Patient serum miRNA expression level detection method and system

Abstract

The invention provides a patient serum miRNA expression level detection method and system, and relates to the field of biomedical detection.The method comprises the steps that on the basis of synchronously collecting multi-channel fluorescence signals, a common time-space change mode of all channels is extracted through statistical analysis, and a time-space change mode of all the channels is obtained; then systematic common mode interference signals are separated by means of adaptive filtering or curved surface fitting and the like, interference is accurately deducted for physical coordinates of each reaction channel, a real reaction process is reconstructed, then amplitude standardization and time axis nonlinear alignment are performed on a reconstruction curve, signal scales and dynamic references are unified, and the real reaction process is reconstructed. And finally extracting characteristic parameters and quantifying the expression level of the target miRNA with high precision in combination with a standard curve model. The problems of reaction kinetics heterogeneity caused by a serum matrix effect and difficulty in stripping systematic noise caused by instrument drifting are solved at the same time, and the accuracy, the reliability and the repeatability of multi-sample parallel detection are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of biomedical detection, and in particular to a method and system for detecting the expression level of miRNA in patient serum. Background Art

[0002] In the field of biomedical testing, especially in high-throughput screening and diagnostic applications, accurate and reliable quantification of the expression levels of specific biomarkers (such as miRNAs) in patient serum is of paramount clinical significance. However, existing detection technologies face multiple complex technical challenges when processing large numbers of parallel samples, which severely restrict the accuracy and reliability of test results.

[0003] Differences in serum sample composition (matrix effects) are a key factor affecting miRNA detection reaction kinetics. Small differences in the concentrations of serum components such as proteins, lipids, and ions can alter the fluorescence signal growth rate and plateau arrival time. When two serum samples contain the same initial miRNA concentration, the fluorescence signal growth curves can exhibit significant differences in slope and peak timing. Endpoint methods, which read signals at fixed time points, fail to capture dynamic signal changes, leading to underestimation of slower reaction rates and overestimation of faster reaction rates. Continuous monitoring of the complete reaction process of each serum sample and dynamic analysis of each reaction are essential for accurate miRNA quantification. Systematic common-mode interference signals can occur during long-term continuous acquisition by the detection equipment. Fluctuations in excitation light source power, detector sensitivity drift, and ambient temperature changes can simultaneously affect fluorescence measurements across all reaction channels. These systemic common-mode interference signals intermingle with the miRNA biological reaction signal, making them difficult to separate. Analysts cannot distinguish whether transient fluctuations in a single fluorescence curve are due to the miRNA reaction or instrument interference. Failure to effectively remove systematic common mode interference signals will lead to distortion of the reconstructed reaction process curve, thereby affecting the quantitative accuracy and reliability assessment of miRNA expression levels.

[0004] The present invention was developed to address the common problems in this field, such as the heterogeneity of reaction kinetics caused by serum matrix effects, the difficulty in real-time removal of systematic common mode interference signals caused by detection equipment drift, and the inability of traditional correction methods to simultaneously take into account time domain and spatial domain compensation. Summary of the Invention

[0005] The purpose of the present invention is to address the shortcomings currently existing in the field and propose a method and system for detecting patient serum miRNA expression levels, which has the advantages of being able to effectively eliminate systemic common pattern interference signals and reconstruct the true reaction process, thereby improving the accuracy and reliability of the quantification of target biomarker expression levels.

[0006] In order to overcome the deficiencies of the prior art, the present invention adopts the following technical solutions: S1, synchronously collect fluorescence signals from multiple reaction channels; S2, performing statistical processing on the fluorescence signal to obtain common pattern information; S3, performing signal processing on the common pattern information to separate the systematic common pattern interference signal; S4. Subtracting the systematic common mode interference signal from the fluorescence signal of each reaction channel to reconstruct the true reaction process of each reaction channel; S5. Quantify the expression level of target biomarkers based on the actual reaction progress.

