Intelligent data processing method and system for inductively coupled plasma mass spectrometry
By using intelligent internal standard matching and multi-dimensional parameter correlation analysis, combined with internal standard dimensional correlation judgment and Cook distance, the problem of high manual verification cost in ICP-MS data processing is solved, realizing automated data processing and improving detection efficiency and accuracy.
Patent Information
- Application Number
- CN202610357867.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-23
- Publication Date
- 2026-06-26
Smart Images
Figure CN122290788A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of bioanalytical detection technology, and in particular to an intelligent data processing method and system for inductively coupled plasma mass spectrometry. Background Technology
[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.
[0003] Inductively Coupled Plasma Mass Spectrometry (ICP-MS) has been the "gold standard" in elemental analysis since its commercialization in the 1980s. Its core logic is to combine the high-temperature ionization characteristics of inductively coupled plasma (ICP) with the high-sensitivity detection capability of mass spectrometry (MS), achieving trace element detection capabilities from ppt to ppb levels.
[0004] The working principle of ICP-MS is a highly efficient conversion process from macroscopic liquid samples to microscopic ion counting. Its limitations stem from the interplay between hardware physical boundaries and dynamic matrix interference. First, ionization efficiency is severely affected by matrix suppression; high concentrations of easily ionized elements can cause nonlinear signal drops that are difficult to correct with traditional linear compensation. Second, limited by separation precision, a single quadrupole only provides unit mass resolution and cannot physically distinguish between isotopes and polyatomic ions (e.g., 40Ar16O+ versus 56Fe), resulting in a high risk of false positives.
[0005] Traditional ICP-MS data processing involves four steps: signal acquisition, baseline correction, peak identification, and quantitative analysis. Early methods used moving average algorithms for signal denoising, but these were prone to signal drift when dealing with complex matrix samples. With advancements in computer technology, quantitative calibration methods such as baseline fitting algorithms based on least squares and internal standard methods have emerged. These algorithms have significantly improved the reliability of ICP-MS data processing. Current models are mostly offline, operating on a "acquisition first, processing later" basis, heavily reliant on manual experience for spectral interpretation and outlier removal. Due to the analytical characteristics of ICP-MS, a single injection may generate intensity data involving dozens of elements and hundreds of isotopes. Pre-set linear analysis conditions are often insufficient to meet the demands of batch processing of complex linear data, and the cost of manual verification remains high.
[0006] In the development of ICP-MS, data processing technology has undergone a significant evolution from basic algebraic operations to high-order artificial intelligence algorithms. Since the early stages of its commercialization, interference correction equations have been widely used to handle isotope overlap problems. The core of this approach lies in establishing linear correction logic using a fixed proportion of the natural abundance of isotopes. This method has clear physical meaning, strong interpretability, and low computational resource consumption, perfectly meeting the quality control needs of traditional laboratories. However, its inherently static correction characteristics make it difficult to detect real-time shifts in instrument quality bias, and it is also highly susceptible to error propagation when dealing with high background noise.
[0007] To analyze complex mass spectrometry data from a higher dimension, chemometric methods such as principal component analysis (PCA) have been introduced. Through orthogonal transformations, PCA simplifies the high-dimensional data structure while preserving as much original information as possible, thereby enabling the classification of sample patterns and a more intuitive display of experimental trends. Parallel to PCA, partial least squares (PLS) demonstrates unique advantages in quantitative analysis. By extracting latent variables that represent both signal intensity fluctuations and maximize the prediction of element concentrations, PLS automatically filters out noise signals that do not contribute to the prediction, effectively solving the problem of collinearity interference among multiple elements and significantly improving the robustness of the model under matrix suppression conditions.
[0008] In recent years, as analytical demands have shifted towards extremely low detection limits and complex nonlinear matrices, machine learning intelligent algorithms have become one of the core methods for overcoming the hardware bottlenecks of ICP-MS. Researchers have used artificial neural networks (ANNs) or random forests (RF) to construct interference prediction models, deeply exploring the implicit mapping relationship between instrument parameters such as carrier gas flow rate and RF power and polyatomic ion formation, achieving "intelligent stripping" of interference signals under conditions without physical collision cells. Simultaneously, addressing the pain point of internal standard method failure in long-term testing, the introduction of time series prediction models such as Long Short-Term Memory (LSTM) networks has enabled the system to learn signal attenuation patterns and achieve self-compensation for performance degradation, significantly improving the long-term stability of the data. These intelligent methods have collectively driven a historic leap in ICP-MS data processing from "passive correction" to "active sensing."
[0009] Existing processing methods generally face a deep contradiction in clinical and high-precision analysis due to the mismatch between linear logic and nonlinear physical processes. First, because the interference correction equation follows the error propagation law, it amplifies statistical noise when dealing with "strong interference and weak signals," leading to deterioration in precision or even negative values. Furthermore, the nonlinear characteristics of plasma fluctuations often cause correction inaccuracies. Second, PCA and PLS are essentially linear models, making it difficult to capture nonlinear interferences such as space charge effects or complex ionization suppression. PCA tends to ignore crucial minute differences, while PLS is highly dependent on the manual setting of the number of principal components, posing a risk of underfitting or overfitting, and lacks intuitive explanatory power for physical interference paths.
[0010] While emerging machine learning algorithms possess the potential for proactive modeling, their complexity and reliance on massive amounts of data for training remain significant challenges, resulting in insufficient technological maturity and market trust. Currently, clinical analysis still heavily relies on basic algorithms like internal standards combined with manual verification, failing to substantially reduce the time costs for laboratory personnel. Summary of the Invention
[0011] To address the technical bottlenecks of existing ICP-MS in complex sample analysis, which rely heavily on the experience of experimenters, involve a large workload for data interpretation and verification, and have low efficiency in anomaly identification, this invention provides an intelligent data processing method and system for inductively coupled plasma mass spectrometry, which can maintain the reliability of traditional linear calibration while significantly reducing labor costs.
