Mass spectrometry apparatus and mass spectrometry apparatus program

Abstract

The mass analysis device of the present invention includes an ion source (31), a mass separation unit (32), and a detection unit (33). The mass analysis device includes: a parameter optimization execution unit (531, 532, 533) that optimizes the values of device parameters including various parameters affecting the ionization efficiency in the ion source (31) by using the Bayesian optimization method based on the results obtained by performing measurements while changing the values of the device parameters; a display processing unit (536) that represents, by means of a single heat map-like chart or an arrangement of multiple such charts, the posterior distribution, i.e., the sensitivity model, showing the relationship between multiple parameters, all or part of which are among the device parameters, and the signal intensity estimated at an intermediate stage of the optimization of the device parameters, and displays the same on the display unit (7) while sequentially updating it; and a file creation unit (535) that enables a user to specify an arbitrary position on the displayed chart and creates a method file describing the measurement conditions used during the measurement of the sample based on the combination of the values of the respective parameters corresponding to the specified position.

Description

Technical Field

[0001] The present invention relates to a mass analysis device and a computer program for a mass analysis device. More specifically, the present invention relates to a mass analysis device having a function of adjusting device parameters to an optimal or near-optimal state based on measurement results, and a computer program for the mass analysis device. Background Art

[0002] In a liquid chromatography mass analysis device (LC-MS), in order to ionize a compound in a sample solution eluted from a chromatographic column of a liquid chromatography unit, an ion source based on an atmospheric pressure ionization (API) method such as an electrospray ionization (ESI) method, an atmospheric pressure chemical ionization (APCI) method, or an atmospheric pressure photoionization (APPI) method is used. For example, in an ESI ion source, by applying a high voltage of the kV level near the tip of a capillary tube to which an eluent from the chromatographic column is supplied, while imparting a charge to the eluent, the eluent is sprayed into an ionization chamber in a substantially atmospheric pressure atmosphere. The tiny charged droplets thus generated are exposed to a high-temperature gas in the ionization chamber, promoting the vaporization of the solvent (mobile phase) in the droplets. During the vaporization of the solvent and the splitting of the droplets, the sample components in the droplets are ionized and taken out into the gas phase. The ions derived from the sample components thus generated are collected for mass analysis.

[0003] In order to perform highly sensitive component analysis in an LC-MS equipped with the above-described ion source, it is crucial to optimize device parameters such as the voltage applied to the constituent elements of the ion source, the temperature of these constituent elements themselves or in the ionization chamber, or the flow rate of various gases used during ionization, in order to improve the ionization efficiency as much as possible. The optimal values of such device parameters depend on the type of target component (compound), the conditions of the mobile phase (type of mobile phase, flow rate, etc.). Thus, the optimization of device parameters is generally performed based on the results obtained by actually measuring a sample containing the target component (generally a standard sample) while varying the values of each parameter within a specified range. If the number of measurements for optimizing these device parameters is large, not only does the measurement efficiency decrease, but also the consumption of samples, mobile phases, and other consumables increases, resulting in an increase in measurement costs. Therefore, it is desirable to explore the optimal parameter values, i.e., the optimal measurement conditions, with as few measurements as possible.

[0004] When the number of types of parameters to be adjusted is set to N and the number of values for which one type of parameter is varied is set to L, if it is desired to perform measurements for all combinations of parameter values for all parameters, the number of measurements becomes L N (hereinafter, this method will be referred to as the "exhaustive method"). For example, even if L and N are each around several, the number of measurements will become quite large.

[0005] In contrast, a method of optimizing parameter values one by one for each of a plurality of parameters has been conventionally known (hereinafter, this method will be referred to as the "sequential method"). This sequential method is a method adopted in the control software "Labsolutions Connect MRM" provided by Shimadzu Corporation, which is described in Non-Patent Document 1 and includes an interface parameter optimization function. The number of measurements in this sequential method is L×N, which is extremely small compared to the above-mentioned exhaustive method.

[0006] In the above-mentioned exhaustive method, among the measurement results obtained by combining all parameter values, the best result, that is, the measurement condition that usually maximizes the signal intensity of the target compound, is the optimal value. In contrast, in a method such as the sequential method where measurements are performed only for specific combinations of parameter values, sometimes even at the stage where the optimization of all parameters is completed, it is not necessarily possible to achieve parameters with higher detection sensitivity overall. That is, sometimes the result of the optimization is not the global optimal solution but falls into a local optimal solution. To avoid this situation, the following function is incorporated in the software described in Non-Patent Document 1: when changing the value of one parameter and performing a measurement at the same time, the change in the signal intensity obtained from each measurement is displayed in a graph. Thus, the user can verify whether appropriate optimization is being performed by confirming such a graph. In addition, after confirming the graph, the user can select appropriate measurement conditions and create a method file that records information showing the measurement conditions.

[0007] Prior Art Documents

[0008] Patent Documents

[0009] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2018-73360

[0010] Patent Document 2: US Patent No. 8,039,795 Specification

[0011] Patent Document 3: US Patent No. 6,759,650 Specification

[0012] Patent Document 4: US Patent No. 7,098,452 Specification

[0013] Patent Document 5: WO 2018 / 078693 Pamphlet

[0014] Non-Patent Documents

[0015] Non-Patent Document 1: "LabSolutions Connect MRM", [Online], Shimadzu Corporation, [Retrieved on July 17, 2019], URL <URL:https: / / www.an.shimadzu.co.jp / data-net / labsolutions / connect_mrm / index.htm>

[0016] Non-Patent Document 2: 5 people including Tanikawa, "Optimization of Interface Parameters for High-Sensitivity LC-MS Measurement", published by the editorial department of Shimadzu Review on March 20, 2019, Shimadzu Review, Vol. 75, No. 3 / 4, pp. 131-135

[0017] Non-Patent Document 3: 3 people including K. Swersky, "Multi-Task Bayesian Optimization", [Online], [Retrieved on July 18, 2019], NIPS, 2013, URL <URL:https: / / papers.nips.cc / paper / 5086-multi-task-bayesian-optimization.pdf>

[0018] Non-Patent Document 4: 3 people including J. Snoek, "Practical Bayesian Optimization of Machine Learning Algorithms", [Online], [Retrieved on July 18, 2019], NIPS, 2012, URL <URL:https: / / papers.nips.cc / paper / 4522-practical-bayesian-optimization-of-machine-learning-algorithms.pdf> Summary of the Invention

[0019] Technical Problem to be Solved by the Invention

[0020] The present applicant proposes a method using the Multi-Task Bayesian Optimization method as a method for more efficiently optimizing device parameters than the above-described sequential method (see Non-Patent Document 2). The details of the Multi-Task Bayesian Optimization method will be described later. However, in the Multi-Task Bayesian Optimization method, one of its features is that instead of sequentially optimizing each of the multiple parameters, measurements are taken while changing the values of the multiple parameters simultaneously, and based on the measurement results, appropriate values are determined as the measurement conditions, i.e., the values of the device parameters, in the next measurement. While repeating such operations, these multiple parameters are optimized. Therefore, even in the Multi-Task Bayesian Optimization method, similar to the sequential method, if the optimization is not properly performed, it may fall into a local solution. To avoid this situation, it is important for the user to confirm the optimization process accompanied by the actual measurement results.