[0007] Through the above scheme, by synchronously collecting fluorescence signals, statistically processing the signals, and separating systematic common mode interference signals, the true reaction process is reconstructed, and finally the target biomarker expression level is quantified based on the true reaction process, which effectively solves the influence of systematic common mode interference signals on the accuracy of detection results in the existing technology and improves the reliability of the quantification results.

[0008] Optionally, the present application further proposes that the specific steps of step S2 include: S21, obtaining the grouping information of the reaction channel; S22, selecting the fluorescence signal of the control group channel according to the grouping information; S23. Perform statistical processing on the fluorescence signals of the control group channel to obtain common pattern information.

[0009] Through the above scheme, by obtaining grouping information and selecting the fluorescence signal of the control group channel for statistical processing, the common pattern information can be obtained more accurately, providing a more reliable basis for the subsequent separation of interference signals.

[0010] Optionally, the present application further proposes that the specific steps of step S3 include: S31, analyzing the temporal fluctuation characteristics of common pattern information in real time; S32. Adjusting adjustment parameters for separating systematic common mode interference signals according to the time fluctuation characteristics; S33. Process the common mode information using adjustment parameters to separate the systematic common mode interference signal.

[0011] Through the above scheme, by real-time analysis of the time fluctuation characteristics of the common mode information and adjustment of the separation parameters, it is possible to dynamically adapt to the changes of the interference signal and achieve more accurate systematic common mode interference signal separation.

[0012] Optionally, the present application further proposes that the specific steps of step S4 include: S41, obtaining the physical coordinates of the reaction channel; S42, forming a data point set by combining the fluorescence signal and the physical coordinates of the corresponding reaction channel; S43, performing spatial surface fitting on the data point set to obtain the spatial distribution of the systematic common mode interference signal; S44. Determine, based on the physical coordinates of the reaction channel, a specific interference value corresponding to the reaction channel from the spatial distribution of the systematic common mode interference signal; S45. Subtract a specific interference value from the fluorescence signal of each reaction channel to reconstruct the actual reaction process of each reaction channel.

[0013] Through the above scheme, by obtaining the physical coordinates of the reaction channel and performing spatial surface fitting, the specific interference value of each channel can be determined in a refined manner, thereby more accurately deducting the interference and reconstructing a more realistic reaction process.

[0014] Optionally, the present application further proposes that the specific steps of step S5 include: S51. determining a characteristic signal level for each real reaction process; S52, performing amplitude normalization processing on the signal value of each real reaction process according to the characteristic signal level; S53, performing nonlinear adjustment on the time axis of each real reaction process so that the characteristic time point of the real reaction process is aligned with the preset reference time point; S54, extracting quantitative parameters from the actual response process after amplitude normalization and time axis nonlinear adjustment; S55. Quantify the target biomarker expression level based on the quantitative parameters.

[0015] Through the above scheme, by determining the characteristic signal level, performing amplitude normalization processing, nonlinearly adjusting the time axis, and extracting quantitative parameters, it is possible to effectively compensate for matrix effects and reaction kinetic differences, and further improve the accuracy of quantification of target biomarker expression levels.

[0016] Optionally, the present application further proposes that the specific steps of step S51 include: S511, identifying the maximum signal value in each real reaction process; S512. The maximum signal value is used as the characteristic signal level of each real reaction process.

[0017] Through the above scheme, by identifying the maximum signal value as the characteristic signal level, a simple and effective amplitude normalization benchmark is provided, which is helpful for subsequent signal processing.

[0018] Optionally, the present application further proposes that the specific steps of step S52 include: S521, determining the baseline signal level and saturation signal level of each real reaction process; S522, performing interval normalization processing on the signal value of each real reaction process according to the baseline signal level and the saturation signal level; S523. Perform nonlinear correction on the signal value after the interval normalization process to compensate for the nonlinear relationship between the signal response and the expression level of the target biomarker.

[0019] Through the above scheme, by determining the baseline and saturation signal levels to perform interval normalization and perform nonlinear correction, it is possible to more comprehensively process the signal amplitude, compensate for the nonlinear relationship, and further improve the quantization accuracy.