[0012] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: In a first aspect, the present invention provides an intelligent data processing method for inductively coupled plasma mass spectrometry, comprising: The inductively coupled plasma mass spectrometry (ICP-MS) data of the element to be measured were acquired and preprocessed to obtain the preprocessed ICP-MS data. Multidimensional parameter correlation analysis was performed on the preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results; The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
[0013] A further technical solution is to trigger a system risk warning when at least one of the following situations occurs in the internal standard dimension correlation judgment: the internal standard recovery rate is continuously higher than the preset upper limit threshold, instrument physical drift or hardware performance deterioration is detected, abnormality in the pretreatment or injection process is detected, or memory effect or residue is detected.
[0014] In a further technical solution, the internal standard intelligent matching introduces the nearest mass number and the first ionization energy as the core matching dimensions.
[0015] In a further technical solution, the intelligent matching of internal standards follows the nearest-neighbor matching principle and automatically assigns priority to the internal standards of the elements to be tested.
[0016] A further technical solution involves determining the correlation between internal standards. If the internal standard is not affected by mass spectrometry, the current internal standard is used to correct the signal of the element to be measured, resulting in net response data, which is then processed using linear point intelligent processing. If a systemic risk warning is triggered, the test is interrupted.
[0017] A further technical solution involves constructing an automated residual monitoring threshold using intelligent linear point processing. When the absolute value of the relative deviation of the residual of a data point is greater than or equal to the automated residual monitoring threshold, the point is automatically identified as a rejection point, and the rejection priority is determined based on the degree of dispersion.
[0018] A further technical solution is that the precise identification of linear outliers specifically involves: constructing an automated residual monitoring operator to calculate the residuals of each calibration point and the sample point to be tested in real time; when the absolute value of the standardized residual is detected to be greater than or equal to the standardized residual threshold and the Cook distance meets the set conditions, the point is automatically determined to be an abnormal discrete point and an anomaly flag is triggered.
[0019] Secondly, the present invention provides an intelligent data processing system for inductively coupled plasma mass spectrometry, comprising: The data preprocessing module is configured to: acquire the inductively coupled plasma mass spectrometry data of the element to be measured and preprocess it to obtain the preprocessed inductively coupled plasma mass spectrometry data; The multidimensional parameter correlation analysis module is configured to perform multidimensional parameter correlation analysis based on preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results. The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
[0020] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of an intelligent data processing method for inductively coupled plasma mass spectrometry as described in the first aspect.
[0021] Fourthly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the intelligent data processing method for inductively coupled plasma mass spectrometry as described in the first aspect.
[0022] The above one or more technical solutions have the following beneficial effects: This invention solves the problem of cumbersome manual calculation costs caused by nonlinear distortion. Addressing the industry pain point of linear distortion in calibration curves and dense clusters of abnormal response points due to complex matrix interference, this invention constructs a simulated intelligent data processing system. This system can autonomously identify outlier data affected by nonlinear factors and replace the traditional manual verification process with an algorithmic closed-loop system, effectively reducing the time spent by laboratory personnel in data processing and achieving cost reduction and efficiency improvement in clinical mass spectrometry analysis. Attached Figure Description
[0023] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0024] Figure 1 This is a flowchart of an intelligent data processing method for inductively coupled plasma mass spectrometry according to an embodiment of the present invention; Figure 2 This is a flowchart of intelligent internal standard matching in an embodiment of the present invention. Detailed Implementation
[0025] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0026] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0027] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.
[0028] Existing technologies have significant gaps in adaptive nonlinear compensation and fully automated data processing, and there is an urgent need for a new intelligent solution that can be quickly implemented to solve the most challenging problem of manual verification costs.
[0029] Example 1 like Figure 1 As shown, this embodiment discloses an intelligent data processing method for inductively coupled plasma mass spectrometry (ICP-MS). Addressing the low automation level and manual reliance bottlenecks in clinical ICP-MS, it constructs an intelligent system of "white-box leadership and black-box evolution." Integrating an interference knowledge graph and multi-dimensional association algorithms, it transforms sensory judgment into a logical matrix supported by a chain of physical evidence. Through a closed-loop linkage of "judgment-rejection-matching," it achieves a paradigm shift from manual review to automated approval, physical diagnosis, and proactive risk isolation. While ensuring compliance and transparency, it utilizes data training to drive the dynamic evolution of the algorithm, significantly improving the efficiency of large-scale sample testing and realizing an intelligent leap from passive data correction to proactive preventative maintenance.
[0030] The system includes a built-in experimental method editing module, allowing users to select internal standard elements and establish correlations between the analytes and the internal standards. Taking fully quantitative analysis as an example, the system includes an internal standard matching system and a linear fitting system.
[0031] The method includes the following steps: S1: Acquire the inductively coupled plasma mass spectrometry data of the element to be measured and preprocess it to obtain the preprocessed inductively coupled plasma mass spectrometry data; In this embodiment, the raw count frequency (CPS, Counts Per Second) is obtained after mass spectrometry analysis of the sample. The CPS value is a direct indicator of ion detection intensity and is the core basis for subsequent quantitative calculations. Based on the CPS value and the current experimental preset method, data preprocessing is performed, and statistical parameters are calculated.
[0032] Statistical parameters include concentration residuals Standardized residuals Distance from Cook .
[0033] S2: Multidimensional parameter correlation analysis was performed on the preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results; The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
[0034] (a) Determining the correlation between internal standard dimensions In this embodiment, the internal standard dimension association judgment is used to detect the rationality and interference status of the current internal standard matching mechanism.
[0035] Internal standard dimension correlation judgment: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed to determine whether the current internal standard is affected by mass spectrometry. If not, the current internal standard is used to directly correct the signal of the analyte element; if so, the internal standard intelligent matching is triggered, and the internal standard intelligent matching module is entered to start the internal standard polling and matching process, and the multi-dimensional parameter correlation analysis module under the current internal standard matching mechanism is entered simultaneously. The internal standard dimension correlation judgment module determines whether to trigger a system risk warning. The internal standard dimension correlation judgment module takes into account both clinical regulatory requirements and user personalized needs, and supports custom configuration of whether to execute the experiment termination command.