[0021] However, in the Multi-Task Bayesian Optimization method, generally, for each measurement, the values of multiple parameters change simultaneously. In addition, when optimizing the parameters, since it is not predetermined which parameter will change to what extent in the measurement to be performed next during the optimization process. Therefore, with the method implemented in the software described in Non-Patent Document 1, it is not possible to perform a display that enables the user to easily grasp the relationship between the changes of each of the multiple parameters and the signal intensity. In addition, if the relationship between such multiple parameters and the signal intensity cannot be confirmed, it is also difficult for the user to judge and set appropriate measurement conditions that do not fall into a local solution.

[0022] The present invention has been completed to solve such technical problems, and an object thereof is to provide a quality analysis device and a program for a quality analysis device, which enable a user to easily confirm whether the optimization is being properly performed or has been properly performed when optimizing device parameters, and to easily set appropriate measurement conditions based on the results of the optimization.

[0023] A solution for solving the above technical problems

[0024] One aspect of the quality analysis device of the present invention includes: an ion source based on an atmospheric pressure ionization method that ionizes components contained in a liquid sample; a mass separation unit that separates ions derived from the sample components according to the mass-to-charge ratio; and a detection unit that detects the separated ions. The quality analysis device includes:

[0025] a parameter optimization execution unit that optimizes the parameter values of device parameters including multiple parameters that affect the ionization efficiency in the ion source by using the Bayesian optimization method based on the results obtained by performing measurements while changing the values of the device parameters.

[0026] A display processing unit displays, via a single heat map - type chart or an arrangement of multiple such charts, a posterior distribution, i.e., a sensitivity model, which shows the relationship between multiple parameters, all or part of the device parameters, and the signal intensity, estimated at an intermediate stage of the optimization of the device parameters performed by the parameter optimization execution unit, and displays it on the display unit while sequentially updating it;

[0027] A file creation unit enables a user to specify an arbitrary position on the chart displayed by the display processing unit, and creates a method file that describes the measurement conditions used during sample measurement based on the combination of the values of the respective parameters corresponding to the specified position.

[0028] In addition, one aspect of the program for a mass spectrometry device of the present invention is for a mass spectrometry device, the mass spectrometry device comprising: an ion source based on an atmospheric pressure ionization method for ionizing components contained in a liquid sample; a mass separation unit for separating ions derived from the sample components according to the mass - to - charge ratio; and a detection unit for detecting the separated ions. The program for the mass spectrometry device is used to optimize the parameter values of the device parameters including multiple parameters that affect the ionization efficiency in the ion source. The program for the mass spectrometry device causes a computer to act as the following functional units:

[0029] A measurement control functional unit controls the operations of the ion source, the mass separation unit, and the detection unit to perform a measurement at the value of the device parameters determined by an optimization - time parameter determination functional unit described later;

[0030] An optimization - time parameter determination functional unit uses the Bayesian optimization method to optimize the parameter values based on the results obtained from the measurements performed under the control of the measurement control functional unit, and determines the value of the device parameters for the next measurement;

[0031] A display processing functional unit displays, via a single heat map - type chart or an arrangement of multiple such charts, a posterior distribution, i.e., a sensitivity model, which shows the relationship between multiple parameters, all or part of the device parameters, and the signal intensity, estimated at an intermediate stage of the optimization of the device parameters performed by the measurement control functional unit and the optimization - time parameter determination functional unit, and displays it on the display unit while sequentially updating it;

[0032] A file creation functional unit enables a user to specify an arbitrary position on the chart displayed on the display unit, and creates a method file that describes the measurement conditions used during sample measurement based on the combination of the values of the respective parameters corresponding to the specified position.

[0033] Of course, in one aspect of the mass spectrometry device of the present invention and one aspect of the program for the mass spectrometry device, the above - mentioned Bayesian optimization method also includes the multi - task Bayesian optimization method disclosed in Non - Patent Documents 2 to 4, etc.

[0034] In addition, in one aspect of the mass spectrometry apparatus of the present invention and in the mass spectrometry apparatus according to one aspect of the program for a mass spectrometry apparatus, an ion source based on atmospheric pressure ionization method for ionizing components contained in a liquid sample typically refers to an ion source that can be set to an ion source based on the ESI method, the APCI method, or the APPI method.

[0035] Advantages of the Invention

[0036] In the above-described aspect of the mass spectrometry apparatus of the present invention and in the mass spectrometry apparatus according to the above-described aspect of the program for a mass spectrometry apparatus of the present invention, in the process of optimizing apparatus parameters using the Bayesian optimization method, each time the sensitivity model as the posterior distribution is updated, the heat map-like chart of the sensitivity model or its arrangement depicted on the screen of the display unit is sequentially updated. In the parameter optimization using the Bayesian optimization method, the change is relatively large with respect to the initial sensitivity model, and as the parameters approach the optimal value, the change in the sensitivity model becomes smaller. Therefore, the user can visually recognize the chart or its arrangement and grasp the situation that the apparatus parameters are approaching the optimal value. In addition, it is possible to confirm whether the optimization process is being properly performed based on the chart of the sensitivity model or its arrangement.

[0037] Then, at the stage when the optimization process is completed, if the user designates, for example, a position presumed to have the maximum signal intensity (highest sensitivity) on the chart of the sensitivity model at that time, the file creation functional unit creates a method file based on the combination of the values of the respective parameters corresponding to the designated position. Thus, for example, the user can simply set appropriate measurement conditions that do not fall into a local solution after making a judgment. That is, it is possible to appropriately set the measurement conditions by a simple operation performed by the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a schematic module configuration diagram of an LC-MS as an embodiment of the present invention.

[0039] Figure 2 is a schematic configuration diagram of an ion source in the LC-MS of the present embodiment.

[0040] Figure 3 is a flowchart showing the process of parameter exploration by the multi-task Bayesian optimization method in the LC-MS of the present embodiment.

[0041] Figure 4 is a diagram showing an example of problem setting and its solution in parameter exploration.

[0042] Figure 5 is a diagram showing an example of parameter exploration by the Bayesian optimization method.

[0043] Figure 6 This is an example of a diagram showing the sensitivity model as the posterior distribution when performing parameter exploration by the multi-task Bayesian optimization method in the LC-MS of the present embodiment.

[0044] Figure 7 This is an example of a diagram showing a comparison between the true sensitivity model and the posterior distribution sensitivity model when performing parameter exploration by the multi-task Bayesian optimization method in the LC-MS of the present embodiment.

[0045] Figure 8 This is an example of a diagram showing a comparison between the true sensitivity model and the posterior distribution sensitivity model when performing parameter exploration by the multi-task Bayesian optimization method in the LC-MS of the present embodiment. Detailed implementation manners

[0046] With reference to the accompanying drawings, an embodiment of the LC-MS including the mass spectrometry device of the present invention will be described.

[0047] [Overall configuration of the LC-MS of the present embodiment]

[0048] Figure 1 This is a module configuration diagram of the main part of the LC-MS of the present embodiment.

