Machine learning techniques for discovering errors and system ready conditions in liquid chromatograph
By applying artificial intelligence/machine learning models to mass spectrometry and liquid chromatography-mass spectrometry systems, fault conditions can be automatically diagnosed, solving the problems of time-consuming and expert-dependent judgment in existing technologies, and achieving rapid and accurate fault identification and improved maintenance efficiency.
Patent Information
- Application Number
- CN202380031965.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-02-02
- Filing Date
- 2023-02-02
- Publication Date
- 2024-11-19
AI Technical Summary
The fault diagnosis process of existing mass spectrometry and liquid chromatography-mass spectrometry systems is time-consuming and relies on expert judgment, resulting in low customer satisfaction and high maintenance costs. Novice users may not be able to identify fault conditions, and the varying levels of expert experience can lead to misdiagnosis and unnecessary parts replacement.
Employing artificial intelligence/machine learning models, based on time-series signal values and experimental results data, it identifies fault states and predicts potential problems. Combined with micro-applications and AI/ML systems, it automates diagnosis and provides user feedback.
It enables rapid and accurate fault diagnosis, reduces downtime, improves user experience and service efficiency, and lowers maintenance costs.
Smart Images

Figure CN118984935A_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 305,867, filed on February 2, 2022. The entire disclosure of that application is hereby incorporated by reference herein. Background Art
[0003] Mass spectrometry (MS) equipment and liquid chromatography-mass spectrometry (LCMS) equipment are used to analyze chemical samples to study the characteristics, quality or structure of the sample. Such systems are composed of many parts and subsystems, each of which may be associated with an error (e.g., valve leak, pump failure, etc.) or a system readiness condition (e.g., balance, startup, etc.). Such systems can provide output signals (such as pressure signals) that can be analyzed by experts to detect errors or readiness conditions. However, this is a time-consuming process that requires the presence of an expert to review the data. In addition, this is a subjective process, and different experts may detect errors / readiness conditions in the same data at different times.
[0004] For example, LC users must know a lot about the normal and abnormal states and operations of LC devices. In addition to some simple (and not widely used) indicators, users must understand and recognize when the instrument is in one of the following states:
[0005] ●Infused
[0006] Balanced
[0007] Check valve leakage
[0008] ●Pressure seal leakage
[0009] Degasser failure
[0010] ●Clogged injection valve
[0011] Partially blocked needle
[0012] ●Contaminated columns
[0013] The column is in chemical and thermal equilibrium
[0014] ●The detector is stable and does not drift
[0015] In some cases, there are existing system checks that detect these conditions, but they are either not run by the user or take too long. In other cases, it is up to the user to determine and resolve the problem, or the user is at least aware that there is a problem and calls a service engineer. This results in lower customer satisfaction because of longer downtime when parts fail, and longer experimentation time because of the skills and maintenance required from the user. In some cases, a novice user may not know that the system is in an error state and continue to collect problematic data.
[0016] Likewise, while service specialists are well trained, their experience levels on each of these issues and / or some of the instruments vary. Due to this, and the lack of easily human-understandable diagnostics, properly functioning parts are replaced while troubleshooting the root cause of the problem. This results in increased costs and waste for the company. Accurate, fast, and easy-to-understand diagnostics can greatly improve service efficiency. Summary of the invention
[0017] Exemplary embodiments relate to a computer-implemented method for building and training an artificial intelligence / machine learning (AI / ML) model that will identify fault states and readiness conditions based on time series signal values and / or experimental result data, and predict when a fault may occur in the future. Embodiments also relate to a non-transitory computer-readable medium storing instructions for performing the method, a device configured to perform the method, and the like.
[0018] The models generated by these techniques can be generalized to solve many different problems. For example, one embodiment can predict a column plugging condition. A user can queue multiple injections (e.g., 50) to run over a period of time (e.g., overnight). Exemplary embodiments can predict error conditions and thus warn the user that the instrument may only be able to perform a subset of injections (e.g., the first 20) before the error condition prevents the instrument from continuing.
[0019] For those skilled in the art, other technical features can be easily seen from the following figures, descriptions and claims. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] To easily identify the discussion of any particular element or act, the most significant digit(s) in a reference number refers to the figure numeral that first introduces that element.
[0021] Figure 1A An aspect of the subject matter according to one embodiment is shown.
[0022] Figure 1B An aspect of the subject matter according to one embodiment is shown.
[0023] Figure 2 An exemplary artificial intelligence / machine learning (AI / ML) system suitable for use with the exemplary embodiments is shown.
[0024] Figure 3 An exemplary computer system architecture is depicted that can be used to practice the exemplary embodiments described herein.
[0025] Figure 4 An example of a mass spectrometry system according to an exemplary embodiment is shown.