[0020] Optionally, the present application further proposes that the specific steps of step S53 include: S531, identifying multiple key time points in each real reaction process; S532: Determine the correspondence between multiple key time points and preset reference time points; S533: According to the corresponding relationship, the time axis of each real reaction process is segmented and stretched or compressed so that the characteristic time point of the real reaction process is aligned with the preset reference time point.

[0021] Through the above scheme, by identifying key time points and performing segmented stretching or compression, the time axes of different reaction processes can be flexibly aligned, effectively eliminating the asynchrony on the time axis and ensuring the comparability between different samples.

[0022] Optionally, the present application further proposes that the specific steps of step S531 include: S5311, performing sliding window averaging processing on the signal value of the actual reaction process to obtain a smoothed signal value; S5312. Analyze the change rate of the smoothed signal value in a continuous time period to obtain signal change rate information; S5313. Based on the signal change rate information, identify the time points when the change rate reaches a preset threshold or the change trend undergoes a significant change, as multiple key time points.

[0023] Through the above scheme, the key time points can be robustly identified through sliding window averaging and smoothing signal change rate analysis, which improves the accuracy and automation of time axis adjustment.

[0024] Optionally, the present application also proposes a patient serum miRNA expression level detection system for detecting the patient serum miRNA expression level, the system comprising a signal acquisition module, an interference signal separation module, a reaction process reconstruction module and an expression level quantification module; A signal acquisition module is used to synchronously acquire fluorescence signals from multiple reaction channels; Interference signal separation module, used to perform statistical processing on the fluorescence signal to obtain common pattern information, and perform signal processing on the common pattern information to separate the systematic common pattern interference signal; Reaction process reconstruction module, used to deduct the systematic common mode interference signal from the fluorescence signal of each reaction channel and reconstruct the true reaction process of each reaction channel; The expression level quantification module is used to quantify the expression level of target biomarkers based on the actual reaction progress.

[0025] Through the above scheme, a system for implementing the above detection method is provided. Through modular design, it can efficiently and automatically complete signal acquisition, interference separation, process reconstruction and expression level quantification, thereby improving the convenience and efficiency of detection.

[0026] From the above, it can be seen that the method and system for detecting patient serum miRNA expression levels provided in the present application effectively eliminates systematic common pattern interference signals, reconstructs the true reaction process, and accurately quantifies it on this basis. It has the advantages of being able to effectively eliminate systematic common pattern interference signals, reconstruct the true reaction process, and thus improve the accuracy and reliability of the quantification of the target biomarker expression level. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but rather the emphasis is placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0028] Figure 1 Schematic diagram of the process of the method for detecting the expression level of miRNA in patient serum of the present invention.

[0029] Figure 2 Schematic diagram of the modularization of the patient serum miRNA expression level detection system of the present invention.

[0030] Explanation of the accompanying figures: 100 - signal acquisition module; 200 - interference signal separation module; 300 - reaction process reconstruction module; 400 - expression level quantification module. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with its embodiments; it should be pointed out that the specific embodiments described herein are only used to explain the present invention and are not used to limit this case. For those skilled in the art, after reviewing the following detailed description, other systems, methods and / or features of this embodiment will become apparent. In addition, the terms describing the positional relationship in the accompanying drawings are only for illustrative purposes and cannot be understood as limiting this patent. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to the specific circumstances.

[0032] Example 1: Combined with the attached Figure 1 ,This embodiment constructs a method for detecting the expression level of patient serum miRNA.

[0033] Existing methods for detecting miRNA expression levels in patient serum face two core technical challenges when performing parallel testing on multiple serum samples: First, differences in the concentrations of components such as proteins, lipids, and ions in each serum sample (matrix effects) can alter the fluorescence signal growth rate and plateau arrival time, resulting in significantly different slopes and peak times of reaction progression curves in different channels for the same initial miRNA concentration. Second, during long-term continuous acquisition, drift in the excitation light source power and detector sensitivity of the detection instrument can be synchronously superimposed on the fluorescence signals of all channels, forming a systematic common mode interference signal. These two factors, combined, make it extremely difficult to separate the biological reaction from the collected fluorescence signal, directly affecting the accuracy and reliability of miRNA quantification results.

