techniques may use monitoring data (e.g., pressure of a pump used during an injection) already typically available in the analyzer (e.g., to control pressurization of a portion of the LC stream by the pump). In these cases, additional hardware may not be required to perform the detection and/or monitoring techniques of the present disclosure.

Second, the monitoring techniques of the present disclosure may be employed to distinguish between different states of the analyzer and trigger a particular response. In this manner, the detection and/or monitoring techniques may facilitate more efficient use of resources (e.g., operator time or outside service personnel) by allowing these resources to be more accurately allocated due to improved knowledge of the status of the analyzer. In some cases, downtime of the analyzer (or its modules) may also be reduced, as improved knowledge of the state of the analyzer may be used to select the most appropriate response.

Third (and in relation to the second point), monitoring of the present disclosure may allow less experienced operators to perform service and maintenance operations that may require the participation of outside service personnel when using some known automated analyzers. By classifying the time series into one of two or more predetermined categories indicative of different states of the LC flow and triggering a response based on the classification results, an inexperienced operator may be able to identify and fix errors and other problems with the LC flow of an automated analyzer.

Several terms having specific meanings are used in this disclosure.

A "time series" according to the present disclosure refers to at least two values (e.g., system pressure) of a particular parameter at two different points in time (e.g., at least one earlier point in time and at least one later point in time). In some examples, a time series may include (much more than) more than two values at respective points in time. The term "point in time" shall not limit the measurement window for obtaining the measurement values comprised in the time series to a certain accuracy. For example, an average measurement value obtained by averaging a plurality of measurements of a parameter may also be included in the time series according to the present disclosure. The time series may comprise values for equidistant time points or non-equidistant time points. In the present disclosure, the term "time series" is used to refer to both "raw data" (e.g., retrieved from a pressure sensor) and to processed raw data (e.g., by using signal processing techniques) as long as the processing steps still reflect the system pressure of the injection assembly.

The terms "automated" or "automatically" in accordance with this disclosure refer to operations performed by a machine without user interaction. The automated steps may be part of a method that further includes steps requiring user interaction. For example, a user may schedule or trigger automated steps of the techniques of this disclosure.

An "automated analyzer" according to the present disclosure is a device dedicated to performing analytical functions. In some examples, the analyzer may be configured to perform an analysis of a sample (e.g., a sample for in vitro diagnosis). For example, the analyzer may be a clinical diagnostic system for performing in vitro diagnostics. The automated analyzer of the present disclosure includes at least one Liquid Chromatography (LC) stream.

The analyzers of the present disclosure may have different configurations as needed and/or according to a desired workflow. Additional configurations may be obtained by coupling multiple devices and/or modules together. A "module" is a working unit, typically smaller in size than the entire analyzer, and has specialized functionality. This function may be an analytical function, but may also be a pre-analytical or post-analytical function, or may be any ancillary function to a pre-analytical, analytical or post-analytical function. In particular, a module may be configured to cooperate with one or more other modules to perform a dedicated task of a sample processing workflow, for example by performing one or more pre-analysis steps and/or post-analysis steps.

In particular, the analyzer may comprise one or more analysis devices designed to perform individual workflows optimized for certain types of analysis.

The analyzer may include an analytical device for one or more of clinical chemistry, immunochemistry, coagulation, hematology, and the like.

Thus, the analyzer may comprise one analysis device or any such analysis device in combination with a respective workflow, wherein the pre-analysis module and/or the post-analysis module may be coupled to the respective analysis device or shared by a plurality of analysis devices. In the alternative, the pre-analysis function and/or the post-analysis function may be performed by a unit integrated in the analysis device. The analyzer may comprise functional units such as liquid handling units for aspirating and/or pumping and/or mixing samples and/or reagents and/or system fluids, and also functional units for sorting, storing, transporting, identifying, separating, detecting.

The term "sample" refers to a biological material suspected of containing one or more analytes of interest, whose qualitative and/or quantitative detection may be correlated with a specific condition (e.g., clinical symptom).