[0026] Figure 5 is a flow chart depicting an exemplary method for detecting system ready / error conditions according to an exemplary embodiment. DETAILED DESCRIPTION
[0027] Note about data privacy
[0028] Some embodiments described herein utilize training data or metrics that may include information voluntarily provided by one or more users. In such embodiments, data privacy may be protected in a variety of ways.
[0029] For example, users may be required to opt-in to any data collection before their data is collected or used. Users may also be provided with the opportunity to opt-out of any data collection. Before opting in to data collection, users may be provided with a description of how their data will be used, how long the data will be retained, and the security measures in place to protect the data from being disclosed.
[0030] Any information identifying the user from which data is collected may be removed or separated from the data. In the event that any identifying information needs to be retained (e.g., to meet regulatory requirements), the user may be informed of the collection of the identifying information, the purposes for which the identifying information will be used, and the amount of time the identifying information will be retained. Information that specifically identifies the user may be removed and may be replaced with, for example, a universal identification number or other non-specific form of identification.
[0031] Once collected, the data may be stored in a secure data storage location that includes security measures to prevent unauthorized access to the data. The data may be stored in an encrypted format. After a predetermined period of time, identifying information and / or non-identifying information may be cleared from the data storage device.
[0032] Although specific privacy protection techniques are described herein for purposes of illustration, one of ordinary skill in the art will recognize that privacy can also be protected in other ways. Further details about data privacy will be discussed in the section describing network implementation schemes below.
[0033] Assuming that the user's privacy conditions are met, the exemplary embodiments can be deployed in a variety of messaging systems, including messaging in social networks or on mobile devices (e.g., via a messaging client application or via a short message service), among other possibilities. An overview of exemplary logic and processes for participating in a synchronous video conversation in a messaging system is provided next.
[0034] Reference is now made to the accompanying drawings, in which like reference numerals are used to refer to like elements throughout. In the following description, for the purpose of explanation, many specific details are set forth in order to provide a thorough understanding thereof. However, novel embodiments may be practiced without these specific details. In other cases, well-known structures and devices are shown in the form of block diagrams for ease of description thereof. It is intended to cover all modifications, equivalents, and substitutes consistent with the claimed subject matter.
[0035] In the drawings and accompanying description, the names "a", "b", and "c" (and similar designations) are intended to be variables representing any positive integer. Thus, for example, if an implementation sets a value of a=5, the complete set of components 122, shown as components 122-1 through 122-a, may include components 122-1, 122-2, 122-3, 122-4, and 122-5. The embodiments are not limited in this context.
[0036] Exemplary embodiments
[0037] An exemplary implementation may generally follow the following steps.
[0038] Generate fault data
[0039] A controlled data set can be generated that exhibits normal behavior as well as abnormal, suboptimal or outright failed behavior. This may involve replacing a working part on a test liquid chromatograph (LC) with a failed part retrieved from the field and running an injection using the failed part, or retrieving data from a previous abnormal, suboptimal or outright failed user analysis.
[0040] For example, a data collection user might install a faulty part and run various types of injections to represent a wide range of chemistries, columns, ambient temperatures, and other external variables that affect the time trajectory.
[0041] Examples of data sets that suggest anomalies or system readiness conditions include, but are not limited to:
[0042] ● Equilibrium data, for which chromatographic data and pressure traces can be collected; the system can check chemical equilibrium and thermal equilibrium
[0043] ● Pressure traces indicating that the LC device has experienced a loss of perfusion conditions; pressure traces showing loss of perfusion under a variety of different circumstances can be collected
[0044] ● Check valve leakage data
[0045] ●Pressure seal leakage data
[0046] Degasser fault data
[0047] ● Data of blocked injection valve
[0048] ● Partially blocked needle data
[0049] ●Contaminated column data
[0050] Detector stability (e.g., lack of drift) data
[0051] Build and train classification and prediction models
[0052] Example embodiments may provide a machine learning model configured to perform two different functions: classification and prediction.
[0053] The classification paradigm is to record the data (e.g., the most recent time series of data from an instrument) and identify any problems in it. The model will use this signal to determine:
[0054] 1. Is there a problem (i.e., classifying the signal as normal or abnormal behavior)
[0055] 2. What is the root cause of the problem (i.e., categorize the problem into a category with actionable solutions, such as a leaking valve)
[0056] 3. Predict the time in the future when a failure may occur (e.g., the remaining useful life of a part, or the amount of time required to balance an instrument for the next injection)
[0057] Based on the model that classifies the pressure signal as anti-phase injection balanced or not, a further model can be built such that:
[0058] 1. Indicate the balance of injection
[0059] 2. Indication of loss of perfusion
[0060] 3. Indicates the re-acquisition of perfusion
[0061] 4. Indicates that there is a leak and the cause of the leak is the check valve
[0062] Exemplary embodiments may extract data from an analytical chemistry device (eg, LC device) at a rate of at least 10 Hz; in some cases, the analytical chemistry device may naturally provide diagnostic data at this rate. Using a frequency of at least 10 Hz may increase successful diagnosis of error conditions.