[0034] In a central laboratory tasked with regional disease screening, operators use an automated dispensing system to dispense hundreds of patient serum samples into individual reaction wells of microplates. The fluorescence signal generated by the enzymatic reaction within the reaction wells should be proportional to the miRNA content. However, due to serum matrix effects, the same initial miRNA concentration can exhibit slowed or accelerated growth rates, and early or delayed plateau phases in different wells. Furthermore, the output power of the solid-state laser experiences nonlinear drift with temperature fluctuations, and the detector sensitivity exhibits slight shifts with changes in hardware status. These systematic common-mode interference signals appear simultaneously across all channels, further contaminating the reaction progress curves. Faced with hundreds of mixed curves containing interfering signals, analysts are unable to determine the true source of fluctuations in individual fluorescence signals.

[0035] If the problems caused by the matrix effect and systemic common pattern interference signals mentioned above cannot be solved, the quantification results of miRNA expression levels will be systematically biased and unable to accurately reflect the true content of biomarkers, thereby affecting the reliability of early screening and diagnosis, increasing the risk of misdiagnosis or delayed treatment; at the same time, inaccurate data will also reduce the repeatability of scientific research results and hinder the discovery and verification of new markers. In response to the above difficulties, the present invention proposes a miRNA quantitative detection scheme that synchronously collects fluorescence signals, statistically extracts common pattern information, separates systemic common pattern interference signals, and reconstructs the true reaction process of each channel, thereby achieving high-precision and high-reliability quantification of patient serum miRNA expression levels.

[0036] In this regard, the present application proposes a method for detecting the expression level of miRNA in patient serum, comprising the following steps: S1, synchronously collect fluorescence signals from multiple reaction channels; S2, performing statistical processing on the fluorescence signal to obtain common pattern information; S3, performing signal processing on the common pattern information to separate the systematic common pattern interference signal; S4. Subtracting the systematic common mode interference signal from the fluorescence signal of each reaction channel to reconstruct the true reaction process of each reaction channel; S5. Quantify the expression level of target biomarkers based on the actual reaction progress.

[0037] Among them, the fluorescence signal refers to the fluorescence intensity data emitted by multiple independent reaction units obtained simultaneously at the same time point or within a very short time interval. Synchronous acquisition is achieved using equipment such as fluorescence detectors equipped with multi-channel detectors, confocal scanning systems, or high-speed camera arrays. For example, multi-channel photomultiplier tube arrays or charge-coupled device cameras are used for parallel data reading. Common pattern information is obtained using statistical methods such as principal component analysis, independent component analysis, factor analysis, or simple average or median calculations. For example, correlation analysis or cluster analysis is performed on the signals of all channels in time series. The main purpose is to identify and characterize common fluctuations or drifts that may be caused by systematic factors. Systematic common pattern interference signals are separated by adaptive filtering, Kalman filtering, wavelet denoising, or signal separation techniques based on model fitting. For example, by establishing a mathematical model of light source drift or detector response changes for fitting and subtraction, the main purpose is to accurately separate non-specific background noise or drift caused by the instrument itself or environmental factors from the observed signal. The true reaction process is reconstructed using methods such as point-by-point subtraction, curve fitting correction, or residual analysis. For example, the estimated interfering signal is subtracted from the original signal. This is primarily to eliminate external interference and restore a pure biological response that accurately reflects the concentration or activity of the target biomarker. Target biomarker quantification is achieved using methods such as standard curve analysis, kinetic model fitting, characteristic parameter extraction, or machine learning regression models. For example, by comparing the curve characteristics of a sample curve with those of a standard of known concentration, this is primarily to provide accurate and reliable biomarker concentration information for disease diagnosis or prognosis assessment.