The sample may be from any biological source, such as physiological fluids, including blood, saliva, ocular lens fluid, cerebrospinal fluid, sweat, urine, milk, ascites, mucus, synovial fluid, peritoneal fluid, amniotic fluid, tissue, cells, and the like. The sample may be pre-treated prior to use, such as preparing plasma from blood, diluting viscous liquids, dissolving, etc.; the treatment methods may involve filtration, centrifugation, distillation, concentration, inactivation of interfering components, and addition of reagents. In some cases, samples obtained from a source can be used directly, or following pre-processing and/or sample preparation workflows to alter the characteristics of the sample, e.g., after addition of an internal standard, after dilution with another solution, or after mixing with reagents, e.g., to enable one or more in vitro diagnostic tests to be performed, or for enrichment (extraction/separation/concentration) of analytes of interest and/or for removal of matrix components that may interfere with the detection of one or more analytes of interest.

The term "sample" is often used to refer to a sample before sample preparation or a sample after sample preparation, or both.

Examples of analytes of interest are typically vitamin D, drugs of abuse, therapeutic drugs, hormones and metabolites. However, this list is not exhaustive.

In particular, the analyzer may include a sample preparation station for automatically preparing the sample. A "sample preparation station" is a pre-analysis module coupled to one or more analysis devices or units in an analysis device and designed to perform a series of sample processing steps aimed at removing or at least reducing interfering matrix components in a sample and/or enriching a sample for an analyte of interest. Such processing steps may include any one or more of the following processing operations performed on one or more samples sequentially, in parallel, or in an interleaved manner: aspirating (aspirating and/or dispensing) fluid, pumping fluid, mixing with reagents, incubating at a temperature, heating or cooling, centrifuging, separating, filtering, sieving, drying, washing, resuspending, aliquoting, transferring, storing … …)

The sample may, for example, be provided in a sample container, such as a sample tube (including primary and secondary tubes), or a multi-well plate, or any other sample carrier. The reagents may be arranged, for example, in the form of containers or cassettes containing individual reagents or groups of reagents, and placed in appropriate receptacles or locations within the storage chamber or conveyor. Other types of reagents or system fluids may be provided in bulk containers or via a pipeline supply.

The term "about" in relation to the values of a parameter is meant to encompass a deviation of +/-10% of the value specified in the disclosure, unless otherwise specified in the respective context.

Drawings

FIG. 1 is a flow diagram of a monitoring technique of the present disclosure.

Fig. 2a, 2b and 2c show the injection process in an exemplary LC flow.

Fig. 3 depicts an exemplary set of system pressure time series for multiple injection procedures.

Fig. 4 shows four different sets of system pressure time series for an injection process when one of the different error conditions of the LC flow occurs.

Fig. 5 is a flow diagram of an exemplary monitoring technique of the present disclosure.

Detailed Description

First, an overview of the monitoring technique of the present disclosure will be given in conjunction with fig. 1. In the following sections, further aspects of the monitoring technique will be discussed in connection with fig. 2a to 5.

General overview

FIG. 1 is a flow diagram of a monitoring technique of the present disclosure.

The method comprises the following steps: automatically monitoring 101 a system pressure of an injection assembly of a liquid chromatography flow to generate a time series of system pressures; classifying 103 the time series into one of two or more predetermined categories indicative of different states of the LC flow; and triggering 105 a response based on the classification result.

The monitoring and classification steps will be discussed in detail in the following sections.

In some examples, the monitoring occurs during a sample injection process. For example, the injection process may include injecting the sample into an LC column of the LC stream. The system pressure may be generated by a pump connected to the injection assembly (e.g., an elution pump of an automated analyzer for injecting the sample into the LC flow column). Further aspects of the components for monitoring system pressure and components of the injection assembly will be discussed below in connection with fig. 2 a-c.

The time series may span at least a portion of the injection process of the sample into the LC flow column (e.g., a window of 20% or more or 50% or more of the duration of the injection process, or at most 20% or at most 50% of the duration of the injection process). In some examples, the time series may span the entire injection process of the sample into the LC flow column. In all cases, the period of time during which the injection occurs may be the focus of the monitoring techniques of the present disclosure. As discussed herein, the pressure curve obtained during this time period may be of interest to derive different states of LC flow.

In some cases, the LC gradient with a particular pressure characteristic may vary with the sample injection process. This portion of the pressure time series may be of less interest to the techniques of this disclosure. Thus, in some examples, no pressure values are sampled for the duration of the LC gradient. Additionally, in some examples, the time series does not include pressure values sampled prior to the injection procedure. However, in some examples, the time series may span at least a portion of the injection process of the sample and a time period before and/or after the injection process of the sample. For example, the time series may include a time period prior to the injection process and a portion (or the entire duration) of the injection process into the LC column of the LC flow.