[0036] Specifically, traditional methods rely on the ideal assumption that the internal standard and the analyte have completely identical response characteristics. The software system monitors the response intensity of the internal standard element in each sample in real time, performs diagnostic analysis on the internal standard volatility, and diagnoses various situations related to the internal standard dimension as follows: (1) The internal standard recovery rate is consistently higher than the preset upper limit threshold. Correlation analysis: Under certain specific conditions (such as the ionization enhancement effect of a carbon-containing matrix or the reduction of charge repulsion caused by a light matrix), the ionization efficiency of the internal standard element may increase, or the ion current entering the mass spectrometer cone may increase. Check whether the CPS of the analyte increases synchronously. If all analytes shift linearly upwards and the internal standard increases, it is determined to be a systematic matrix enhancement.
[0037] Diagnostic conclusion: Systemic risk warning, suggesting possible carbon enhancement effect or insufficient sample dilution.
[0038] (2) Instrument physical drift and hardware performance degradation Correlation analysis: As the number of batches increases, the sampling cone / removal cone aperture may decrease due to matrix deposition (cone carbon buildup), or the peristaltic pump tubing may age and deform. This can lead to a slow decrease in the amount of aerosol or ion flow entering the instrument. A moving average algorithm is introduced to monitor the internal standard response. If the internal standard signal shows a monotonically decreasing trend across multiple consecutive samples (e.g., a decrease of 0.5% per sample over 20 consecutive samples), the system will trigger an alert even if the hard threshold has not been exceeded.
[0039] Diagnostic conclusion: Systemic risk warning, indicating "sample injection system status warning" or "recommendation to clean the sampling cone / replace the pump tubing".
[0040] (3) Abnormalities in pretreatment and injection processes Correlation analysis: Common human errors in clinical testing (such as failure to add internal standard, doubling the internal standard, or depletion of internal standard solution) or physical malfunctions (such as air bubbles in the injection tube or blockage of the injection needle). When the internal standard recovery rate drops sharply or doubles, or when the relative standard deviation (RSD) of the internal standard signal calculated in real time during the acquisition process increases abnormally.
[0041] Diagnostic conclusion: Systemic risk warning. The error may be due to "internal standard omission / doubling" or "physical injection failure (bubbles / blockage)".
[0042] (4) Internal standard is subject to specific mass spectrometry interference. Correlation analysis: Matrix components in certain specific clinical samples may produce isotopic or polyatomic ion interference with internal standard elements (e.g., samples containing high concentrations of chlorine may produce ArCl+ interference with certain internal standards of medium mass number). If the system is configured with multiple alternative internal standards distributed across different mass number ranges, when some internal standards respond normally while the response of a certain internal standard is abnormally elevated, it is determined that the internal standard is subject to specific mass spectrometry interference.
[0043] Diagnostic conclusion: The "intelligent internal standard matching" logic was triggered, and the system automatically switched to an undisturbed alternative internal standard for correction.
[0044] (5) Memory effect and residue Correlation analysis: Certain elements (such as mercury, gold, or certain easily adsorbed internal standard elements) tend to adhere to the inner wall of the injection tubing. After analyzing high-concentration samples, this can lead to an increase in the internal standard background signal in subsequent samples. In the wash blank after analyzing high-matrix samples, it is monitored whether the internal standard signal can quickly return to the background level.
[0045] Diagnostic conclusion: Systemic risk warning, indicating "internal standard residue risk", it is recommended to extend the cleaning time.
[0046] If the internal standard is interfered with by specific mass spectrometry, intelligent internal standard matching is triggered. Intelligent internal standard matching is a crucial component of this invention. In clinical testing and analysis, simultaneous analysis of multiple internal standards is a common method for compensating for matrix effects and improving detection precision. Traditional processes rely on pre-set calibration equations in software, determining calibration coefficients using blank samples, and then correcting the count rate (CPS) of the analyte for linear fitting and quantification. In this invention, the intelligent internal standard matching module introduces the nearest mass number (m / z) and first ionization energy (IP) as core matching dimensions. Following the "nearest match" principle, it automatically prioritizes internal standards for the analyte; while maintaining transparency and intuitiveness in the calibration logic, it effectively eliminates the randomness of manual selection, significantly reducing the cost of manual intervention while ensuring high throughput and accuracy in clinical testing. The intelligent internal standard matching module considers both clinical regulatory requirements and user-specific needs, supporting custom configuration. Researchers can freely enable or disable this module's function according to their actual validation protocols.
[0047] When the internal standard polling and matching process begins, a multi-dimensional parameter correlation analysis is simultaneously performed under the current internal standard matching mechanism. In other words, the intelligent internal standard matching updates the internal standard matching mechanism, and the updated internal standard data is re-input into the multi-dimensional parameter correlation analysis module for processing, forming a closed-loop iterative verification. The multi-dimensional parameter correlation analysis includes internal standard dimensional correlation judgment and accurate identification of linear outliers, generating a multi-dimensional diagnostic report under the current internal standard matching mechanism. By constructing a multi-dimensional parameter correlation analysis system, the limitations of single-indicator judgment are improved, and a deep correlation diagnostic and processing logic is established.
[0048] Furthermore, such as Figure 2 As shown, the current internal standard matching mechanism is as follows: Based on the mass spectrometry behavior consistency, interference avoidance, and signal stability of the "analyte-internal standard element," it automatically matches the "internal standard element without interference, with synchronized response, and optimal correction effect" from the internal standard candidate pool (pre-defined by element grouping) through preset multidimensional correlation rules. The signal of the selected internal standard element is then used in the algorithm system for drift / matrix effect correction of the analyte. Specifically, a polling process is initiated, internal standard elements are retrieved from the internal standard candidate pool, and multidimensional parameter correlation analysis is performed simultaneously to determine whether the candidate internal standard meets the matching threshold. If not, the next internal standard is polled; if so, the optimal internal standard is locked, and this internal standard is used to complete the signal correction of the analyte, obtaining the net response intensity after signal drift and matrix effect correction, which is then used for subsequent quantification.