[0038] The core innovation of the present invention lies in that by synchronously collecting fluorescence signals from multiple reaction channels and combining statistical processing of the fluorescence signals to obtain common pattern information with signal processing of the common pattern information to separate systematic common pattern interference signals, the systematic common pattern interference signals that affect all channels can be identified and stripped out, and then the systematic common pattern interference signals can be deducted from the fluorescence signals of each reaction channel to reconstruct the true reaction process of each reaction channel, thereby effectively eliminating the influence of instrument drift and environmental fluctuations on the detection results, making the subsequent quantification of the target biomarker expression level more accurate and reliable.

[0039] To more fully understand the operating mechanism of this embodiment, this embodiment achieves accurate detection of patient serum miRNA expression levels through the following logically progressive steps: First, the fluorescence signals of multiple reaction channels are synchronously collected to ensure data consistency of all channels in the time dimension, providing a unified observation basis for subsequent common pattern information extraction; Second, the synchronously collected fluorescence signals are statistically processed to extract common pattern information that is prevalent among the channels. This common pattern information reflects the overall signal fluctuation caused by detection device drift and environmental fluctuations; Then, signal processing methods such as high-pass filtering or spatial surface fitting are applied to the extracted common pattern information to separate systematic common pattern interference signals; Next, the separated systematic common pattern interference signals are subtracted point by point from the original fluorescence signal of each reaction channel to reconstruct the true reaction process of each channel; Finally, kinetic characteristics such as half-peak time and maximum reaction rate are extracted from the reconstructed true reaction process, and the miRNA content is quantified in combination with the standard curve model. This embodiment can completely eliminate systematic interference such as excitation light source power drift and detector sensitivity drift, significantly improving the accuracy and reliability of miRNA expression level quantification.

[0040] In some preferred embodiments, the present application is implemented as follows: When testing patient serum miRNA expression levels, this embodiment utilizes a fluorescence quantitative PCR instrument equipped with a CCD camera array to synchronously capture fluorescence images of all reaction wells in a microplate at predetermined time points during each reaction cycle. Image processing software then converts the captured fluorescence images into fluorescence signal time series for each reaction channel. To obtain common pattern information, this embodiment performs principal component analysis on the fluorescence signal time series for each reaction channel to extract a first principal component; this first principal component represents the most significant common signal variation trend across all channels. Subsequently, this embodiment applies a low-pass filter to this first principal component, combined with an adaptive threshold algorithm, to remove high-frequency noise and identify slowly drifting components, which are systematic common pattern interference signals. This embodiment deducts the corresponding systematic common pattern interference signal estimate from the raw fluorescence signal of each reaction channel point by point, thereby obtaining a true reaction progress curve after interference correction. Finally, this embodiment uses a sigmoid kinetic model to fit the true reaction progress curve after interference correction, extracting key kinetic characteristic parameters such as half-peak time and maximum fluorescence intensity. These extracted characteristic parameters are then converted into miRNA expression level values ​​by combining them with a pre-established standard curve model.

[0041] Through the above-mentioned technical solution, this embodiment effectively overcomes the impact of systematic common-mode interference signals introduced by factors such as instrument drift such as excitation light source power drift and detector sensitivity drift in multi-channel parallel detection on the accuracy of fluorescence measurement. This embodiment completes the reconstruction of the true reaction process of each channel by identifying the systematic common-mode interference signal and stripping it from the original fluorescence signal of each reaction channel, thereby eliminating the contamination of non-specific background signals. This embodiment significantly improves the accuracy and reliability of the quantification of patient serum miRNA expression levels, improves the signal-to-noise ratio and data credibility of the test results, and provides high-quality quantitative data support for clinical diagnosis and scientific research applications.

[0042] However, directly performing statistical processing on the fluorescence signals of all reaction channels may introduce information unrelated to the systematic common pattern interference signals, such as biological differences between different samples, thereby affecting the accuracy of the common pattern information and, in turn, affecting the subsequent separation of interference signals and the reconstruction of the true reaction process.