[0063] The application can read a 96 hour plot file where 10 Hz data is written by the analytical chemistry device firmware. In one embodiment, the following steps can be performed by a computer processor:
[0064] ●Call the instrument API and use it to read the binary 96-hour plot data
[0065] ● Embed the new code into the appropriate location in the edge device code so that the pipeline can use it
[0066] ●Call this code from your microapp to collect feedback on the imbalanced model
[0067] Figure 1A An example of this step is depicted.
[0068] Developing Balanced Micro Applications
[0069] The micro-app can be a fat client web app that will employ suitable code to read the data at (at least) 10Hz, employ the trained model to determine the equilibrium time, present the equilibrium time to the user, and obtain feedback from the user as to whether the model is correct, early, or late. The code can also extract a lot of metadata about solvent composition, flow rates, etc. in order to define the context of the model performance. The micro-app can be written in python, using a low-level code library to generate a web app, or if creating a fat client, a simple GUI library.
[0070] Figure 1B An example of this step is depicted.
[0071] For each of the situations to be addressed, such as perfusion, balance, and leakage, many test cases can be generated. The performance of the model can be compared to human subject matter expert performance. Any relevant metric for calculating the performance of the model relative to a human can be used, such as correlation, RMSE, accuracy, etc.
[0072] Example embodiments may utilize artificial intelligence / machine learning (AI / ML). Figure 2 An AI / ML environment 200 suitable for use with an exemplary embodiment is depicted.
[0073] First, note that Figure 2 A specific AI / ML environment 200 is depicted and discussed in conjunction with a specific type of AI / ML architecture. However, other AI / ML systems exist, and one of ordinary skill in the art will recognize that any suitable technology may be used to implement the AI / ML environment in addition to the depicted AI / ML environment.
[0074] The AI / ML environment 200 may include an AI / ML system 202, such as a computing device that applies an AI / ML algorithm to learn the relationships between the above-mentioned protein parameters.
[0075] The AI / ML system 202 can utilize training data 208. In some cases, the training data 208 can include pre-existing labeled data from a database, library, repository, etc. The training data 208 can include, for example, rows and / or columns of data values 214. The training data 208 can be co-located with the AI / ML system 202 (e.g., stored in a storage device 210 of the AI / ML system 202), can be remote from the AI / ML system 202 and accessed via a network interface 204, or can be a combination of local and remote data. Each unit of the training data 208 can be labeled with an assigned category 216 (or multiple assigned categories); for example, each row and / or each column can be labeled with a classification. In some embodiments, the training data can include individual data elements (e.g., not organized into rows or columns) and can be labeled on an individual basis.
[0076] As described above, the AI / ML system 202 may include a storage device 210 , which may include a hard drive, a solid-state storage device, and / or a random access memory.
[0077] The training data 212 can be used to train a model 222. Depending on the particular application, different types of models 222 may be suitable for use. For example, exemplary embodiments may utilize a Bayesian hierarchical model, or a gradient boosted tree may be particularly suitable for learning the association of data values 214 and assigned categories 216. In other examples, deep learning architectures such as recurrent neural networks (RNNs). Other types of models 222 or non-model-based systems may also be well suited for the tasks described herein, depending on the designer's goals, available resources, the amount of available input data, etc.
[0078] Any suitable training algorithm 218 may be used to train the model 222. Nevertheless, Figure 2 The example shown in may be particularly suitable for supervised training algorithms. For supervised training algorithms, AI / ML system 202 can apply data value 214 as input data, and the resulting assigned category 216 can be mapped to the input data to learn the association between the input and the label. In this case, the assigned category 216 can be used as a label for the data value 214.
[0079] The training algorithm 218 may be applied using processor circuitry 206, which may include suitable hardware processing resources to operate on the logic and structures in the storage device 210. The development of the training algorithm 218 and / or training model 222 may depend at least in part on model hyperparameters 220; in an exemplary embodiment, the model hyperparameters 220 may be automatically selected based on hyperparameter optimization logic 228, which may include any known hyperparameter optimization techniques appropriate to the selected model 222 and the training algorithm 218 to be used.
[0080] Optionally, model 222 may be retrained over time.
[0081] In some embodiments, some of the training data 212 can be used to initially train the model 222, and some of the training data can be reserved as a validation subset. The portion of the training data 212 that does not include the validation subset can be used to train the model 222, while the validation subset can be reserved and used to test the trained model 222 to verify that the model 222 can generalize its predictions to new data.