[0049] Furthermore, the essence of intelligent internal standard matching is to quantify the suitability of each candidate internal standard by constructing an "internal standard diagnostic analysis and evaluation matrix" on the preprocessed ICP-MS raw signal data, and finally select the internal standard that is "interference-free, highly stable, and has the best correction effect", and output the net response data of the analyte after correction by the internal standard.
[0050] If the internal standard is not affected by mass spectrometry and does not trigger intelligent matching, the current internal standard is used to directly correct the analyte signal, obtaining standardized net response data, including the internal standard ratio and corrected intensity. Internal Standard Ratio (IS Ratio): analyte signal / internal standard signal (acquired in the same batch and at the same time); Signal drift correction coefficient: calculated from the fluctuation of the internal standard ratio at the calibration curve points, used to correct all signals; Net Intensity: the final value after subtracting the blank and multiplying by the drift coefficient, which is the basis for linear fitting.
[0051] Linear point intelligent processing is performed on the standardized net response data. Specifically, based on the boundary protection verification principle, the first and last two points of the calibration curve (i.e., the lowest and highest concentration points) are determined and forcibly retained. The linear point intelligent processing result under the current matching internal standard condition is obtained through linear standard judgment. After verifying the curve refit: 1) Evaluation of linear regression coefficients: The linear regression coefficients after verifying the curve refit... r 1) Whether the requirements are met; 2) Outlier threshold assessment: After executing the intelligent processing flow, verify whether the outlier threshold meets the requirements. For the intelligent processing results of linear points, perform linearity standard judgment again. If the requirements are met / although not met, the internal standard has been polled and matched, then output the result; if not met and the internal standard polling and matching has not been completed, then perform the internal standard intelligent matching loop.
[0052] Furthermore, intelligent processing of linear points is an important component of this invention, specifically: by constructing an automated residual monitoring threshold. Simulates the sensory recognition of rejection points by inspectors. When a point is identified as a rejection point, the system automatically determines the rejection priority based on its dispersion. In accordance with clinical data standards, the system automatically skips the highest and lowest calibration points of the curve during rejection. While meeting the preset number of retention points, the system considers both the outlier severity of the rejection point and the principle of "maximizing the number of retention points," while also taking into account the linear regression coefficient as much as possible. r In cases where the value is not specified, intelligent rejection is implemented. This automated judgment mechanism replaces the manual operation of observing the residual distribution of the calibration curve one by one, significantly reducing the labor intensity of manual review.
[0053] For the The relative deviation of the concentration residuals for each data point. The formula is:
[0054] in, The vertical axis represents the linear regression equation, which is generally the CPS value of the test element after internal standard correction or the ratio of the CPS of the test element to the CPS of the internal standard element. The slope of the linear regression equation represents the sensitivity of the instrument analysis. The intercept of the linear regression equation represents the instrument background signal or blank value. Indicates the first The theoretical concentration value of each sample point; Indicates the first The relative deviation of the residuals for each sample point.
[0055] The Linear Point Intelligent Processing Module takes into account both clinical regulatory requirements and personalized user needs. This module supports custom configuration, and experimenters can freely enable or disable the module's function according to the actual validation protocol.
[0056] If a systemic risk warning is triggered during the internal standard dimension association judgment, the test will be interrupted.
[0057] (II) Precise identification of linear outliers Accurate identification of linear outliers introduces standardized residuals z i Distance from Cook D i A dual evaluation system is implemented. An automated residual monitoring operator is constructed to calculate the residuals of each calibration point and the sample point under test in real time. The system sets a standardized residual threshold. (For example ), simulating sensory recognition of abnormal discrete points. When detected When this happens, the system automatically identifies the point as an abnormal discrete point and triggers an anomaly flag. If the data point ( x i , y i The absolute value of the standardized residuals Exceeding the preset threshold (e.g.) (principle), and its Cook distance satisfy ( n If the sample size is given, then this point is defined as an outlier affecting the reliability of the model in the ICP-MS experiment. The standardized residual is used as a unified criterion for outlier judgment, expressed as:
[0058] in, Indicates the first The actual measured value of each sample point is generally the CPS value of the element to be measured after internal standard correction or the ratio of the element to be measured to the internal standard element. Indicates the first The ordinate value of the linear regression equation for each sample point; The standard deviation of the residuals; The value represents the leverage, which indicates the influence of that point on the fitted curve.
[0059] Table 1 Cook Distance Threshold Evaluation Table
[0060] Based on the judgment results, the following system analysis conclusions are given: (1) Slight deviation. (2) Significant outlier with low leverage: judged as local interference (such as injection fluctuation). (3) Significant outlier with high leverage: judged as systemic risk (such as standard dilution error or matrix suppression).
[0061] In summary, multidimensional parameter correlation analysis ultimately completes the entire intelligent process of internal standard selection and outlier removal. After completing automated data processing, the intelligent system integrates multidimensional parameter correlation analysis diagnosis and simultaneously outputs a report containing intelligent data processing results and risk warnings. This report aims to assist laboratory personnel in quickly identifying potential risks, reduce the burden of manual review, and provide efficient and accurate reference for clinical auxiliary diagnosis.
[0062] In this embodiment, traditional outlier handling algorithms (such as residual analysis and Cook distance) are essentially numerical logic; they can tell experimenters that "the data is wrong," but cannot answer "why it's wrong." Expert systems based on interference knowledge graphs, by introducing semantic logic, structure fragmented physicochemical principles, instrument state parameters, and clinical matrix characteristics, constructing a diagnostic system with "human-like expert" thinking, providing transparent and traceable underlying evidence for clinical auxiliary diagnosis.