[0043] In this regard, the present application further proposes that the specific steps of step S2 include: S21, obtaining the grouping information of the reaction channel; S22, selecting the fluorescence signal of the control group channel according to the grouping information; S23. Perform statistical processing on the fluorescence signals of the control group channel to obtain common pattern information.

[0044] Among them, the reaction channel is realized by using a pre-configured experimental design table, identifying through sample labels, or automatically classifying based on specific marker signals in the channel, etc., the purpose of which is to provide a basis for subsequent selective processing. The fluorescence signal of the control group channel refers to the fluorescence signal data collected from the reaction channel designated as the control group. Statistical processing refers to the use of principal component analysis, independent component analysis, singular value decomposition, or calculation based on central tendency quantities such as mean, median, mode, etc., or pattern recognition based on regression analysis or machine learning models. Common pattern information refers to data extracted from the fluorescence signal of the control group channel, reflecting the systematic interference characteristics that are prevalent in all channels and change over time. The common pattern information is expressed as a time series curve, eigenvector or mathematical model parameter. Through this refined selection and processing of the data source, the present application effectively solves the problem that direct statistical processing of all reaction channels may introduce biological difference noise, ensures the accurate identification and stripping of systematic interference signals, and thus achieves accurate quantification of miRNA expression levels.

[0045] In some preferred embodiments, the present application is implemented as follows: When testing patient serum miRNA expression levels, a microplate layout can be pre-designed, with some wells designated as controls. For example, in a 96-well plate, the first and last columns of the wells can be reserved as negative or positive controls. Before the experiment begins, the system can read the reaction channel grouping information from a pre-set experimental configuration file. This information clearly identifies which channels belong to the control group and which to the experimental group. For example, the configuration file can be a table containing channel IDs and corresponding grouping types (e.g., control, experimental). Based on the read grouping information, the system can automatically filter and select the fluorescence signals of all reaction channels marked as "control." For example, if channels with IDs 1 to 8 and 89 to 96 are designated as controls, only the fluorescence signal data collected synchronously from these channels throughout the entire reaction process is extracted. The fluorescence signals of these selected control channels are then statistically processed to obtain common pattern information. Specifically, principal component analysis can be performed on the fluorescence signal data of these control channels. The fluorescence signal time series data for all control group channels are constructed into a data matrix, where each row represents a time point and each column represents the fluorescence signal for a control group channel. A principal component analysis algorithm is then performed on this data matrix to extract the first principal component, which typically represents the most dominant common variation trend in the dataset, i.e., the systemic common pattern interference signal. The time series data of this extracted principal component is the common pattern information obtained by this protocol, reflecting the drift or fluctuation generated by the system itself in the absence of a specific response to the target biomarker.

[0046] Through the above technical solution, non-systematic noise introduced by biological differences in the experimental groups (such as matrix effects of serum samples from different patients) was effectively eliminated, so that the common pattern information obtained can more purely and accurately reflect the systematic interference of the detection system itself, thereby improving the overall reliability of the detection of patient serum miRNA expression levels.

[0047] Because the detection equipment itself exhibits systematic drift, this drift acts as a common-mode interference signal that is superimposed on the true signals of all parallel channels. This results in the common-mode information containing both true reaction information and the systematic common-mode interference signal. Directly subtracting the common-mode information containing the systematic common-mode interference signal from the fluorescence signal will prevent accurate reconstruction of the true reaction progress in each reaction channel. Therefore, separating the systematic common-mode interference signal from the common-mode information is crucial for accurately reconstructing the true reaction progress.

[0048] In this regard, the present application further proposes step S3, wherein the specific steps of performing signal processing on the common pattern information and separating the systematic common pattern interference signal include: S31, analyzing the temporal fluctuation characteristics of common pattern information in real time; S32. Adjusting adjustment parameters for separating systematic common mode interference signals according to the time fluctuation characteristics; S33. Process the common mode information using adjustment parameters to separate the systematic common mode interference signal.