[0082] Once the model 222 is trained, it can be applied (by the processor circuit 206) to new input data. The new input data can include unlabeled data stored in a data structure, potentially organized into rows and / or columns. This input to the model 222 can be formatted according to a predefined input structure 224, which reflects the manner in which the training data 212 is provided to the model 222. The model 222 can generate an output structure 226, which can be, for example, a prediction of the assigned category 216 to be applied to the unlabeled input.
[0083] The above description relates to a specific kind of AI / ML system 202 that applies supervised learning techniques given available training data with input / result pairs. However, the present invention is not limited to use with a specific AI / ML paradigm, and other types of AI / ML techniques may be used.
[0084] Figure 3An example of a system architecture and data processing device that can be used to implement one or more illustrative aspects described herein in a standalone and / or networked environment is shown. Various network nodes such as data server 310, web server 306, computer 304, and laptop computer 302 can be interconnected via a wide area network (WAN) 308 such as the Internet. Other networks can also or alternatively be used, including a private intranet, a company network, a LAN, a metropolitan area network (MAN), a wireless network, a personal network (PAN), etc. Network 308 is for illustrative purposes and can be replaced with fewer or additional computer networks. A local area network (LAN) can have one or more topologies in any known LAN topology, and can use one or more protocols in a variety of different protocols, such as Ethernet. Device data server 310, web server 306, computer 304, laptop computer 302 and other devices (not shown) can be connected to one or more networks in the network via twisted pair, coaxial cable, optical fiber, radio waves or other communication media.
[0085] The computer software, hardware, and networks may be used in a variety of different system environments, including stand-alone, networked, remote access (also known as remote desktop), virtualized, and / or cloud-based environments, among others.
[0086] The term "network" as used herein and depicted in the accompanying drawings refers not only to a system in which remote storage devices are coupled together via one or more communication paths, but also to independent devices that may be coupled to such a system from time to time with storage capabilities. Thus, the term "network" includes not only a "physical network" but also a "content network" consisting of data residing on all physical networks that can be attributed to a single entity.
[0087] Components may include a data server 310, a web server 306, a client computer 304, and a laptop computer 302. The data server 310 provides overall access, control, and management of databases and control software for performing one or more illustrative aspects described herein. The data server 310 may be connected to the web server 306, through which a user interacts with the requested data and obtains the data. Alternatively, the data server 310 itself may act as a web server and be directly connected to the Internet. The data server 310 may be connected to the web server 306 via a network 308 (e.g., the Internet), via a direct or indirect connection, or via some other network. A user may interact with the data server 310 using a remote computer 304, a laptop computer 302, such as using a web browser to connect to the data server 310 through one or more external public web sites hosted by the web server 306. The client computer 304, the laptop computer 302 may be used together with the data server 310 to access data stored therein, or may be used for other purposes. For example, from a client computer 304, a user may access the web server 306 using an Internet browser (as is known in the art) or by executing a software application that communicates with the web server 306 and / or data server 310 via a computer network (such as the Internet).
[0088] The server and application can be combined on the same physical machine and retain separate virtual or logical addresses, or can reside on separate physical machines. Figure 3 Only one example of a network architecture that may be used is shown, and those skilled in the art will appreciate that the specific network architecture and data processing devices used may vary and are secondary to the functionality they provide, as further described herein. For example, the services provided by web server 306 and data server 310 may be combined on a single server.
[0089] Each component data server 310, web server 306, computer 304, laptop computer 302 can be any type of known computer, server or data processing device. Data server 310 can include, for example, a processor 312 that controls the overall operation of data server 310. Data server 310 can also include RAM 316, ROM 318, network interface 314, input / output interface 320 (e.g., keyboard, mouse, display, printer, etc.) and memory 322. Input / output interface 320 can include various interface units and drivers for reading, writing, displaying and / or printing data or files. Memory 322 can also store operating system software 324 for controlling the overall operation of data server 310, control logic 326 for instructing data server 310 to perform various aspects described herein, and other application software 328 that provides auxiliary, support and / or other functions that can be used in conjunction with or without the various aspects described herein. Control logic may also be referred to as data server software control logic 326 in this article. The functionality of the data server software may refer to a combination of operations or decisions made automatically based on rules encoded into the control logic, manually by a user providing input to the system, and / or automatic processing based on user input (e.g., queries, data updates, etc.).
[0090] The memory 1122 may also store data for performing one or more aspects described herein, including a first database 332 and a second database 330. In some embodiments, the first database may include the second database (e.g., as a separate table, report, etc.). That is, the information may be stored in a single database, or divided into different logical, virtual, or physical databases, depending on the system design. The web server 306, computer 304, laptop computer 302 may have a similar or different architecture as described with respect to the data server 310. Those skilled in the art will appreciate that the functionality of the data server 310 (or web server 306, computer 304, laptop computer 302) as described herein may be distributed across multiple data processing devices, for example, to distribute processing loads across multiple computers, to separate transactions based on geographic location, user access level, quality of service (QoS), etc.