[0063] The specific steps for establishing a dynamically updated interference knowledge graph are as follows: (1) Constructing the multidimensional topology of the knowledge graph. The multidimensional topology of the knowledge graph includes an entity layer and a relation layer.
[0064] The entity layer includes target elements and isotopes: covering clinically essential elements (such as As, Cd, Hg, Pb) and their natural abundance, ionization energy, and mass number; interfering entities: including polyatomic ions (such as ArCl). + , ClO + ), isotopes (such as) 114 Cd and 114 Sn), dicharged ions (M) 2+ Sample matrix entities: refine clinical sample types (whole blood, serum, urine, hair) and associate them with their unique high-content components (such as high sodium, high chloride, high lipid); hardware parameter entities: pump speed, nebulizer gas flow rate, plasma power, cone tip lifespan, etc.
[0065] The relationship layer includes interference triggering relationships: for example, a "high-chlorine matrix" "generates" ArCl in plasma. +"Interference" m / z is 75 As detection; internal standard coupling relationship: determining the influence weight of a specific matrix on different internal standards (such as Ge, In, Bi); causal diagnostic link: establishing a logical link from "decreased internal standard recovery" to "physical obstruction of sample loading" or "space charge effect".
[0066] (2) Through the diagnostic mechanism of multidimensional pattern matching, an expert system is established and the system is trained in real time. The inference engine dynamically matches the real-time collected data stream with the knowledge graph.
[0067] 1) Automatic deconvolution of spectral interference The expert system accurately identifies linear outliers through residual analysis and Cook distance judgment. It immediately searches the spectrum for relevant polyatomic interferences. If an abnormal signal is simultaneously detected in the Cl monitoring channel (or internal standard response fluctuation), the system will use its built-in Interference Correction Equation (IEC) or Kinetic Discrimination (KED) principle for logical comparison to determine whether the outlier is due to a genuine increase in concentration or an increase in ArCl concentration. + His contributions caused this.
[0068] 2) In-depth profiling of non-spectral interference (matrix effects) The expert system uses internal standard dimension correlation judgment to perform volatility analysis on internal standards and match matrix effect models in the spectrum. For example, if the recovery deviation between light mass number internal standards (such as Sc) and heavy mass number internal standards (such as Bi) shows a nonlinear expansion, the expert system will call the space charge effect knowledge node to diagnose the presence of high concentration of salts or heavy metal background in the sample and provide suggestions for optimizing the dilution ratio or changing the collision cell parameters.
[0069] Proactive Risk Isolation: Current intelligent data processing systems primarily focus on data verification after collection. This invention, by introducing time series analysis, upgrades the system into a lifecycle management system with proactive early warning capabilities. When the current interference pattern is identified as a systemic risk (such as stepwise signal attenuation caused by gradual clogging of the atomizer), the system will proactively trigger an alarm / interception mechanism, not only eliminating the current outlier but also issuing warnings for subsequent samples in the batch to prevent erroneous diagnostic reports.
[0070] Adaptive parameter optimization: The diagnostic conclusion of the expert system is not only a result, but also a chain of evidence based on a knowledge graph. For example, the report shows: "Determined as an outlier, cause: suspected chloride interference; basis: abnormal 75As residual, simultaneous monitoring of Cl channel response increased, consistent with the ArCl+ polyatomic ion interference model." This transparency ensures that the system meets the stringent requirements of clinical laboratories for data traceability.
[0071] With the massive accumulation of clinical testing data, the system inputs "latent anomalies" not fully covered by white-box rules in intelligent data processing into black-box models (such as ensemble learning or deep neural networks). The final output analysis report contains both high-order pattern predictions from the black-box model and underlying evidence support from the white-box model. This not only eliminates the opacity brought about by data black-boxing but also endows the system with dynamic evolutionary processing capabilities that transcend traditional rules, truly realizing a complete transformation from "manual full review" to "automatic system approval and processing."
[0072] This invention addresses the issues of "algorithm black box" and lack of trust in artificial intelligence models. Specifically, it addresses the drawbacks of black-box models such as deep learning in clinical diagnosis, including uninterpretability, missing physical evidence chains, and difficulties in tracing compliance. This invention constructs a progressive intelligent processing system based on white-box logic guidance and black-box data evolution. Specifically: (1) Establishing a compliance foundation based on white-box logic: This invention rejects blind reliance on end-to-end deep learning, but instead establishes a statistical discrimination mechanism (such as Cook distance) by simulating the judgment logic of senior analysts. D i Standardized residuals z i The white-box decision matrix is based on physicochemical criteria (such as internal standard recovery rate and first ionization energy matching). This transparent and intuitive architecture ensures that the removal of every outlier and the switching of every internal standard are traceable, fundamentally eliminating the sense of "loss of control" and crisis of trust that clinical laboratories have with automated algorithms.
[0073] Furthermore, the white-box decision matrix = statistical discrimination rules + physicochemical criteria + interpretable decision logic. This invention proposes a white-box decision matrix as the core cornerstone of algorithm compliance and credibility. This matrix abandons the uninterpretable end-to-end deep learning black-box model and adopts fully explicit, reproducible, and verifiable decision logic. It simulates the manual interpretation thinking of senior mass spectrometry analysts and constructs a two-dimensional comprehensive decision system composed of statistical discrimination indicators and physicochemical criteria. This enables transparent decision-making throughout the entire process, including calibration point validity, outlier identification and removal, internal standard selection and switching, and interference cause diagnosis.
[0074] The white-box decision matrix uses the sample points to be decided as rows and the quantitative decision dimensions as columns. Each cell in the matrix is a calculable, comparable, and traceable indicator value or qualified decision result. Finally, the decision conclusion is output through preset explicit combination rules to ensure that every anomaly removal and every internal standard switch is based on evidence and is traceable.
[0075] 1) Statistical discriminant submatrix Used to mathematically determine the quality of curve fitting and the influence of sample points, including: Standardized residuals: Used to determine the degree of deviation of a single point from the regression line and to identify obvious outliers; Cook distance: By combining residuals and leverage values, it identifies high-influence points that have a strong impact on the slope and intercept of the curve.