The classification step will be discussed next. As part of the classification step, the time series can be automatically processed to help determine which of the two or more categories is the appropriate category for the current time series. In some examples, classifying the time series includes feature analysis of the time series (e.g., feature analysis of a plurality of predetermined features of the time series). The category determination may then comprise selecting an appropriate category based on the determined characteristics of the time series. The features may be dynamically adjusted or updated during operation of the automated analyzer.

In some examples, classifying the time series includes using a classifier trained by a machine learning algorithm. In the present context, the term "trained" means that a classifier is performed on a plurality of training samples within a training period during which the properties of the classifier are adapted to produce or improve a classification result. The training may be performed prior to delivery of the automated analyzer or its control software and/or after deployment of the automated analyzer or its control software (e.g., at the customer). As previously discussed, the classifier may be updated periodically or continuously.

In some examples, the classifier may be trained on historical pressure time series or simulated pressure time series. In some examples, the classifier may be updated based on the automated analyzer and/or a pressure time series of different automated analyzers (e.g., data of different automated analyzers may be collected and processed to train the classifier, and the trained classifier may be provided to the automated analyzer).

In some examples, the classifier may be trained in terms of metadata generated by processing a time series (e.g., feature analysis as described in this disclosure). In this example, the step of classifying may include a time series of initial processing steps to generate the metadata, and may include subsequent steps of applying the classifier. However, in other examples, the classifier may be trained and used by directly processing the time series. Other aspects of processing the time series will be discussed below in conjunction with fig. 3 and 4.

In some examples, the classifier may be model-based or model-free.

In some examples, the classifier may employ an artificial neural network. As discussed above, the artificial neural network may be trained on sample data.

In other examples, the classifier may employ other algorithms trained through machine learning techniques. For example, digital pattern recognition techniques may be employed in the classification step of the monitoring techniques of the present disclosure. In other examples, classification may include using a decision tree algorithm, a probabilistic classifier (e.g., a classifier using a naive bayes-type algorithm), a support vector machine, a relevance classifier, a nearest neighbor classifier, or other suitable numerical classification technique trained through machine learning techniques.

In other examples, other numerical classification techniques that are not trained by machine learning techniques may be used. For example, some algorithms may be configured with explicit instructions to classify a time series into one of two or more predetermined categories. For example, rule-based classification techniques may be configured with explicit instructions to classify a time sequence. However, in some cases, machine learning techniques may be superior to techniques that require explicit directives for some classification techniques.

In this manner, the monitoring techniques of the present disclosure may be used to automatically retrieve information about different states of an analyzer. The system pressure profile of the injection assembly contains information that can be automatically retrieved using the techniques of this disclosure to improve the operation of the automated analyzer (e.g., reduce downtime).

Automatic analyzer

Additional aspects of the components of the automated analyzer and injection assembly will be discussed in the following sections in conjunction with fig. 2a, 2b and 2 c. Fig. 2a, 2b and 2c show these components and the injection process of the LC flow.

Fig. 2a shows an injection assembly 10 and an LC column 5 for LC flow connected to injection assembly 10. Furthermore, a mass spectrometer 6 is connected to the LC column 5. The injection assembly 10 is configured to inject a sample into the LC column 5. In the LC column 5, the sample is separated (by time) into its components. The separated components of the sample are introduced into a mass spectrometer to resolve the mass-to-charge ratios of the components (or fragments thereof).

The injection assembly comprises: an injection port 3 for inputting a sample to the injection assembly 10; and a transfer system 4, 7 for transferring the sample from the injection port to a port of the LC column 5 of the LC flow column. The syringe assembly 10 is connected to the pump 1 and the waste container 8. The transfer systems 4, 7 are configured to switch between different states connecting two or more of the pump 1, the waste container 8, the injection port and the port of the LC column 5 in different configurations. In the example of fig. 2a, the transfer system 4, 7 comprises a rotary valve 4 and a sample ring 7. In addition, the injection system 10 includes a sample dispensing component 2 (e.g., an automated syringe).