[0049] The LC-MS of the present embodiment generally includes a measurement unit 1, a control unit 4, a data processing unit 5, an input unit 6, and a display unit 7. The measurement unit 1 includes a liquid chromatography unit (LC unit) 2 and a mass spectrometry unit (MS unit) 3. The liquid chromatography unit 2 includes a liquid delivery pump 21, a syringe 22, a chromatographic column 23, etc., and the mass spectrometry unit 3 includes an ion source 31, a mass separation unit 32, a detection unit 33, etc.

[0050] In the liquid chromatography unit 2, a sample is injected from the syringe 22 into the mobile phase delivered by the liquid delivery pump 21, and the sample is sent into the chromatographic column 23 along with the liquid flow of the mobile phase. Various components (compounds) in the sample are separated over time during passage through the chromatographic column 23, and are eluted from the outlet of the chromatographic column 23 and introduced into the mass spectrometry unit 3. In the mass spectrometry unit 3, the ion source 31 ionizes the components in the eluate from the chromatographic column 23, and the mass separation unit 32 separates the generated various ions according to their mass-to-charge ratio m / z. The detection unit 33 detects the ions separated according to the mass-to-charge ratio and generates a detection signal corresponding to the ion amount.

[0051] The control unit 4 controls the operations of the measurement unit 1 and the data processing unit 5, and includes functional modules such as a measurement control unit 41 during parameter optimization, a normal measurement control unit 42, and a measurement method storage unit 43. The data processing unit 5 receives the data obtained by the measurement unit 1 and performs various data processes, and includes functional modules such as a peak intensity calculation unit 51, a data storage unit 52, and a parameter optimization processing unit 53. The parameter optimization processing unit 53 includes a reference model storage unit 531, a posterior distribution model estimation unit 532, a next exploration parameter determination unit 533, an exploration end determination unit 534, a method creation unit 535, and an optimization-time display processing unit 536 as its subordinate functional modules.

[0052] In addition, generally, most of the functional modules of the control unit 4 and the data processing unit 5 can use a personal computer as a hardware resource, and are specifically implemented by executing a dedicated control and processing program installed on the computer on the computer. One embodiment of the program for the quality analysis device of the present invention is a part of the control and processing program. Of course, such a computer program can be incorporated into non-temporary recording media such as CD-ROMs, DVD-ROMs, memory cards, and USB memories (Dongles) and provided to users. Alternatively, it can also be provided to users in the form of data transmission via communication lines such as the Internet.

[0053] [Configuration and Schematic Operation of the Ion Source in LC-MS of the Present Embodiment]

[0054] Figure 2 is a schematic configuration diagram of the ion source 31 in the LC-MS of the present embodiment. The ion source 31 is an ESI ion source, which is one of the atmospheric pressure ion sources, and includes an ESI probe 312 that ionizes the components in the eluent in an ionization chamber 311 in a substantially atmospheric pressure atmosphere formed inside the chamber 310.

[0055] The ESI probe 312 includes: a capillary 3121 through which the eluent flows; an atomizing gas tube 3122 configured to surround the capillary 3121; a heating gas tube 3123 configured to surround the atomizing gas tube 3122; an interface heater 3124 that mainly heats the gas in the heating gas tube 2123; and a high-voltage power supply 3125 that applies a high voltage to the capillary 3121. The ionization chamber 311 and the lower-stage intermediate vacuum chamber (not shown) are connected through a thin desolvation tube 313. In order to heat the desolvation tube 313, a desolvation tube heater 315 is provided around the desolvation tube 313. In addition, a drying gas tube 314 that ejects drying gas into the ionization chamber 311 is arranged around the inlet portion (ion introduction port) of the desolvation tube 313. A block heater 316 mainly heats the gas in the drying gas tube 314.

[0056] A brief description of the ion generation operation in the ion source 31 having the above configuration is given below.

[0057] When the eluate eluted from the outlet of the chromatographic column 23 reaches the vicinity of the front end of the capillary 3121, a direct current electric field formed by a high voltage (up to about several kV) applied from the high voltage power supply 3125 to the capillary 3121 imparts a charge to the eluate to bias it. The eluate with the charge becomes fine droplets (charged droplets) by means of the nebulizing gas ejected from the nebulizing gas tube 3122 and is sprayed into the ionization chamber 311. The sprayed droplets come into contact with the gas molecules in the ionization chamber 311, split, and are made finer. Since the high-temperature heating gas ejected from the heating gas tube 3123 flows in a manner surrounding the spray flow from the above-mentioned eluate, the vaporization of the solvent from the droplets is promoted, and the diffusion of the spray flow is suppressed. In the process of promoting the vaporization of the solvent from the droplets, the component molecules in the droplets have a charge and fly out of the droplets to become gas ions.

[0058] Since there is a pressure difference at both open ends of the desolvation tube 313, an air flow is formed from the ionization chamber 311 into the desolvation tube 313 by this pressure difference. As described above, the ions derived from the sample components generated in the ionization chamber 311 and the fine charged droplets in which the solvent has not been completely vaporized are sucked into the desolvation tube 313 along with the above-mentioned air flow. Since the high-temperature drying gas is ejected from the drying gas tube 314 in the opposite direction to the suction direction, the vaporization of the solvent from the charged droplets is further promoted by being exposed to this drying gas. Furthermore, since the desolvation tube 313 itself is heated to a high temperature by the desolvation tube heater 315, the vaporization of the solvent from the charged droplets is also promoted in the desolvation tube 313. Thereby, ionization is further promoted, and a large number of ions derived from the sample components are sent to the downstream intermediate vacuum chamber for mass analysis.

[0059] The following seven parameters exist in the ion source 31 as the main device parameters that affect the ionization efficiency.

[0060] (1) Interface temperature (hereinafter, sometimes simply referred to as "IFT" or "I / F temperature")

[0061] It is the temperature of the interface heater 3124 that heats the gas in the heating gas tube 3123 and the vicinity of the front end of the ESI probe 312.

[0062] (2) Block heater temperature (hereinafter, sometimes simply referred to as "BHT" or "BH temperature")

[0063] It is mainly the temperature of the block heater 316 that heats the gas in the drying gas tube 314.

[0064] (3) Desolvation Tube Temperature (hereinafter, sometimes simply referred to as "DLT" or "DL Temperature")

[0065] It is mainly the temperature of the desolvation tube heater 315 that heats the desolvation tube 313.

[0066] (4) Interface Voltage (hereinafter, sometimes simply referred to as "IFV" or "I / F Voltage")

[0067] It is the voltage used to form an electric field for charging the eluent sprayed as droplets. In this embodiment, it is the high voltage applied to the tip of the ESI probe 312 (where the polarity depends on the ionization mode and can be either positive or negative).

[0068] (5) Nebulizing Gas Flow Rate (hereinafter, sometimes simply referred to as "NebGas")

[0069] It is the flow rate of the nebulizing gas that assists the ejection of the eluent flowing near the ejection port at the tip of the ESI probe 312 through the nebulizing gas tube 3122.