[0076] The system uses a dual evaluation system of standardized residuals and Cook distance to make joint judgments: only points with large residuals but weak impacts are retained; points with Cook distances exceeding the standard and standardized residuals exceeding the threshold are judged as strong impact anomalies and are removed; at the same time, the entire calculation process is retained for review and traceability.
[0077] 2) Physicochemical Criterion Submatrix To ensure the physical rationality of the decision from the perspective of mass spectrometry behavior and interference mechanisms, including: Internal standard recovery rate: to determine whether the internal standard is affected by matrix effects, ionization inhibition, or mass spectrometry interference; First ionization energy matching degree: to determine the consistency of the ionization behavior between the internal standard and the analyte; Mass spectrometry interference matching rules: Based on mass-to-charge ratio, polyatomic ions, and isotope interference libraries, determine whether the current internal standard or analyte has known mass spectrometry interference; Synchronization of the response of the internal standard and the element under test: Determine whether the signal change trends of the two are consistent.
[0078] 3) Comprehensive decision logic of white-box matrix The white-box decision matrix executes decisions according to explicit hierarchical rules: First, the internal standard is judged to be stable and whether there is matrix effect or mass spectrometry interference by physical and chemical criteria. If the internal standard recovery rate exceeds the limit, the response synchronization is poor or there is obvious interference, the internal standard intelligent matching and polling optimization are automatically triggered. Then, outlier identification is performed on the calibration curve using a statistical discrimination system. Combining standardized residuals and Cook distance as dual indicators, normal discrete points, ordinary outliers, and strong-influence outliers are accurately distinguished. Only sample points that simultaneously meet both the physicochemical criteria and statistical indicators are considered valid calibration points; if either dimension is not met, a clear label, removal suggestion, or interfering diagnostic conclusion will be given.
[0079] This invention achieves intelligent data processing that is interpretable, verifiable, and auditable throughout the entire process through the aforementioned white-box decision matrix. This fundamentally eliminates the sense of loss of control and trust crisis that clinical laboratories face regarding automated algorithms, and meets the stringent requirements for algorithm transparency, compliance, and stability in the fields of in vitro diagnostics and clinical mass spectrometry.
[0080] (2) Functional leap driven by interference knowledge graph: This invention overcomes the limitations of traditional numerical cleaning by constructing an expert system, and integrates matrix effects (such as space charge effect and ionization suppression) with mass spectrometry interference (such as polyatomic ions ArCl).+ The expert-level interference knowledge graph enables a leap from simple data filtering to auxiliary diagnosis of anomalies. When the system detects linear deviation, it automatically associates it with internal standard fluctuation characteristics and outputs diagnostic suggestions with physical context based on the knowledge graph, providing a scientific basis for laboratory quality control decisions.
[0081] (3) Achieving intelligent evolution through the fusion of black-box and white-box: While maintaining white-box logic as the core decision-making criterion, the system continuously trains the black-box model using massive amounts of historical clinical data. The black-box model is responsible for perceiving subtle nonlinear patterns and latent interference fingerprints, while the white-box logic is responsible for verifying the physical rationality of the black-box output. This deep fusion of black-box anomaly perception and white-box approval logic enables the system to have the vitality of self-learning and continuous optimization while ensuring clinical compliance.
[0082] Furthermore, in this invention, the white-box component is a white-box decision matrix, which consists of statistical discriminant indicators such as standardized residuals and Cook distance, and physicochemical criteria such as internal standard recovery rate, first ionization energy matching, and mass spectrometry interference knowledge graph. It adopts fully explicit and interpretable rules to achieve outlier removal, internal standard optimization, and interference diagnosis, providing the system with a compliance foundation and final decision-making power.
[0083] The black box part is a machine learning perception model, which adopts gradient boosting tree (LightGBM) or random forest model and is trained based on massive clinical mass spectrometry historical data. The input includes multi-dimensional features such as the test element and internal standard signal, time series features, residuals, Cook distance, and interference labels, and learns latent interference fingerprints and nonlinear deviation patterns.
[0084] The trained black-box model is deployed between data preprocessing and white-box decision-making to detect latent anomalies and early interference in advance, and output the probability of anomalies, the type of interference, and suspected abnormal sites. The black-box output needs to be physically and statistically verified by the white-box decision matrix before a final decision can be made, thereby achieving intelligent self-evolution of the system while ensuring algorithm transparency and clinical compliance.
[0085] This invention implements an active risk isolation and early warning mechanism based on an interference knowledge graph. In response to the batch quality risks caused by hardware performance degradation or systematic drift in clinical testing, this invention constructs an active risk isolation mechanism that shifts from post-correction to pre-blocking. This mechanism deeply integrates physical condition parameters and interference knowledge graphs, realizing preventive quality control throughout the entire clinical testing process. Specifically: (1) Evolving from static truncation to dynamic trend perception, traditional quality control often relies on passive error reporting after exceeding a fixed threshold, which has obvious lag. This invention introduces an internal standard dimension data diagnosis mechanism, which can monitor the drift trend of internal standard response intensity and matrix signal in real time. (2) Active risk isolation and early warning mechanism. Relying on the built-in expert-level interference knowledge graph, the system transforms real-time signal anomalies into physical cause diagnoses with guiding significance. Through multi-dimensional parameter correlation analysis, the system can effectively distinguish between occasional matrix interference and trend hardware failures. If it is determined to be an unacceptable systematic risk, the expert system will actively intercept subsequent analysis sequences and output a diagnostic report containing physical background (such as atomization efficiency fluctuations, mass spectrometry interference fingerprints) to ensure the physical authenticity of the output results. This mechanism represents a leap from passive firefighting to proactive risk avoidance. By providing early warnings of potential hardware hazards, it significantly reduces the sample retesting rate and provides laboratories with scientific and traceable digital evidence of quality compliance.