The sample injection process can be performed as follows:

in a first step, as shown in figure 2a, a sample may be drawn from a vial or other sample container into the sample distribution section 2. During this operation, the injection port 3 is connected to the waste container 8. In a further step depicted in fig. 2b, the injection valve 3 is switched to the loading position, in which the sample ring 7 connects the injection port 3 and the waste container 8. The sample dispensing component 2 (e.g., an automated syringe) transfers the sample through the injection port 3 into the sample ring 7. In some examples, the amount of sample transferred may be a few microliters (e.g., less than 10 microliters).

In a further step, shown in fig. 2c, the injection valve 3 switches the sample ring 7 to the injection position. In the injection position, the pump 1 is connected to the sample loop 7 at one end of the sample loop 7. The other end of the sample loop 7 is connected to the LC column 5. In this position, the pump pressurizes the fluid path including the sample loop 7 to inject the sample into the LC column 5 (where it may be further processed as discussed above). The system pressure of injection assembly 10 during the injection procedure (during the entire injection procedure or a portion thereof) may be monitored to generate a time series for further processing in the techniques of this disclosure. Fig. 3 shows an exemplary set of time series that may be obtained.

For example, the system pressure may be monitored by a pressure sensor disposed in the injection assembly 10 or a component attached to the injection assembly. For example, a pressure sensor of the pump 1 may be used to monitor the system pressure. The pressure sensor may already be provided, for example, to detect a malfunction of the pump 1 or for monitoring the pressurizing operation performed by the pump 1. In other examples, pressure sensors may be disposed elsewhere in injection assembly 10, or in a component of the LC flow connected to injection assembly 10, to monitor the system pressure of injection assembly 10.

In some examples, system pressure may be monitored directly (via a pressure sensor) or indirectly through the use of any sensor capable of monitoring a parameter directly related to system pressure.

It should be noted that the layouts of fig. 2a, 2b and 2c are merely exemplary. The monitoring techniques of the present disclosure may also be applied if one or more LC streams of the automated analyzer are set differently. For example, the injection assembly 10 may include different components than a rotary valve.

Furthermore, the LC column 5 of the (alternatively or additionally) LC flow may be coupled to other detectors than the mass spectrometer 6. For example, the detector may be an optical or magnetic detector configured to analyze the sample provided by the LC column 5.

Additionally or alternatively, the fluid path connecting the injection assembly 10 and the LC column 5 may be arranged differently. For example, the fluid path may include additional components (e.g., additional valves or capillaries). In some examples, injection assembly 10 may be coupled to multiple LC columns of multiple LC flows (e.g., by providing a flow selector valve). In addition, the automated analyzer may include multiple injection assemblies (e.g., for multiple LC flows) as discussed above. In such a case, the monitoring techniques of the present disclosure may be employed to automatically monitor the system pressure (in parallel or at different times) for multiple or each injection assembly.

The monitoring techniques of the present disclosure may be applied regardless of the layout of one or more LC streams of an automated analyzer.

Details of processing time series

In subsequent sections, different aspects of the time series of system pressures that address the monitoring techniques of the present disclosure will be discussed in greater detail in connection with FIG. 3. Fig. 3 depicts an exemplary set 31 of system pressure time series for multiple injection procedures.

It can be seen that the pressure drops at the beginning of the injection process and then rises again (e.g., over a period of less than 3 minutes or less than 2 minutes). In general, the shape of the pressure curve can be explained as follows: when the sample and additional liquid and additional air gap reside in the sample ring (e.g., of the injection assembly as discussed above), they are not pressurized. At the moment of switching the injection valve, the pressurized system is open to the ring (unpressurized) and equilibrium occurs. This may result in an initial pressure drop. The contents of the sample loop are pressurized to the system pressure over a subsequent period of time. When the contents of the sample loop are pressurized to the system pressure, the pressure is obtained prior to the switching event of the injection valve.

In some examples, the time series of system pressures may be pre-processed. For example, a single time series of system pressures may be smoothed or outliers may be eliminated. Additionally or alternatively, multiple time series (e.g., each spanning at least a portion of a single injection procedure) may be averaged. The averaged time series may then be further processed. In other examples, one or more portions of the time series may be eliminated.

In some examples, one or more features (e.g., a plurality of features) are extracted from the time series. These features may relate to one or more of the following: a pressure value at a predetermined location (e.g., a pressure value at a maximum or minimum in a time series), a magnitude of a pressure change (e.g., a pressure rise or a pressure fall), a speed of the pressure change.