[0070] (6) Heating Gas Flow Rate (hereinafter, sometimes simply referred to as "HeatGas")

[0071] It is the flow rate of the high-temperature gas that flows from the periphery of the capillary 3121 in the same direction as the spray flow of the droplets through the heating gas tube 3123.

[0072] (7) Flow Rate of Dry Gas (hereinafter, sometimes simply referred to as "DryGas")

[0073] It is the flow rate of the dry gas that flows in the direction opposite to the suction direction of the gas into the desolvation tube 313 through the dry gas tube 314.

[0074] If the values of the above 7 parameters are changed, the ionization efficiency changes, the amount of ions for mass analysis changes, and the detection sensitivity (signal intensity) in the detection unit 33 also changes. Since the degree and direction of the change in detection sensitivity depend on the component (compound), in order to perform high-sensitivity measurement, it is necessary to optimize the parameter values for each compound.

[0075] In the LC-MS of this embodiment, in order to optimize the above device parameters that affect the ionization efficiency, the multi-task Bayesian optimization method obtained by improving the general Bayesian Optimization method is used. Here, first, regarding the Bayesian optimization method, use Figure 4 、 Figure 5 to briefly explain.

[0076] [Overview of Parameter Optimization Using Bayesian Optimization Method]

[0077] When searching for the optimal value of a parameter, it is desirable to search for the optimal value with as few measurements as possible. If this problem is generalized, it belongs to the problem of "searching for the value of a parameter that can obtain the best possible measurement value in repeated measurements". The good measurement value mentioned here generally refers to the maximum signal strength value, but depending on the purpose of the measurement, it may also be the measurement value with the maximum SN ratio or the measurement value with the minimum SN ratio.

[0078] Now as an example, Figure 4 As shown in (a), the measurement of four different parameter values ​​about a certain parameter (such as voltage) is completed, and four measurement values ​​shown as black dots are obtained in the figure. Now consider the situation of exploring the next parameter value, which is expected to obtain a measurement value larger than these four measurement values.

[0079] When the user (operator) selects the parameter value to be set next according to his / her own judgment, Figure 4 As shown by the dotted line and the dashed line in (b), it is inevitable that the selection of the next parameter value will vary depending on what measurement target model is assumed (A and B in the figure).

[0080] In contrast, if the Bayesian optimization method (see Patent Document 1, etc.), which is widely known as a method of parameter search, is used, the next parameter value (measurement condition) that is likely to give a good result can be determined based on the acquired measurement data. Figure 4 (c) is a diagram showing the result of estimating the model of the measurement object using the Bayesian optimization method based on the above four measurement values, and then exploring the parameter values ​​to be measured next. Figure 4 In (c), the solid line curve shows the mean value of the posterior distribution of the model function inferred by the Bayesian optimization method, the range filled with slashes shows the uncertainty (or variance) of the posterior distribution of the model function, and the vertical bold solid line marked as "Next" shows the next parameter value automatically selected.

[0081] More specifically, in the Bayesian optimization method, under the assumption that the model of the object to be measured (hereinafter, the "model" refers to the distribution substantially showing the relationship between the parameter value and the signal intensity value or sensitivity) follows a Gaussian process, based on the acquired measurement data, the mean value and the variance value of the posterior distribution of the model function are calculated. Then, based on these calculated values, the next measurement condition (parameter value) that is expected to improve the model in a way closer to the true value is determined. Then, under the determined measurement condition, the next measurement is carried out, new measurement data is acquired, and this data is added to the already acquired measurement data, and then the posterior distribution of the model function is estimated to improve the accuracy of the posterior distribution of the model function. By repeating such processing, a parameter value that can obtain a better measurement value can be obtained.

[0082] Figure 5 FIG. is an example showing parameter exploration using the Bayesian optimization method. Figure 5 (a) of is a diagram of a state in which one more measurement value is added to the initial four measurement values (the same as that in Figure 4 (a) of ). Figure 5 (b) of is a diagram of a state in which seven more measurement values are added to the initial four measurement values. Similar to Figure 4 (c) of, the black dots show the measurement values, the solid line curve shows the mean value of the estimated posterior distribution of the model function, the dashed line curve shows the true model, the range filled with oblique lines shows the uncertainty of the posterior distribution of the model function, and the vertical bold solid line marked as "Next" shows the automatically selected next parameter value. As shown in Figure 5 (b) of, if the measurement is repeated while changing the parameter value, the parameter exploration is preferentially carried out in the range near the parameter value with the highest measurement value. Thus, a parameter value with the highest measurement value or close to it can be found.

[0083] The above description is an example in the case of optimizing only one type of device parameter. Needless to say, in the Bayesian optimization method, it can be extended to optimize multiple parameters in parallel.

[0084] In the above-described general Bayesian optimization method, sometimes the estimation accuracy of the posterior distribution of the model function at the initial stage of exploration with a small number of measurement points is low, making it difficult to conduct exploration efficiently. To overcome this situation, in recent years, a multi-task Bayesian optimization method obtained by improving the Bayesian optimization method has been proposed (Non-Patent Documents 2 to 4). In the multi-task Bayesian optimization method, on the premise that data has been obtained in an experiment to be measured (hereinafter referred to as the "target experiment") and an experiment associated therewith (hereinafter referred to as the "reference experiment"), the model of the target experiment is estimated under the assumption that there is a correlation between the data generated in the target experiment and the data generated in the reference experiment. As shown in Non-Patent Document 2, in the multi-task Bayesian optimization method, since the model of the target experiment can be estimated with very high accuracy even at the initial stage of exploration with a small number of observation points, the exploration efficiency, especially at the initial stage of exploration, can be improved.

[0085] For the above reasons, in the LC-MS of the present embodiment, the multi-task Bayesian optimization method is used for parameter optimization, but the ordinary Bayesian optimization method can also be used.

[0086] [Processing Steps of Parameter Optimization in LC-MS of the Present Embodiment]

[0087] Figure 3 is a flowchart showing the process of processing when parameter optimization is performed by the multi-task Bayesian optimization method in the LC-MS of the present embodiment. According to Figure 3 , the operation during parameter optimization in the LC-MS of the present embodiment will be described.

[0088] In addition, the algorithms of multi-task Bayesian optimization have been explained in detail in Non-Patent Documents 3, 4, etc. Moreover, since the algorithm itself is not the gist of the present invention, detailed description thereof is omitted.

[0089] To perform multi-task Bayesian optimization, a reference model based on measurement values obtained by this apparatus or an apparatus similar to this apparatus is required. For example, the reference model can be created in advance based on the measured data obtained by performing an exhaustive measurement using another apparatus of the same model as this apparatus. The "model" in this case is a sensitivity model indicating the distribution of measurement values in a 7-dimensional space in which the above 7 parameters are each set to 1 dimension, that is, the signal intensity values obtained by LC-MS. The data constituting the reference model is stored in the reference model storage unit 531 in advance.