[0086] In summary, the technical effects achieved by the solution of this invention are as follows: This invention significantly improves analytical efficiency and reduces the manual cost of multi-dimensional review. It transforms the complex logic that previously relied on manual review by experienced personnel into an automated closed-loop process of decision-making, elimination, and matching. Through multi-dimensional parameter correlation analysis, the system can not only automatically identify nonlinear outliers but also simultaneously perform cross-validation by correlating physical conditions such as internal standard recovery rates, replacing the tedious point-by-point verification and manual recalculation process of CPS (Continuous Physical System). In high-volume clinical testing tasks, this automated decision-making system significantly reduces the time invested in data post-processing, freeing laboratory personnel from inefficient numerical calibration and allowing them to focus on high-value diagnostic decisions.
[0087] A black-box and white-box integrated system is constructed, taking into account both transparency and evolution capabilities. In response to the extreme requirements of clinical diagnosis for interpretability and physical authenticity, this invention breaks through the black-box limitation of traditional AI: (1) White-box logic establishes trust: By simulating the logic of human judgment, a physical evidence chain based on the matching of mass number and ionization energy is constructed to ensure that the underlying correction logic is intuitive and traceable, and completely eliminates the crisis of trust in automated processing in clinical practice. (2) Black-box drives intelligent evolution: The functional leap from numerical cleaning to physical diagnosis is realized by using interference knowledge graph. The system can not only handle outliers, but also provide accurate auxiliary diagnostic suggestions based on the knowledge base, and realize the dynamic self-optimization of the algorithm with the accumulation of data, thus constructing a high-transparency and sustainable evolution quality control system.
[0088] Shifting from passive correction to proactive prevention, optimizing compliance and operating costs. This invention achieves closed-loop management through software algorithms, eliminating the need for expensive hardware facilities such as online dilution, and realizing true zero-hardware performance upgrades: (1) Proactive risk isolation: The system has a forward-looking early warning mechanism. By monitoring signal drift trends, it implements preventive maintenance prompts before data deterioration, significantly reducing sample retesting rates and reagent consumption. (2) Automated compliance traceability: A complete electronic evidence chain and knowledge graph-based anomaly diagnosis records provide the laboratory with detailed digital quality control evidence, greatly simplifying traceability work during compliance audits and realizing scientific control and cost optimization of the entire testing process.
[0089] Specific examples: 1. Create a blank method and edit the experimental method flow according to the process requirements. The editing content includes: selection of acquisition parameters (including selection of internal standards and pre-matching of internal standards), quantitative methods (including linear fitting functions, weights, and forced methods), etc.
[0090] (1) Selecting elements: 45 Sc、 57 Fe、 66 Zn, 89 Y、 103 Rh; (2) Preset quantitative parameters.
[0091] Table 2 Quantitative Parameter Table
[0092] (3) Example of preset internal index element: a. Check the internal standard: 45 Sc、 89 Y、 103 Rh; b. Internal standard preset: 57 Fe、 66 Zn corresponding internal standard 45 Sc; 2. Obtain raw data: The sample enters the ion source through the injection system, and the raw count cps value is obtained by mass spectrometry analysis.
[0093] Table 3 Original CPS Table of Elements
[0094] Table 4 Element Recovery Rate Table
[0095] 3. Calculate statistical parameters: calculate concentration residuals, standardized residuals, and Cook distance.
[0096] Table 5 Statistical Parameter Table
[0097] 4. Multidimensional parameter correlation analysis.
[0098] (1) Internal standard dimension correlation judgment: 80%~120% is set as normal, and below 80% is abnormal; after judgment, the current matching internal standard is within the normal fluctuation range and there is no abnormality.
[0099] (2) Accurate identification of linear outliers: setting the absolute value of standardized residuals The value of k is set to 3.5; it is determined that under the current internal standard matching mechanism, 57 The linear outlier for Fe is "standard curve 05". 66 Zn has no wireless anomaly, Cook distance D i It is within the normal range and is judged to be a slight deviation.
[0100] 5. Internal standard dimension association judgment: "Intelligent internal standard matching" was not triggered.
[0101] 6. Intelligent processing of linear points: Setting thresholds for outlier points in concentration residuals. r i The threshold for the lowest calibration point is set to ±20%, and values exceeding ±20% are identified as outliers by the system. The threshold for other calibration points is set to ±15%. Based on the degree of outlier, linear points are intelligently identified and removed until all removal combinations are traversed. Under the premise of meeting the linearity standard, the best processing result under the current internal standard matching mechanism is output.
[0102] 7. Linearity Criterion Judgment: Setting the linear regression coefficients r ≥0.995, outlier threshold is set: minimum calibration point ≤±20%, other calibration points ≤±15%. Values exceeding the threshold range are identified as outliers by the system; based on the intelligent processing results of linear points under the internal standard matching mechanism, 57 Fe and 66 Since Zn does not meet the linearity criterion and the internal standard polling matching has not yet been completed, the process will re-enter the internal standard intelligent matching process.
[0103] 8. Intelligent matching of internal standards: settings 57 Fe、 66 Zn corresponding internal standard 89 Y.
[0104] 9. Multidimensional parameter correlation analysis.
[0105] Table 6 Multidimensional Parameter Correlation Analysis Table
[0106] (1) Internal standard dimension association judgment: After judgment, the current matching internal standard is within the normal fluctuation range and there is no abnormality.
[0107] (2) Accurate identification of linear outliers: Based on the current internal standard matching mechanism, 57 Fe and 66 The linear outlier in Zn is "standard curve 07", Cook distance D i It is within the normal range and is judged to be a slight deviation.
[0108] 10. Intelligent processing of linear points: Based on the degree of outliers, intelligently identify and remove linear points until all removal combinations are traversed. Under the premise of satisfying the linearity criterion, output the best processing result under the current internal standard matching mechanism: 57 Fe removes the linear point "standard curve 05". 66 Zn has no discarded points. The calculation results are as follows: Table 7 Intelligent Processing Results
[0109] 11. Linearity Criterion: Based on the intelligent processing results of linear points under the internal standard matching mechanism, 57 Fe satisfies the outlier threshold requirement, and the linear regression coefficients r ≥0.995, equals 1.0000; 66 Zn satisfies the outlier threshold requirement, and the linear regression coefficients r ≥0.995, which is 0.9996; meets the linearity standard requirements.