Additionally or alternatively, the features may relate to global characteristics of the time series. For example, the features may quantify the number of maxima and minima or the number of oscillations of the time series. Additionally or alternatively, the features may be time series of spectral features. For example, the features may relate to a magnitude at a particular frequency (i.e., a magnitude of a fourier coefficient), or to spectral energy at a particular frequency. In other examples, the features may relate to global or local extrema (e.g., global maximum, local maximum, global minimum, or local minimum) of the time series. Other features may combine different ones of the above features in the meta-feature.

The extracted features may be organized in any suitable form (e.g., as feature vectors) and input into a classifier. As discussed above, the classifier may be any type of classifier suitable for classifying extracted features (e.g., feature vectors) into two or more classes indicative of different states of the LC flow. In one example, the classifier is a classifier trained by a machine learning algorithm.

In other examples, the time series may be used directly as an input to the classifier. In these examples, no previous feature extraction steps occurred. However, there may be signal processing steps to bring the time series into a form suitable for inputting them into the classifier (e.g., one or more of the pre-processing steps discussed above). For example, the time series may be used directly as an input to an artificial neural network or another classifier trained through machine learning.

Category and response

In subsequent sections, different possible categories and exemplary responses will be discussed in more detail in connection with fig. 4 and 5.

Fig. 4 shows four different sets of a system pressure time series 40 of the injection process when one of the different error states of the LC flow occurs. In one example, each of the error states may belong to a different class of states of the LC flow.

In some examples, the two or more predetermined categories include at least one category indicative of injection of gas into the injection assembly. For example, the two or more categories may include at least two categories (or three or more categories or four or more categories) indicative of different amounts of gas to inject. In the environment surrounding the automated analyzer, the gas may be air or another gas.

The curve on the right side of fig. 4 is an exemplary curve monitored when injecting a lower air volume (e.g., 1 μ L-upper right curve 40b) and a higher air volume (e.g., 5 μ L-lower right curve 40 d). It can be seen that the curve has a generally more pronounced drop at the beginning of the injection process (i.e. the minimum pressure at the beginning of the injection process is at a lower pressure value at the time of injection of the gas) than the curve measured during normal operation of the specification of the injection assembly as shown in fig. 3. This may occur due to the higher compressibility of air or other gases compared to the sample. In this way, pressurizing the volume of the injection assembly comprising the sample and, for example, a certain amount of air results in a greater compression of said volume and therefore a lower pressure than in the absence of gas.

Additionally or alternatively, the two or more predetermined categories include at least one category indicative of injection below a nominal amount of the sample injected during the injection (e.g., two or more indications indicating two different levels of below nominal sample amounts, etc.). For example, a category may indicate that the sample amount is below a threshold percentage of the nominal sample amount (e.g., less than 90% of the nominal amount or less than 50% of the nominal amount). In other examples, the first category may indicate that the sample amount is within a first range below the nominal sample amount, and the second category may indicate that the sample amount is within a second range below the nominal sample amount that is below the first range. The collection of curves 40c on the lower left of FIG. 4 are monitoring curves measured when the sample amount is below normal (e.g., 1 μ L instead of a nominal amount of 5 μ L). The difference between the curves in this state and the normal state is not clearly visible. However, the automatic classifier of the present disclosure may distinguish between the two categories based on time series.

Additionally or alternatively, the two or more predetermined categories include at least one category indicative of higher than normal amount of injection of the sample injected during the injection (e.g., two or more indications indicative of two different levels of higher than normal sample amount, etc.).

Additionally or alternatively, the two or more predetermined categories include at least one category indicative of abnormal sample composition. For example, an abnormal sample composition is a composition that includes at least one undesired species (e.g., an undesired organic species). In other examples, the abnormal sample composition is a composition in which the desired substance is absent. The collection of curves 40a on the upper left of fig. 4 are monitoring curves measured during injection of water instead of a mixture of water and organic constituents. It can be seen that in this case the time series may have an oscillating characteristic with several minima and maxima.

Additionally or alternatively, the two or more reservations include at least one category indicative of defective injection valves. This may mean that the time series is a (substantially) flat curve (not shown in fig. 4). Other categories may indicate other defects in the components of the injection assembly.

Additionally or alternatively, the two or more predeterminations include at least one classification that indicates the presence of particles (not shown in fig. 4) in the injected sample.