[0090] The above sensitivity model is represented as a signal intensity distribution in a seven-dimensional space corresponding to seven device parameters. Although arithmetic processing based on such a sensitivity model can be performed, it is difficult to display the sensitivity model in a manner that is easily understandable to the user. Therefore, the data processing in the parameter optimization processing unit 53 itself is performed while reflecting all seven device parameters, but the display of the sensitivity model on the display unit 7 reduces the number of dimensions to four. According to the studies by the present inventors, the parameters that particularly affect the ionization efficiency are four parameters: the interface voltage (IFV), the nebulizing gas flow rate (NebGas), the heating gas flow rate (HeatGas), and the drying gas flow rate (DryGas). Therefore, in the display, Figure 6 a chart obtained by arranging heat map-type charts two-dimensionally (hereinafter, referred to as a "heat map two-dimensional arrangement chart") as shown represents the sensitivity model with respect to these four parameters.

[0091] In Figure 6 , the horizontal axis of a substantially square heat map-type chart is the interface voltage (IFV), the vertical axis is the nebulizing gas flow rate (NebGas), and the signal intensity value (or sensitivity) is represented according to a prescribed color scale. In addition, the axis for the horizontal arrangement of a plurality of heat map-type charts is the drying gas flow rate (DryGas), and the axis for the vertical arrangement is the heating gas flow rate (HeatGas). In this way, by arranging two-dimensional charts two-dimensionally, information on the signal intensity distribution in a four-dimensional space is represented as a whole. Therefore, a point (strictly speaking, one pixel) in one of the heat map-type charts in the heat map two-dimensional arrangement chart shows the signal intensity value corresponding to one combination of the values of the above four parameters.

[0092] First, the next exploration parameter determination unit 533 determines the values of the seven device parameters, which are the initial measurement conditions, based on the reference model stored in the reference model storage unit 531. When optimizing the parameters, the measurement control unit 41 controls the operation of the measurement unit 1 so that measurement (LC / MS analysis) is performed after setting the determined measurement conditions in the measurement unit 1. Generally, since the device parameters are optimized for each compound, a standard sample containing the target compound is used as the sample at this time, and measurement targeting this compound is performed (step S1). In the data processing unit 5, the peak intensity calculation unit 51, for example, obtains the height (or area) of a peak on the extracted ion chromatogram at a specific mass-to-charge ratio derived from the target compound as the signal intensity value.

[0093] The posterior distribution model estimation unit 532 estimates the posterior distribution (posterior distribution model) of the model function by calculating the mean and variance of the model function using Gaussian process regression based on the data constituting the reference model and the measured data (step S2). The next exploration parameter determination unit 533 calculates the measurement conditions to be measured next, that is, the values of seven parameters, by performing an operation using an acquisition function based on the estimated posterior distribution model (step S3). The measurement control unit 41 during parameter optimization controls the operation of the measurement unit 1 to perform measurement (LC / MS analysis) after setting the determined new measurement conditions in the measurement unit 1, thereby acquiring measurement data (step S4).

[0094] The exploration end determination unit 534 determines whether a pre-determined repetition end condition is satisfied (step S5). The repetition end condition can be set, for example, when the number of repetitions of a routine reaches a pre-set value, or when the measured value (signal intensity value) reaches a pre-determined target, etc. In addition, in the LC-MS of the present embodiment, as described later, when the user gives an instruction to abort the optimization, it is also determined that the end condition is satisfied. Then, in step S5, if it is determined that the repetition end condition has not been satisfied ("No" in step S5), the process returns to step S2, and the estimation of the posterior distribution model based on the newly obtained measured value and the parameter value at this time is performed again. In this way, the processing of steps S2 to S5 is repeated, and if the repetition end condition is satisfied, the exploration of the parameters is ended (step S6).

[0095] After that, the method creation unit 535 extracts a combination of parameter values based on the exploration result according to the user's instruction as described later, and creates a measurement method in which the combination of the parameter values is set as the measurement condition. Then, it is stored as a measurement method corresponding to the target compound in the measurement method storage unit 43 (step S7). In this way, since a measurement method for a certain compound is created, in the case of having a plurality of compounds to be measured, it is only necessary to perform parameter optimization and create a measurement method for each of these compounds in the same manner.

[0096] [Display processing during parameter optimization in the LC-MS of the present embodiment]

[0097] As described above, in the LC-MS of the present embodiment, while repeatedly performing measurement execution, estimation of the posterior distribution model based on the measurement result, and determination of parameters for the next measurement, the device parameters are optimized. During this processing, the display processing unit 536 during optimization performs characteristic display processing.

[0098] That is, as described above, in the execution of the optimization of the device parameters, every time a measurement under a new measurement condition is carried out, the posterior distribution model is updated. Generally speaking, this posterior distribution model changes in such a way that it gradually approaches the sensitivity model that is the correct solution as the measurements are repeated (of course, the sensitivity model that is the correct solution is unknown). If the difference from the sensitivity model that is the correct solution becomes smaller, the change in the posterior distribution model for each measurement becomes smaller. That is, the change in the posterior distribution model becomes more saturated as it gets closer to the sensitivity model that is the correct solution. This means that if the posterior distribution model approaches the sensitivity model that is the correct solution to a certain extent, the effect of additionally carrying out measurements weakens. On the contrary, the disadvantage that the time required for optimization becomes longer becomes obvious.

[0099] Therefore, in the LC-MS of the present embodiment, during optimization, the display processing unit 536 represents the posterior distribution model by a Figure 6 heat map type two-dimensional arrangement chart as shown, and depicts it on the screen of the display unit 7. Every time the posterior distribution model is updated according to the execution of the measurement, the heat map type two-dimensional arrangement chart is updated to the latest.

[0100] Figure 7 And Figure 8 is a diagram showing an example of the comparison between the change of the posterior distribution model as the number of measurements increases and the sensitivity model that is the correct solution. In addition, the observed points (measurement conditions) and the next observation point (that is, the point obtained as a result of exploration) are also shown in the posterior distribution model. In addition, the correct solution sensitivity model shown on the Figure 7 left side of Figure 8 is only shown as a reference for comparison. As described above, the correct solution sensitivity model is actually unknown. In addition, on the Figure 7 right side of Figure 8 is shown a heat map type two-dimensional arrangement chart of the posterior distribution model. Here, in order to avoid making the drawings complicated, only about half of the two-dimensional arrangement in a roughly rectangular shape is shown. In fact, it is better to display the whole as Figure 6 shown.

[0101] According to Figure 7 and Figure 8 it can be seen that as the number of observation points increases, a posterior distribution model closer to the sensitivity model that is the correct solution can be deduced. In addition, it can be seen that according to the positions of the additional observation points in the process of increasing the number of observation points, it is possible to preferentially explore the measurement conditions with higher signal intensity.

[0102] If for the 10 observation points shown in Figure 7 shown, Figure 8Comparing the charts of the posterior distribution models for the 13 observation points and the 16 observation points shown, it can be seen that the change in the posterior distribution model is relatively large when the number of observation points increases from 10 to 13. In contrast, the change in the posterior distribution model becomes smaller when the number of observation points increases from 13 to 16. If the user observes the change in such a posterior distribution model through a heatmap-style two-dimensional arrangement chart, it can be inferred that a correct estimation is made to a certain extent near the 13 observation points.