[0091] One or more aspects may be embodied in computer-usable or readable data and / or computer-executable instructions, such as embodied in one or more program modules executed by one or more computers or other devices as described herein. Typically, program modules include routines, programs, objects, components, data structures, etc., which perform specific tasks or implement specific abstract data types when executed by a processor in a computer or other device. These modules can be written in source code programming languages, which are then compiled for execution, or these modules can be written in scripting languages such as (but not limited to) HTML or XML. Computer executable instructions can be stored on a computer-readable medium such as a non-volatile storage device. Any suitable computer-readable storage medium can be used, including hard disks, CD-ROMs, optical storage devices, magnetic storage devices, and / or any combination thereof. In addition, various transmission (non-storage) media representing data or events as described herein can be transmitted between a source and a destination in the form of electromagnetic waves propagated through a signal-conducting medium (such as a metal wire, an optical fiber, and / or a wireless transmission medium). The various aspects described herein can be embodied as a method, a data processing system, or a computer program product. Thus, the various functions may be embodied in whole or in part in software, firmware, and / or hardware or hardware equivalents such as integrated circuits, field programmable gate arrays (FPGAs), etc. Certain data structures may be used to more efficiently implement one or more aspects described herein, and such data structures are contemplated to be within the scope of the computer-executable instructions and computer-usable data described herein.
[0092] For illustration purposes, Figure 4 is a schematic diagram of a system that can be used in conjunction with the techniques herein. Figure 4 A specific type of device in a specific LCMS configuration is depicted, but one of ordinary skill in the art will appreciate that different types of chromatography devices (eg, MS, tandem MS, etc.) may also be used in conjunction with the present disclosure.
[0093] Sample 402 is injected into liquid chromatograph 404 via injector 406. Pump 408 pumps the sample through column 410 to separate the mixture into component parts according to retention time through the column.
[0094] The output from the column is input into a mass spectrometer 412 for analysis. Initially, the sample is desolvated and ionized by a desolvation / ionization device 114. Desolvation can be any desolvation technique, including, for example, a heater, a gas, a heater in combination with a gas, or other desolvation techniques. Ionization can be achieved using any ionization technique, including, for example, electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI), matrix-assisted laser desorption (MALDI) or other ionization techniques. The ions produced by the ionization are sent into a collision chamber 418 by the voltage gradient being applied to an ion guide 416. The collision chamber 418 can be used to transfer ions (low energy) or to fragment ions (high energy).
[0095] Various techniques can be used, including those described in U.S. Pat. No. 6,717,130 to Bateman et al., which is incorporated herein by reference, in which an alternating voltage can be applied across the collision cell 418 to induce fragmentation. Spectra are collected of the precursor (no collision) at low energy and the fragments (products of collision) at high energy.
[0096] The output of the collision cell 418 is input to a mass analyzer 420. The mass analyzer 420 can be any mass analyzer, including a quadrupole, a time of flight (TOF), an ion trap, a magnetic sector mass analyzer, and a combination thereof. A detector 422 detects ions emitted from the mass analyzer 122. The detector 422 can be integrated with the mass analyzer 420. For example, in the case of a TOF mass analyzer, the detector 422 can be a microchannel plate detector that counts ion intensities (i.e., counts incident ions).
[0097] The raw data repository 424 can provide permanent storage for storing ion counts for analysis. For example, the raw data repository 424 can be an internal or external computer data storage device, such as a disk, a flash-based storage device, etc. The acquisition device 426 analyzes the stored data. The data can also be analyzed in real time without being stored in the storage medium 124. In real-time analysis, the detector 422 passes the data to be analyzed directly to the computer 126, rather than first storing it in a permanent storage device.
[0098] Collision cell 418 performs the fragmentation of precursor ions. Fragmentation can be used to determine the primary sequence of peptides, and then the origin protein is identified. Collision cell 418 contains gas, such as helium, argon, nitrogen, air or methane. When charged precursor interacts with gas atoms, the collision produced can fragment the precursor by decomposing the precursor into the resulting fragment ions. This fragmentation can use the technology described in Bateman, by switching the voltage in the collision cell between a low voltage state (e.g., low energy, <5V) and a high voltage state (e.g., high energy or elevated energy, >15V), wherein the low voltage state is used to obtain the MS spectrum of the peptide precursor, and the high voltage state is used to obtain the MS spectrum of the collision-induced fragment of the precursor. High voltage and low voltage can be referred to as high energy and low energy, because high voltage or low voltage is used respectively to give kinetic energy to ions.