In some examples, the two or more categories include at least one category indicative of normal operation of the injection assembly and at least one category indicative of an error condition of the injection assembly. In one example, the classifier has only two classes: one category indicative of normal behavior and a second category indicative of abnormal behavior (e.g., combining two or more or all of the abnormal behaviors discussed above). In this case, the classifier performs binary classification. In other examples, the classifier has only three or more classes: one category indicative of normal behavior and two or more categories indicative of different abnormal behavior (e.g., including two or more or all of the abnormal behavior discussed above). In this case, the classifier performs multi-label classification.

Classification based on a time series of pressure values can occur with high accuracy. However, it should be appreciated that the classification may not work optimally (e.g., not every time the LC flow is in a certain state-e.g., a certain amount of air is injected-the classifier will obtain a corresponding classification result). In some examples, if multiple injection procedures are classified into a particular category, only a particular state is assumed and/or only a particular response is triggered.

In all cases, the monitoring techniques of the present disclosure may include automatically triggering a response. The following section will discuss exemplary responses that may be triggered by classification. In case a specific class indicative of the LC flow state is detected, the different exemplary responses listed below may be combined into a response.

In some examples, the response includes recording the classification result. For example, if the sorting of the time series results in normal operation of the injection assembly, a response to the recorded sorting results may be triggered. In other examples, a response to the recorded classification results occurs for each classification operation.

Additionally or alternatively, the response may include initiating or scheduling an automated maintenance operation. For example, the maintenance operation may be one or more of checking for bubbles in the syringe assembly or checking for sample dilution. If the classification of the time series yields an abnormal sample composition, a response to the initiation or scheduling of an automated maintenance operation may be triggered.

Additionally or alternatively, the response may include generating an error message. This response may be triggered whenever the sorting produces a result that the injection assembly is not within specification.

Additionally or alternatively, the response may include requiring an operator to perform a predetermined inspection or maintenance operation. This may include outputting a message on an interface (e.g., a graphical user interface) of the operator. The message may include information about the scheduled inspection or maintenance operation (e.g., instructions on how to perform the scheduled inspection or maintenance operation.

For example, the maintenance operation is one or more of checking for bubbles in the syringe assembly or checking for sample dilution. If the sorting of the time series produces an indication that gas is to be injected into the jetting assembly, a response may be triggered that requires an operator to perform a predetermined inspection or maintenance operation.

Additionally or alternatively, the response may include notifying the service provider.

In some examples, the response may include a flag that one or more measurements performed on one or more samples using LC flow may be erroneous. Additionally or alternatively, the response may be to stop operation of the automated analyzer or a module of the automated analyzer (e.g., a particular LC flow or LC flow column).

If the classification result is that the LC flow is in a particular state, two or more of the responses discussed above may be triggered. For example, recording the classification result and generating an error message may be triggered for a particular abnormal state of the LC flow.

Exemplary monitoring techniques

Fig. 5 is a flow chart 50 of an exemplary monitoring technique of the present disclosure. In one step 52, a sample injection process in an injection assembly of an LC flow is initiated. The injection process is monitored 53 for a system pressure time series (or multiple time series spanning multiple injection processes). The obtained time series may be preprocessed for classification. In another step, the time series is sorted. In the example of fig. 5, the classifier can distinguish between five different classes and trigger a specific response for each class.

The first category (category 1) indicates normal operation of the injection assembly. In other words, the injection assembly operates according to specifications. The triggered response includes generating an entry in a log (e.g., in a log of control software of an automated analyzer). The entries in the log may include information that the classification results have resulted in normal operation of the injection assembly. Additionally or alternatively, the entries in the log may include information related to a time series of system pressures (e.g., a time series of system pressures).

The second category (category 2) indicates samples that have been injected with the wrong organic content. This is a category that indicates abnormal sample composition as discussed above. A trigger response to this type of detection may trigger a check procedure for air bubbles by the injector of the injection assembly. This may involve sending a message (e.g., including instructions on how to perform the inspection process) to an operator or service personnel of the automated analyzer. In other examples, the inspection process may be performed automatically (at least in part) by an automated analyzer.

The third category (category 3) indicates partial air injection. As discussed above, such categories may include situations where a particular portion of the nominal sample volume (rather than the entire sample volume) is occupied by air. Determining that the LC flow is in this state may trigger a response that includes sending a message to the operator to check IF the minimum sample size is provided and following instructions regarding this ("check IF ready condition").