[0103] Then, the user confirms the heatmap-style two-dimensional arrangement chart updated each time the measurement is performed. When it is determined that the optimization has progressed to a satisfactory level, the user gives an optimization termination instruction by performing a specified operation on the input unit 6. In this way, when the exploration end determination unit 354 that has received this instruction performs the processing of step S5 described above next, it determines that the end condition is satisfied. That is, for example, even before the number of repetitions of the routine reaches a preset value, the exploration of the parameters can be stopped according to the user's judgment. Thus, when the optimization of the parameters has progressed to a level that the user can satisfy to a certain extent and it is difficult to expect a significant improvement in sensitivity even if more time is spent, the exploration of the parameters can be stopped at this point.

[0104] In addition, after the exploration of the parameters is completed as described above, in the state where the heatmap-style two-dimensional arrangement chart of the final posterior distribution model shown Figure 6 is displayed on the screen of the display unit 7, the user indicates, through the input unit 6, the position of the measurement condition that is presumed to be the most appropriate and generally has the maximum signal intensity value. In this way, the combination of parameter values corresponding to the position indicated by the user is specified on the heatmap-style two-dimensional arrangement chart, and a measurement method is created with the specific combination of parameter values as the measurement condition. Thus, the user can create a measurement method storing good measurement conditions for a certain compound, for example, only by performing a click operation with a pointing device.

[0105] In addition, as described above, in the parameter optimization process in the LC-MS of the present embodiment, the measurement conditions for the next measurement are automatically determined based on the measurement results. However, it is also possible for the user to set arbitrary measurement conditions ( Figure 6 any point on the chart shown) during the process, and perform the processing of steps S2 to S4 based on the results of the measurement under the measurement conditions. In this case, the measurement conditions arbitrarily set by the user are generally measurement conditions that have not been measured in the parameter optimization process, but can also be measurement conditions that have been measured one or more times in the parameter optimization process.

[0106] Setting measurement conditions by the user as described above is useful, for example, in the following case: as a certain prior information, it is known that good measurement results can be obtained under a certain measurement condition, and the measurement under this measurement condition has not been carried out in the parameter optimization process.

[0107] [Modified Example]

[0108] In the LC-MS of the above embodiment, seven device parameters that affect ionization efficiency were optimized. However, of course, not all of the seven device parameters are required. When the number of parameters to be optimized is 3 or less, the dimension of the heat map two-dimensional arrangement chart can be reduced according to the number.

[0109] In addition, even if the ion source 31 is an ESI ion source, Figure 2 The configuration shown is only an example, and its configuration can be changed in various ways. Then, as the configuration changes, the content of the parameters that affect ionization efficiency can be appropriately changed. That is, in an approximately atmospheric pressure atmosphere, the charged state of the eluate being sprayed, the size of the droplets being sprayed, the ease of vaporization of the solvent from the droplets, the collection efficiency of guiding the generated ions to the desolvation tube or the ion intake part equivalent thereto (such as a sampling cone, etc.), and other factors that affect ionization efficiency in the ESI ion source are various, and the parameters that affect any of such factors can be the object of the above parameter optimization.

[0110] For example, in the mass analyzer described in Patent Document 2, the capillary for transporting the liquid sample is grounded, and a high voltage is applied to the nozzle provided so as to surround the capillary. The droplets are charged by the electric field formed by the potential difference between the nozzle and the capillary. If the voltage applied to the nozzle is changed, the intensity of the electric field changes, and thus the charged state of the droplets also changes, and the ionization efficiency is also affected. Therefore, it is obvious that in such a configuration, the voltage applied to the nozzle instead of the capillary is equivalent to the interface voltage in the above embodiment.

[0111] In addition, in the ion source 31 in the above-described embodiment, the blowing direction of the heating gas that promotes the vaporization of the solvent from the droplets is substantially the same as the direction of the spray flow of the eluate. However, for example, in the mass spectrometer described in Patent Document 3, it is configured to blow a high-temperature gas flow to the spray flow of the eluate from two directions different from the direction of the spray flow of the eluate. In addition, in the mass spectrometer described in Patent Document 4, a gas flow is formed so as to face the spray flow of the eluate. In such a configuration, although the direction of the gas flow with respect to the spray flow of the eluate is different from that in the above-described embodiment, the purpose of improving the ionization efficiency by promoting the vaporization of the solvent from the droplets is the same. Therefore, it is obvious that these gases correspond to the heating gas and the drying gas in the above-described embodiment. In addition, the gas blown to the spray flow of the eluate in such a manner does not necessarily have to be a heated gas. Even if there is a difference in degree, it is obvious that this gas also has the effect of promoting the vaporization of the solvent from the droplets, and thus heating is not necessarily required.

[0112] In addition, in the ion source 31 in the above-described embodiment, the ions generated in the ionization chamber 311 are mainly guided to the desolvation tube 313 by the gas flow. However, a configuration can also be adopted in which the ions generated in the spray flow are induced to the inlet of the desolvation tube 313 by the action of an electric field. For example, in the mass spectrometer described in Patent Document 5, a converging electrode and a reflecting electrode are arranged in front of the direction in which the eluate is sprayed from the capillary. By the action of the electric field formed by using these electrodes, the generated ions are converged and induced to the vicinity of the inlet of the desolvation tube. As a result, the utilization efficiency (ion collection efficiency) of the generated ions is improved, and substantially the same effect as improving the ionization efficiency is achieved. Therefore, the voltage applied to such a converging electrode and reflecting electrode can also be an object of parameter optimization.

[0113] In addition, when the ion source is an ion source based on an ionization method other than the ESI method, of course, the parameters unique to the configuration of the ion source can be parameters that affect the ionization efficiency. For example, in an APCI ion source, the high voltage applied to the needle electrode is an important parameter. In addition, in an APPI ion source, the intensity (power) of the laser irradiated to the spray flow of the eluate is an important parameter.

[0114] The parameters optimized by the Bayesian optimization method or the multi-task Bayesian optimization method may include device parameters other than the parameters that affect the ionization efficiency. For example, various parameters in the mass analysis unit 3 can be set as the object of the above-mentioned parameter optimization. The various parameters in the mass analysis unit 3 are: the voltage applied to one or more ion transport optical systems for transporting the generated ions to a mass separation unit 32 such as a quadrupole mass filter; in the case of a configuration having a collision cell for dissociating ions, the collision energy when ions are dissociated in the collision cell, etc.

[0115] Furthermore, the flow rate of the mobile phase, the column oven temperature, the organic solvent ratio of the mobile phase, etc. as the LC separation conditions in the liquid chromatograph unit 2 can also be used as the object of the above-mentioned parameter optimization. When optimizing such LC separation conditions by the above method, sometimes even if the parameter values are changed, although the area of the peak on the chromatogram corresponding to the target compound changes, the height of the peak hardly changes. Therefore, when optimizing the parameters related to the LC separation conditions, it is preferable to set the signal intensity of the target compound as the peak area rather than the peak height.

[0116] In addition, the above-mentioned embodiments and modified examples are only examples of the present invention. For aspects other than these, appropriate modifications, corrections, additions, etc. within the scope of the gist of the present invention should also be included in the scope of the claims of the present invention.