[0099] Various procedures can be used to determine when and how to switch the voltage for this MS / MS collection. For example, conventional methods trigger voltage with a target or data-dependent mode (data-dependent analysis, i.e., DDA). These methods also include gas phase isolation (or preselection) of the coupling of the target precursor. The low-energy spectrum is acquired and checked by software in real time. When the required mass reaches the specified intensity value in the low-energy spectrum, the voltage in the collision cell is switched to a high-energy state. The high-energy spectrum for the preselected precursor ion is then obtained. These spectra include fragments of the precursor peptides seen under low energy. After collecting enough high-energy spectra, data acquisition returns to the low-energy state, and continues to look for the precursor mass with suitable intensity to be used for high-energy collision analysis.
[0100] Different suitable methods can be used together with the system as described herein to obtain ion information, such as information on precursor and product ions of mass spectrometry for analyzing samples. Although conventional switching techniques can be adopted, embodiments can use the technology described in Bateman, which can be characterized as a fragmentation procedure that switches voltage with a simple alternating cycle. This switching is completed at a sufficiently high frequency so that multiple high-energy spectra and multiple low-energy spectra are included in a single chromatographic peak. Different from conventional switching procedures, the cycle is independent of the content of the data. This switching technique described in Bateman provides effective simultaneous mass analysis of both precursor ions and product ions. In Bateman, the use of high-energy and low-energy switching procedures can be applied as a part of LC / MS analysis of a single injection of a peptide mixture. In the data collected from a single injection or experimental run, the low-energy spectrum contains ions mainly from unfragmented precursors, and the high-energy spectrum contains ions mainly from fragmented precursors. For example, a portion of the precursor ions can be fragmented to form product ions, and the precursor ions and product ions can be analyzed substantially simultaneously, or simultaneously, or, for example, using rapid succession, by applying a rapid switching or alternating voltage to the collision cell of the MS module, switching between low voltage (e.g., mainly generating precursors) and high voltage or elevated voltage (e.g., mainly generating fragments) to adjust the fragmentation. According to the above-mentioned Bateman's technique, MS operation using rapid succession alternating between high (or elevated) energy and low energy can also be referred to as Bateman technique and high-low protocol in this article.
[0101] The data collected by the high-low procedure allows the retention time, mass-to-charge ratio and intensity of all ions collected under both low energy mode and high energy mode to be accurately determined. Generally speaking, different ions are seen in two different modes, and the spectrum collected under each mode can be further analyzed individually or in combination. As seen in one or both modes, ions from a common precursor will share the same retention time (and therefore have substantially the same scan time) and peak shape. The high-low procedure allows the different features of ions to be meaningfully compared within a single mode and between modes. This comparison can then be used to group the ions seen in the low energy spectrum and the high energy spectrum.
[0102] In summary, such as when the Bateman technique is used to operate the system, a sample 402 is injected into the LC / MS system. The LC / MS system generates two sets of spectra, a set of low energy spectra and a set of high energy spectra. The set of low energy spectra mainly contains ions associated with precursors. The set of high energy spectra mainly contains ions associated with fragments. These spectra are stored in a raw data repository 424. After data acquisition, these spectra can be extracted from the raw data repository 424 and displayed and processed by a post-acquisition algorithm in an acquisition device 426.
[0103] Metadata describing various parameters related to data collection can be generated with the raw data. This information can include the configuration of the liquid chromatograph 404 or mass spectrometer 412 (or other chromatographic device that collected the data), which can define the data type. An identifier (e.g., a key) of a codec configured to decode the data can also be stored as part of the metadata and / or with the raw data. The metadata can be stored in a metadata catalog 430 in the document repository 428.
[0104] The collection device 426 can provide the analyst with visualization of data at each of the workflow steps according to the workflow operation, and allow the analyst to generate output data by performing processing specific to the workflow step. The workflow can be generated and retrieved via the client browser 432. When the collection device 426 performs the steps of the workflow, it can read the raw data from the data stream located in the raw data repository 424. When the collection device 426 performs the steps of the workflow, it can generate processed data in the metadata directory 430 stored in the document repository 428; alternatively or in addition, the processed data can be stored in different locations specified by the user of the collection device 426. It can also generate audit records that can be stored in the audit log 434.
[0105] The example embodiments described herein may be executed at the client browser 432 and the collection device 426, among other locations.
[0106] Figure 5 is a flow chart depicting exemplary logic suitable for executing a method for identifying a system ready or error condition. The logic may be stored as instructions on a non-transitory computer readable medium and / or executed by one or more processors.