The fourth category (category 4) indicates partial sample injection. This is an example of a category indicating a sub-normal amount of injected sample as discussed above. A trigger response to this class of detection may trigger an examination process of the sample dilution process. This may involve sending a message (e.g., including instructions on how to perform the inspection process) to an operator or service personnel of the automated analyzer. In other examples, the inspection process may be performed automatically (at least in part) by an automated analyzer.

The fifth category (category 5) indicates full air injection. Determining that the LC flow is in this state may trigger a response that includes automatically scheduling a maintenance visit (e.g., visit by an outside service person). In some examples, the automated scheduling process may include sending a message to a service person.

The set of five different categories depicted in fig. 5 is merely exemplary. As discussed above, when using the techniques of this disclosure, a variety of different categories indicative of different states of an LC flow may be distinguished.

Computer implementation

The present disclosure also relates to a computer system configured to perform a technique of monitoring a state of a Liquid Chromatography (LC) flow of an automated analyzer.

In some examples, the computer system may be a controller of the analyzer (or a portion thereof). However, in other examples, the computer system may only be connected to the analyzer via a network and may not be part of the controller of the analyzer. For example, the computer system may be a hospital or laboratory management system, or a computer system of a supplier or service provider of the analyzer.

The computer system is only required to obtain a time series of system pressures for the injection assembly of the liquid chromatography flow. This may mean that the computing system receives this information over the network. However, in other examples, as discussed above, the computing system also controls the functionality of the analyzer (e.g., measuring pressure or triggering a response), which means that it is the controller of the analyzer.

The computing system of the present disclosure is not limited to a particular software or hardware configuration. A computing system may have a software or hardware configuration as long as the software or hardware configuration is capable of performing the steps of the techniques for monitoring the state of a Liquid Chromatography (LC) flow of an automated analyzer according to the present disclosure.

The present disclosure also relates to a computer-readable medium having instructions stored thereon that, when executed by a computer system, prompt the computer system to perform steps of a technique for monitoring the state of a Liquid Chromatography (LC) flow of an automated analyzer according to the present disclosure.

A computer program comprising computer executable instructions for performing the method according to the present disclosure in one or more embodiments attached herein when the program is executed on a computer or a computer network is further disclosed and presented. In particular, the computer program may be stored on a computer readable data carrier. Thus, in particular, one, more than one or even all of the method steps disclosed herein may be performed by using a computer or a network of computers, preferably by using a computer program.

A computer program product is further disclosed and proposed, having program code to perform the method according to the present disclosure in one or more embodiments enclosed herein, when said program is executed on a computer or a computer network. In particular, the program code may be stored on a computer readable data carrier.

Further disclosed and proposed is a data carrier having stored thereon a data structure which, after loading into a computer or computer network, such as into a working memory or a main memory of a computer or computer network, can execute a method according to one or more embodiments disclosed herein.

A computer program product is further disclosed and proposed, having a program code stored on a machine-readable carrier for performing a method according to one or more embodiments disclosed herein, when said program is executed on a computer or a computer network. As used herein, a computer program product refers to the program as a tradable product. The products may generally be present in any format, such as paper format, or on a computer readable data carrier. In particular, the computer program product may be distributed over a data network.

Further disclosed and proposed is a modulated data signal containing instructions readable by a computer system or computer network for performing a method according to one or more embodiments disclosed herein.

With reference to the computer-implemented aspects of the invention, one or more, or even all, of the method steps according to one or more embodiments disclosed herein may be performed by using a computer or a network of computers. Thus, in general, any method steps involving the provision and/or manipulation of data may be performed by using a computer or a network of computers. In general, these method steps may include any method steps, typically other than those requiring manual work, such as providing a sample and/or taking certain aspects of a measurement.

Further disclosed and proposed is a computer or computer network comprising at least one processor, wherein said processor is adapted to perform a method according to one embodiment described in the present specification.

A computer-loadable data structure adapted to perform a method according to an embodiment described in the present specification while the data structure is being executed on a computer is further disclosed and claimed.

Further disclosed and proposed is a storage medium on which a data structure is stored, and wherein the data structure is adapted to perform a method according to one embodiment described in the present description after having been loaded into a main storage and/or a working storage of a computer or a computer network.