[0117] [Various Solutions]

[0118] It is easy for those skilled in the art to understand that the above exemplary embodiments and their modified examples are specific examples of the following solutions.

[0119] (Item 1) The mass analysis device described in Item 1 includes: an ion source based on the atmospheric pressure ionization method for ionizing the components contained in a liquid sample; a mass separation unit for separating the ions derived from the sample components according to the mass-to-charge ratio; and a detection unit for detecting the separated ions.

[0120] The mass analysis device includes:

[0121] a parameter optimization execution unit that optimizes the parameter values of the device parameters including various parameters that affect the ionization efficiency in the ion source by using the Bayesian optimization method based on the results obtained by performing measurements while changing the values of the device parameters;

[0122] a display processing unit that represents, by a single heat map-type chart or an arrangement of multiple such charts, the posterior distribution, i.e., the sensitivity model, showing the relationship between multiple parameters of all or part of the device parameters and the signal intensity estimated in the middle stage of the optimization of the device parameters performed by the parameter optimization execution unit, and displays it on the display unit while sequentially updating it;

[0123] The file creation unit enables a user to specify an arbitrary position on a chart displayed by the display processing unit, and creates a method file that describes the measurement conditions used during sample measurement based on the combination of the values of the respective parameters corresponding to the specified position.

[0124] (Item 7) Further, the program for a mass spectrometry device according to Item 7 is for a mass spectrometry device including: an ion source based on atmospheric pressure ionization method that ionizes components contained in a liquid sample; a mass separation unit that separates ions derived from the sample components according to the mass-to-charge ratio; and a detection unit that detects the separated ions, and the program for the mass spectrometry device is for optimizing the parameter values of the device parameters including various parameters that affect the ionization efficiency in the ion source.

[0125] The program for the mass spectrometry device causes a computer to operate as the following functional units:

[0126] A measurement control functional unit that controls the operations of the ion source, the mass separation unit, and the detection unit to perform measurement with the values of the device parameters determined by the parameter determination functional unit during optimization described later.

[0127] A parameter determination functional unit during optimization that uses the Bayesian optimization method to optimize the parameter values based on the results obtained from the measurement performed under the control of the measurement control functional unit, and determines the values of the device parameters for the next measurement.

[0128] A display processing functional unit that represents, by a single heat map-type chart or an arrangement of multiple such charts, the posterior distribution, i.e., the sensitivity model, that shows the relationship between multiple parameters and signal intensity, which are all or part of the device parameters, estimated during the optimization of the device parameters performed by the measurement control functional unit and the parameter determination functional unit during optimization, and displays the same on the display unit while sequentially updating it.

[0129] A file creation functional unit enables a user to specify an arbitrary position on the chart displayed on the display unit, and creates a method file that describes the measurement conditions used during sample measurement based on the combination of the values of the respective parameters corresponding to the specified position.

[0130] According to the mass spectrometry apparatus described in Item 1 and the program for a mass spectrometry apparatus described in Item 7, in the process of optimizing the apparatus parameters using the Bayesian optimization method, each time the sensitivity model as the posterior distribution is updated, the heat map - type chart of the sensitivity model depicted on the screen of the display unit is sequentially updated. Thus, the user can visually recognize this chart to grasp the situation that the apparatus parameters are approaching the optimal values. In addition, it is possible to confirm whether the optimization process is being properly performed based on the chart of the sensitivity model. Further, according to the mass spectrometry apparatus described in Item 1 and the program for a mass spectrometry apparatus described in Item 7, for example, after the user determines appropriate measurement conditions such that they do not fall into a local solution, the user can simply set the measurement conditions and create a measurement method.

[0131] (Item 2) In the mass spectrometry apparatus described in Item 1, it can be configured as follows: It further includes a suspension instruction reception unit that enables the user to instruct the suspension of the execution of the optimization of the apparatus parameters during the execution of the optimization of the apparatus parameters by the parameter optimization execution unit.

[0132] The parameter optimization execution unit suspends the optimization of the apparatus parameters based on the suspension instruction received by the suspension instruction reception unit.

[0133] (Item 8) In the program for a mass spectrometry apparatus described in Item 7, it can be configured to further have a suspension instruction reception functional unit that enables the user to instruct the suspension of the execution of the optimization of the apparatus parameters during the execution of the optimization of the apparatus parameters by the parameter optimization execution functional unit.

[0134] The parameter optimization execution functional unit suspends the optimization of the apparatus parameters based on the suspension instruction received by the suspension instruction reception functional unit.

[0135] According to the mass spectrometry apparatus described in Item 2 and the program for a mass spectrometry apparatus described in Item 8, the user confirms the heat map - type chart of the sensitivity model on the display screen that is sequentially updated as the parameter optimization progresses, and at the time point when the user determines that the optimization has progressed to an almost satisfactory level, etc., the user can stop the parameter optimization process according to the user's instruction. Thus, it is possible to shorten the time required for parameter optimization while determining the measurement conditions that can achieve sufficient performance.

[0136] (Item 3) In the mass spectrometry apparatus described in Item 1, the parameter optimization execution unit can be configured as follows: In the middle of the optimization of the apparatus parameters, the combination of the values of the respective parameters corresponding to an arbitrary position specified by the user on the chart displayed by the display processing unit is set as the measurement conditions for the next measurement, and the optimization of the parameter values is executed.

[0137] (Item 9) In the program for the quality analysis device described in Item 7, the parameter optimization execution functional unit can be configured to, in the middle of optimizing the device parameters, set the combination of the values of the respective parameters corresponding to any position specified by the user on the chart displayed by the display processing functional unit as the measurement conditions for the next measurement, and execute the optimization of the parameter values.

[0138] According to the quality analysis device described in Item 3 and the program for the quality analysis device described in Item 9, for example, when it is known as a certain prior information that good measurement results can be obtained under certain measurement conditions, it is possible to complete the parameter optimization according to the user's instruction, with such measurement conditions being preferentially set or set without omission. Thereby, it is possible to execute the exploration of parameters that can obtain better measurement results.

[0139] (Item 4) In the quality analysis device described in Item 1, the ion source can be configured as an ion source based on the electrospray ionization method, that is, an ESI ion source.

[0140] Compared with ion sources based on other atmospheric pressure ionization methods, there are more types of parameters that affect the ionization efficiency in an ESI ion source, and its optimization is more complicated. Therefore, in a quality analysis device equipped with an ESI ion source, the effect of applying the quality analysis device of the present invention is particularly remarkable.

[0141] (Item 5) In the quality analysis device described in Item 4, the various parameters included in the device parameters can be configured as parameters related to at least any one of the voltage for forming an electric field for charging the liquid sample sprayed in a substantially atmospheric pressure atmosphere, the size of the droplets of the liquid sample being sprayed, the flow rate of the gas for promoting the vaporization of the solvent from the droplets, the temperature of the gas or the ionization chamber for ionization, or the voltage of the electric field for forming an electric field for guiding the ions from the ion generation position to the ion transport unit for downward transport.