[0110] 12. Results and Risk Warnings: Through multidimensional parameter correlation analysis and expert interpretation of the interference knowledge graph, potential anomalies were detected and analyzed. Although the fluctuation range of the internal standard response was within the threshold range, the black-box model, by linking the interference knowledge graph with the chemical composition characteristics of the injected solution, detected... 57 The "standard curve 05" and "standard curve 07" in the Fe linear fitting may have potential spectral interference risks, i.e. 57 Fe may be present. 16 O 1 H 40 Ar、 12 C 45 Sc、 1 H 56 The diagnosis results, based on the overlapping interference of multi-atom mass spectrometry such as Fe, indicate that the signal drift of outliers is not caused by the increase in the concentration of the actual analyte, but by mass spectrometry interference caused by the complex matrix or by kinetic discrimination (KED) mode bias.
[0111] By combining intelligent linear point processing with intelligent internal standard matching, the system not only obtains intelligent data processing results but also performs precise quantitative verification of black-box diagnostic results through the calculation of standardized residuals. z iDistance from Cook D i The white-box logic confirmed that only "standard curve 05" met the statistical exclusion criteria, successfully eliminating the suspected misjudgment of "standard curve 07" by the black-box model, and achieving a dual correction of diagnostic sensitivity and accuracy.
[0112] The system ultimately performs intelligent elimination and refitting of the "Standard Curve 05" curve, while automatically generating an auxiliary diagnostic report with a transparent physical evidence chain and simultaneously outputting "Standard Curve 05 Risk Warning" to ensure the compliance and traceability of the test results.
[0113] Example 2 This embodiment discloses an intelligent data processing system for inductively coupled plasma mass spectrometry, including: The data preprocessing module is configured to: acquire the inductively coupled plasma mass spectrometry data of the element to be measured and preprocess it to obtain the preprocessed inductively coupled plasma mass spectrometry data; The multidimensional parameter correlation analysis module is configured to perform multidimensional parameter correlation analysis based on preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results. The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
[0114] Example 3 The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method of Embodiment 1.
[0115] Example 4 The purpose of this embodiment is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the method of Embodiment 1.
[0116] The steps and methods involved in the apparatuses of Embodiments 3 and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.
[0117] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.
[0118] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
[0119] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.
Claims
1. A smart data processing method for inductively coupled plasma mass spectrometry, characterized in that, include: The inductively coupled plasma mass spectrometry (ICP-MS) data of the element to be measured were acquired and preprocessed to obtain the preprocessed ICP-MS data. Multidimensional parameter correlation analysis was performed on the preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results; The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
2. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 1, characterized in that, When the internal standard dimension correlation judgment shows at least one of the following situations, a system risk warning is triggered: the internal standard recovery rate is continuously higher than the preset upper limit threshold, instrument physical drift or hardware performance deterioration is detected, abnormality in the pretreatment or injection process is detected, or memory effect or residue is detected.
3. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 1, characterized in that, The internal standard intelligent matching introduces the nearest mass number and the first ionization energy as the core matching dimensions.
4. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 3, characterized in that, The intelligent internal standard matching follows the nearest neighbor matching principle and automatically assigns priority levels to the internal standards of the elements to be tested.
5. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 1, characterized in that, When determining the correlation between internal standards, if the internal standard is not affected by mass spectrometry, the current internal standard is used to correct the signal of the element to be measured, the net response data is obtained, and linear point intelligent processing is performed on it. If a systemic risk warning is triggered, the test will be interrupted.
6. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 5, characterized in that, The linear point intelligent processing constructs an automated residual monitoring threshold. When the absolute value of the relative deviation of the residual of a data point is greater than or equal to the automated residual monitoring threshold, the point is automatically determined as a rejection point, and the rejection priority is determined according to the degree of dispersion.
7. The intelligent data processing method for inductively coupled plasma mass spectrometry as described in claim 1, characterized in that, The precise identification of linear outliers specifically involves: constructing an automated residual monitoring operator to calculate the residuals of each calibration point and the sample point to be tested in real time; when the absolute value of the standardized residual is detected to be greater than or equal to the standardized residual threshold and the Cook distance meets the set conditions, the point is automatically determined to be an abnormal discrete point and an anomaly flag is triggered.
8. An intelligent data processing system for inductively coupled plasma mass spectrometry, characterized in that, include: The data preprocessing module is configured to: acquire the inductively coupled plasma mass spectrometry data of the element to be measured and preprocess it to obtain the preprocessed inductively coupled plasma mass spectrometry data; The multidimensional parameter correlation analysis module is configured to perform multidimensional parameter correlation analysis based on preprocessed inductively coupled plasma mass spectrometry data to obtain the processing results. The multidimensional parameter correlation analysis includes internal standard dimension correlation judgment and accurate identification of linear anomalies, specifically: Based on the preprocessed inductively coupled plasma mass spectrometry data, the internal standard dimension correlation judgment is performed. If it is determined that the current internal standard is interfered with by mass spectrometry, the internal standard intelligent matching is triggered, the internal standard polling matching process is started, the internal standard matching mechanism is updated, and the multi-dimensional parameter correlation analysis under the current internal standard matching mechanism is performed simultaneously to obtain the optimal internal standard. Based on the optimal internal standard, the signal of the analyte element is corrected, and the corrected data is output. The corrected data is subjected to precise linear anomaly identification. Standardized residuals and Cook distance are introduced to remove outliers and obtain the processing results.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps in the intelligent data processing method for inductively coupled plasma mass spectrometry as described in any one of claims 1-7.
10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the intelligent data processing method for inductively coupled plasma mass spectrometry as described in any one of claims 1-7.