[0107] At block 502, information of an analytical chemistry instrument may be accessed. This information may include, for example, instrument diagnostic signal data 502a (such as a pressure trace, temperature readings, etc.) and / or one or more output chromatograms 502b generated in response to a request to perform an analysis on a sample. Different types of ready / error conditions may utilize different types of data; for example, detecting a loss of perfusion state may rely on a pressure trace, while detecting an equilibrium state may rely on both a pressure trace and chromatographic data.
[0108] At block 504, machine learning may be applied to detect instrument errors or ready conditions. Machine learning may be used to train classification and / or prediction models to detect system ready / error conditions. Examples of models well suited for use with exemplary embodiments include Bayesian models 504a, gradient boosted trees 504b, and recursive neural networks 504c. Using these types of models, in some embodiments, ready / error conditions may be detected with only a limited amount of system data (e.g., two minutes of data), which allows for high-speed detection of problems and system ready conditions.
[0109] At block 506, the ML system may display a notification of the error or readiness condition on a display device. In some embodiments, the display may include a graphical element that allows the user to take context-specific actions based on the type of error or readiness condition detected. For example, when a readiness condition such as system balance is detected, a selectable element may appear, allowing the user to start a sample analysis run. When an error condition such as a check valve leak is detected, a visual guide may appear that describes the problem to the user and shows the user how to fix it.
[0110] In some embodiments, the ML system can take automatic actions in response to detecting an error or a ready condition. For example, when a ready condition such as equilibrium is detected, the system can automatically start a queued sample analysis run without further input from the user. When a loss of perfusion conditions is detected, the system can automatically terminate the current analysis run.
[0111] The components and features of the devices described above may be implemented using any combination of discrete circuit systems, application specific integrated circuits (ASICs), logic gates, and / or single chip architectures. In addition, where appropriate, microcontrollers, programmable logic arrays, and / or microprocessors, or any combination of the foregoing, may be used to implement the features of these devices. It should be noted that hardware, firmware, and / or software elements may be collectively or individually referred to herein as "logic" or "circuitry."
[0112] It should be understood that the exemplary devices shown in the above block diagrams may represent a functional descriptive example of many possible specific implementations. Therefore, the division, omission or inclusion of block functions depicted in the drawings does not mean that the hardware components, circuits, software and / or elements used to implement these functions will necessarily be divided, omitted or included in the implementation.
[0113] At least one computer-readable storage medium may include instructions that, when executed, cause a system to perform any of the computer-implemented methods described herein.
[0114] Some embodiments may be described using the expressions "one embodiment" or "an embodiment" and their derivatives. These terms mean that a particular feature, structure, or characteristic described in conjunction with the embodiment is included in at least one embodiment. The appearance of the phrase "in one embodiment" in different places in the specification does not necessarily refer to the same embodiment. In addition, unless otherwise stated, the above-mentioned features are considered to be usable together in any combination. Therefore, any features discussed separately can be used in combination with each other, unless it is stated that these features are incompatible with each other.
[0115] The detailed description herein may be presented in terms of program processes executed on a computer or computer network with general reference to the notations and nomenclature used herein. These program descriptions and representations are used by those skilled in the art to most effectively convey the substance of their work to others skilled in the art.
[0116] A process is generally considered to be a self-consistent sequence of operations that brings about a desired result. These operations are those requiring physical manipulation of physical quantities. Although not required, these quantities usually take the form of electrical, magnetic or optical signals capable of being stored, transferred, combined, compared and otherwise manipulated. In practice, it has proven convenient at times to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, etc., primarily for common reasons. It should be noted, however, that all of these and similar terms are associated with the appropriate physical quantities and are merely convenient labels applied to these quantities.
[0117] In addition, the manipulations performed are often referred to by terms such as adding or comparing, which are often associated with mental operations performed by a human operator. In any of the operations described herein that form a part of one or more embodiments, in most cases, such capabilities of a human operator are not necessary or desirable. Instead, these operations are machine operations. Useful machines for performing the operations of the various embodiments include general purpose digital computers or similar devices.
[0118] The expressions "coupled" and "connected," as well as their derivatives, may be used to describe some embodiments. These terms are not necessarily intended to be synonymous with each other. For example, the terms "connected" and / or "coupled" may be used to describe some embodiments to indicate that two or more elements are in direct physical or electrical contact with each other. However, the term "coupled" may also mean that two or more elements are not in direct contact with each other, but still cooperate or interact with each other.
[0119] Various embodiments also relate to devices or systems for performing these operations. The device may be specially constructed for the desired purpose, or it may include a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. The processes presented herein are not inherently related to a particular computer or other device. Various general-purpose machines may be used with programs written according to the teachings of this article, or it may prove convenient to construct more specialized devices to perform the desired method steps. The desired structures of various of these machines will appear in the description given.