[0142] (Item 6) More specifically, in the quality analysis device described in Item 5, the various parameters can be configured to include at least one of the flow rate of the auxiliary atomizing gas for spraying droplets from the nozzle tip, the flow rate of the heating gas for promoting the vaporization of the solvent in the sprayed droplets, the temperature of the heater for heating the heating gas, and the flow rate of the drying gas blown to the ions and droplets near the ion transport unit into which the generated ions are introduced.

[0143] According to the quality analysis devices described in Items 5 and 6, in a quality analysis device equipped with an ESI ion source, it is possible to set the parameters that have a greater impact on the ionization efficiency to an optimal or near-optimal state, and create a measurement method capable of detecting the target compound with high sensitivity.

[0144] Description of Reference Numerals

[0145] 1 Measurement Department

[0146] 2 Liquid Chromatography Department

[0147] 21 Liquid delivery pump

[0148] 22 Syringe

[0149] 23 Column

[0150] 3 Quality Analysis Department

[0151] 31. Ion source

[0152] 310 Chamber

[0153] 311 Ionization Chamber

[0154] 312 ESI probe

[0155] 3121 Capillary

[0156] 3122 Atomizing Gas Tube

[0157] 3123 Heating gas pipe

[0158] 3124 Interface Heater

[0159] 3125 High Voltage Power Supply

[0160] 313 Desolvation Tube

[0161] 314 Dry gas pipe

[0162] 315 Desolvation Tube Heater

[0163] 316 Block Heater

[0164] 32. Mass separation unit

[0165] 33. Inspection Department

[0166] 4 Control Unit

[0167] 41 Measurement control unit during parameter optimization

[0168] 42 Normal measurement control unit

[0169] 43 Measurement method storage unit

[0170] 5 Data Processing Department

[0171] 51 Peak intensity calculation unit

[0172] 52 Data storage unit

[0173] 53 Parameter Optimization Processing Unit

[0174] 531 Reference Model Storage Unit

[0175] 532 Posterior Distribution Model Estimation Unit

[0176] 533 Next Exploration Parameter Determination Unit

[0177] 534 Exploration End Judgment Unit

[0178] 535 Method Creation Unit

[0179] 536 Display Processing Unit during Optimization

[0180] 6 Input Unit

[0181] 7 Display Unit.

Claims

1. A quality analysis device, comprising: an ion source that ionizes components contained in a liquid sample based on an atmospheric pressure ionization method; a mass separation unit that separates ions derived from the sample components according to the mass-to-charge ratio; and a detection unit that detects the separated ions, characterized in that, Comprising: A parameter optimization execution unit that optimizes the parameter values of device parameters including various parameters that affect the ionization efficiency in the ion source, using the Bayesian optimization method based on the results obtained by performing measurements while changing the values of the device parameters; A display processing unit that represents, through a single heat map - type chart or an arrangement of multiple such charts, the posterior distribution, i.e., the sensitivity model, showing the relationship between multiple parameters, all or part of which are among the device parameters, and the signal intensity, estimated at an intermediate stage of the optimization of the device parameters performed by the parameter optimization execution unit, and displays it on the display unit while sequentially updating it; A file creation unit that enables a user to specify an arbitrary position on the chart displayed by the display processing unit and creates a method file describing the measurement conditions used during sample measurement based on the combination of the values of the respective parameters corresponding to the specified position; 2. The quality analysis device according to claim 1, wherein Further comprising: An abort instruction reception unit that, during the execution of the optimization of the device parameters by the parameter optimization execution unit, enables a user to instruct the abortion of the execution; The parameter optimization execution unit aborts the optimization of the device parameters according to the abort instruction received by the abort instruction reception unit; 3. The mass spectrometry device according to claim 1, wherein: During the optimization of the device parameters, the parameter optimization execution unit sets the combination of the values of the respective parameters corresponding to an arbitrary position specified by the user on the chart displayed by the display processing unit as the measurement conditions for the next measurement and executes the optimization of the parameter values; 4. The mass spectrometry device according to claim 1, wherein: The ion source is an ion source based on the electrospray ionization method; 5. The mass spectrometry device according to claim 4, wherein: The various parameters included in the device parameters are parameters associated with at least any one of the voltage for forming an electric field for charging the liquid sample sprayed in an atmospheric pressure atmosphere, the size of the droplets of the liquid sample being sprayed, the flow rate of the gas for promoting the vaporization of the solvent from the droplets, the temperature of the gas or the ionization chamber where ionization occurs, or the voltage of the electric field for forming an electric field that guides the ions from the ion generation position to the ion transport unit for downstream transport; 6. The mass spectrometry device according to claim 5, wherein: The various parameters include at least one of the flow rate of the auxiliary atomizing gas for spraying droplets from the nozzle tip, the flow rate of the heating gas for promoting the vaporization of the solvent in the sprayed droplets, the temperature of the heater for heating the heating gas, and the flow rate of the drying gas blown to the ions and droplets near the ion transport unit into which the generated ions are introduced; 7. A non - transitory computer - readable medium stores a program for a quality analysis device, and the program for the quality analysis device is used for a quality analysis device, which includes: an ion source that ionizes components included in a liquid sample based on an atmospheric pressure ionization method; A mass separation unit that separates ions derived from sample components according to the mass - to - charge ratio; And a detection unit that detects the separated ions. The program for the mass spectrometry device is used to optimize the parameter values of the device parameters including various parameters that affect the ionization efficiency in the ion source, and is characterized in that it causes a computer to act as the following functional units: The measurement control function unit controls the operations of the ion source, the mass separation unit, and the detection unit to perform measurement on the premise of the value of the device parameter determined by the function unit for determining parameters during optimization described later; The function unit for determining parameters during optimization uses the Bayesian optimization method to optimize the parameter value based on the result obtained from the measurement performed under the control of the measurement control function unit, and determines the value of the device parameter in the next measurement; The display processing function unit represents, through a single heat map-like chart or an arrangement of multiple such charts, the posterior distribution, i.e., the sensitivity model, which shows the relationship between multiple parameters, all or part of which are in the device parameters, and the signal intensity, estimated at an intermediate stage of the optimization of the device parameter performed by the measurement control function unit and the function unit for determining parameters during optimization, and displays it on the display unit while sequentially updating it; The file creation function unit enables the user to specify an arbitrary position on the chart displayed on the display unit, and creates a method file describing the measurement conditions used during the measurement of the sample based on the combination of the values of the respective parameters corresponding to the specified position; 8. The non - transitory computer - readable medium according to claim 7, wherein The computer also operates as the following function units: The abort instruction reception function unit enables the user to instruct the abort of the execution of the optimization of the device parameter during the execution of the optimization of the device parameter by the function unit for determining parameters during optimization; The function unit for determining parameters during optimization aborts the optimization of the device parameter according to the abort instruction received by the abort instruction reception function unit; 9. The non-transitory computer-readable medium according to claim 7, wherein: The measurement control function unit sets the combination of the values of the respective parameters corresponding to an arbitrary position specified by the user on the chart displayed by the display processing function unit as the measurement condition for the next measurement during the optimization of the device parameter, performs the measurement, and the function unit for determining parameters during optimization optimizes the parameter value based on the result obtained from the measurement.

Images