Other aspects

Various aspects of techniques for monitoring the state of Liquid Chromatography (LC) flow of an automated analyzer of the present disclosure have been discussed in the preceding sections. In addition, techniques for monitoring the state of Liquid Chromatography (LC) flow of an automated analyzer of the present disclosure may also be performed according to the following aspects:

1. a method of monitoring the status of a Liquid Chromatography (LC) flow of an automated analyzer, the method comprising automatically performing the steps of:

monitoring a system pressure of an injection assembly of the liquid chromatography flow to generate a time series of system pressures;

classifying the time series into one of two or more predetermined categories indicative of different states of the LC flow; and

a response is triggered based on the classification result.

2. The method of aspect 1, wherein the monitoring occurs during a sample injection process.

3. The method of aspects 1 or 2, wherein the system pressure is generated by a pump connected to the syringe assembly.

4. The method according to any one of aspects 2 or 3, wherein the time series spans at least a portion of an injection process of the sample into an LC column of the LC stream.

5. The method of any of aspects 1-4, wherein classifying the time series comprises using a classifier trained by a machine learning algorithm.

6. The method of aspect 5, wherein the classifier is trained on historical pressure time series or simulated pressure time series.

7. The method of any of aspects 1-6, wherein classifying the time series comprises feature analysis of one or more predetermined features of the time series.

8. The method of aspect 7, wherein the one or more features include one or more of: a pressure value at a predetermined location, a magnitude of a pressure change, a velocity of a pressure change, a maximum and a minimum of a time series, a number of oscillations of the time series, or a spectral feature of the time series.

9. The method of any of preceding aspects 1-8, wherein the two or more predetermined categories include at least one category indicative of injection of gas into the injection assembly.

10. The method of aspect 9, wherein the two or more categories include at least two categories indicative of injections of different amounts of gas.

11. The method according to any one of the preceding aspects 1 to 10, wherein the two or more categories comprise at least one category indicating injections below a nominal sample volume and/or indicating injections above a normal sample volume.

12. The method according to any one of the preceding aspects 1 to 11, wherein the two or more classes comprise at least one class indicative of abnormal sample composition.

13. The method of aspect 12, wherein the abnormal sample composition is a composition comprising at least one undesired species, optionally an undesired organic species, or a composition in which a desired species is absent.

14. The method of any of the preceding aspects 1-13, wherein the two or more categories include at least one category indicative of normal operation of the injection assembly.

15. The method of any preceding aspect 1 to 14, wherein the response comprises recording the classification result.

16. The method of aspect 15, wherein the response to recording the classification result is triggered if the classification of the time series results in normal operation of the injection assembly.

17. The method of any of preceding aspects 1-16, wherein the response comprises initiating or scheduling an automated maintenance operation.

18. The method of aspect 17, wherein the maintenance operation is one or more of checking for bubbles in the syringe assembly or checking for sample dilution.

19. The method of aspect 17 or 18, wherein the response to initiating or scheduling an automated maintenance operation is triggered if the classification of the time series results in an abnormal sample composition.

20. The method of any preceding aspect 1 to 19, wherein the responding comprises generating an error message.

21. The method of any of preceding aspects 1 to 20, wherein the response comprises requiring an operator to perform a predetermined inspection or maintenance operation, optionally comprising providing instructions regarding the respective inspection or maintenance operation.

22. The method of aspect 21, wherein the inspection or maintenance operation is one or more of inspecting the syringe assembly for bubbles or inspecting a sample dilution process.

23. The method of aspect 21 or 22, wherein the response to a request for an operator to perform a predetermined inspection or maintenance operation is triggered if the classification of the time series produces an indication to inject gas into the jetting assembly.

24. The method of any preceding aspect 1 to 23, wherein the response comprises notifying a service provider.

25. The method of any of preceding aspects 1-24, wherein the injection assembly comprises: an injection port for inputting a sample to the injection assembly; and a transfer system for transferring a sample from the injection port to a port of the LC flow column.

26. The method of aspect 25, wherein the injection assembly is connected to a pump and a waste container, and wherein the transfer system is configured to switch between different states connecting two or more of the pump, the waste container, the injection port, and the port of the LC column in different configurations.

27. A computer system configured to perform the steps of any one of the methods of aspects 1-26.

28. The computer system of aspect 27, wherein the computer system is a controller of the automated analyzer.

29. A computer-readable medium having instructions stored thereon that, when executed by a computer system, prompt the computer system to perform the steps of any one of the methods of aspects 1-26.