[0120] It should be emphasized that the abstract of the present disclosure is provided so as to allow the reader to quickly determine the essence of the technical disclosure. It should be understood that the submitted abstract will not be used to interpret or limit the scope or meaning of the claims. In addition, in the aforementioned specific embodiments, it can be seen that various features can be grouped together in a single embodiment for the purpose of simplifying the present disclosure. This method of the present disclosure should not be interpreted as reflecting the intention that the claimed embodiment requires more features than the features explicitly stated in each claim. On the contrary, as reflected in the following claims, the subject matter of the present invention is less than all the features of a single disclosed embodiment. Therefore, the following claims are hereby incorporated into the specific embodiments, wherein each claim exists independently as a separate embodiment. In the appended claims, the terms "including" and "in which" are used as the colloquial English equivalents of the corresponding terms "comprising" and "wherein", respectively. In addition, the terms "first", "second", "third", etc. are used only as markers and are not intended to impose numerical requirements on their objects.
[0121] What has been described above includes examples of the disclosed architecture. Of course, it is not possible to describe every conceivable combination of components and / or methods, but one of ordinary skill in the art will recognize that many further combinations and permutations are possible. Therefore, the novel architecture is intended to encompass all such changes, modifications, and variations that fall within the spirit and scope of the appended claims.
Claims
1. A computer-implemented method, the computer-implemented method comprising: Access information on analytical chemistry instruments; applying a machine learning model to the information, the machine learning model configured to detect one or more of an instrument error condition or an instrument ready condition; and A result of applying the machine learning model to the information is displayed on a display, wherein the result includes a notification that the error condition or the ready condition has occurred.
2. The computer-implemented method of claim 1, wherein the analytical chemistry instrument is a liquid chromatography (LC) device.
3. The computer-implemented method of claim 1, wherein the information is one or more of instrument diagnostic signal data or a chromatogram generated based on an output of the analytical chemistry instrument.
4. The computer-implemented method of claim 1 , wherein the instrument error condition or the instrument ready condition comprises one or more of a primed / unprimed state, a balanced / unbalanced state, a check valve leak, a pressure seal leak, a degasser failure, a clogged injection valve, a partially clogged needle, a contaminated column, a chemically and / or thermally balanced column, or a stable and / or non-drifting detector.
5. The computer-implemented method of claim 1, wherein the machine learning model comprises one or more of a Bayesian hierarchical model, a gradient boosted tree, or a recursive neural network.
6. A non-transitory computer-readable storage medium comprising instructions that, when executed by a computer, cause the computer to: Access information on analytical chemistry instruments; applying a machine learning model to the information, the machine learning model configured to detect one or more of an instrument error condition or an instrument ready condition; and A result of applying the machine learning model to the information is displayed on a display, wherein the result includes a notification that the error condition or the ready condition has occurred.
7. The computer readable storage medium of claim 6, wherein the analytical chemistry instrument is a liquid chromatography (LC) device.
8. The computer-readable storage medium of claim 6, wherein the information is one or more of instrument diagnostic signal data or a chromatogram generated based on an output of the analytical chemistry instrument.
9. The computer-readable storage medium of claim 6, wherein the instrument error condition or the instrument ready condition comprises one or more of a primed / unprimed state, a balanced / unbalanced state, a check valve leak, a pressure seal leak, a degasser failure, a clogged injection valve, a partially clogged needle, a contaminated column, a chemically and / or thermally balanced column, or a stable and / or non-drifting detector.
10. The computer-readable storage medium of claim 6, wherein the machine learning model comprises one or more of a Bayesian hierarchical model, a gradient boosted tree, or a recursive neural network.
11. A computing device, the computing device comprising: processor; and a memory storing instructions that, when executed by the processor, configure the device to: Access information on analytical chemistry instruments; applying a machine learning model to the information, the machine learning model configured to detect one or more of an instrument error condition or an instrument ready condition; and A result of applying the machine learning model to the information is displayed on a display, wherein the result includes a notification that the error condition or the ready condition has occurred.
12. The computing device of claim 11, wherein the analytical chemistry instrument is a liquid chromatography (LC) device.
13. The computing device of claim 11, wherein the information is one or more of instrument diagnostic signal data or a chromatogram generated based on an output of the analytical chemistry instrument.
14. A computing device according to claim 11, wherein the instrument error condition or the instrument ready condition includes one or more of a primed / unprimed state, a balanced / unbalanced state, a check valve leak, a pressure seal leak, a degasser failure, a blocked injection valve, a partially blocked needle, a contaminated column, a chemically and / or thermally balanced column, or a stable and / or non-drifting detector.
15. The computing device of claim 11, wherein the machine learning model comprises one or more of a Bayesian hierarchical model, a gradient boosted tree, or a recursive neural network.