[0049] Temporal fluctuation characteristics refer to signal trends, such as slowly rising or falling drift, periodic variations such as periodic noise caused by power supply frequency or mechanical vibration, or transient jumps or random noise. Adjustment parameters include the filter's cutoff frequency, bandwidth, or order, the learning rate or step size of the adaptive filtering algorithm, or coefficients based on model fitting. The purpose of adjusting parameters is to dynamically optimize the performance of the separation algorithm based on the real-time detected temporal fluctuation characteristics, enabling it to more accurately match and remove the interference signal at the current moment, thereby improving the accuracy and adaptability of the separation. Systematic common mode interference is addressed through filtering, such as high-pass filtering to remove slow drift, notch filtering to remove periodic noise, adaptive noise cancellation, signal decomposition, such as independent component analysis (ICA) and empirical mode decomposition (EMD), or model fitting, such as polynomial fitting and exponential fitting.

[0050] The solution of this application achieves accurate separation of systematic common mode interference signals by performing refined signal processing on common mode information. This dynamic and adaptive separation strategy significantly improves the ability to extract pure interference signals from common mode information, laying a solid foundation for subsequently deducting the interference signal from the fluorescence signal of each reaction channel and then reconstructing the true reaction process. Compared with the method of using fixed parameters to separate interference signals, this solution can better deal with the complex and dynamically changing systematic interference that may occur during long-term operation of the detection equipment, thereby ensuring the accuracy and reliability of the final reconstructed reaction process.

[0051] In some preferred embodiments, S33, using the adjustment parameter to process the common mode information to separate the systematic common mode interference signal is specifically implemented as follows: First, real-time spectrum analysis (such as sliding window Fourier transform or wavelet transform) is performed on the common pattern information, and the mean, variance or slope is calculated within a short time window to automatically identify major interference features such as low-frequency drift and periodic noise.

[0052] Secondly, the filter parameters are dynamically adjusted based on the results of spectrum analysis and time domain statistics. If low-frequency drift is predominant, an appropriate high-pass cutoff frequency is set. If periodic noise occurs, the notch center frequency and bandwidth are configured. For complex or nonlinear drift, Kalman filtering or LMS adaptive filtering can be enabled and the gain updated in real time.

[0053] Finally, the adaptive filter is applied directly to the common mode information, and the output processing result is a systematic common mode interference signal that accurately reflects the instrument drift and environmental noise, which is then subtracted point by point in subsequent steps to reconstruct the true reaction process of each channel.

[0054] The above-mentioned technical solution solves the problem that the fixed parameter separation method cannot accurately remove interference due to the dynamics and complexity of the detection equipment's own drift, and obtains a purer systematic common mode interference signal. This provides accurate input for the subsequent subtraction of the interference signal from the fluorescence signal of each reaction channel, and then reconstructs the true reaction process of each reaction channel, significantly improving the accuracy and reliability of the detection results.

[0055] To address the shortcoming that using only the average value to deduct common interference will ignore spatial heterogeneity, the present application further proposes S4, which involves deducting the systematic common pattern interference signal from the fluorescence signal of each reaction channel to reconstruct the true reaction process of each reaction channel, including the following steps: S41, obtaining the physical coordinates of the reaction channel; S42, forming a data point set by combining the fluorescence signal and the physical coordinates of the corresponding reaction channel; S43, performing spatial surface fitting on the data point set to obtain the spatial distribution of the systematic common mode interference signal; S44. Determine, based on the physical coordinates of the reaction channel, a specific interference value corresponding to the reaction channel from the spatial distribution of the systematic common mode interference signal; S45. Subtract a specific interference value from the fluorescence signal of each reaction channel to reconstruct the actual reaction process of each reaction channel.

[0056] Among them, spatial surface fitting refers to the use of mathematical methods to construct a continuous mathematical surface model that can describe the trend of a set of discrete three-dimensional data points (for example, X, Y coordinates and corresponding Z values). It can be achieved using techniques such as least squares, spline interpolation, polynomial fitting, or radial basis function (RBF) interpolation. The purpose is to expand the limited discrete measurement values ​​to the entire spatial area, thereby estimating the signal value or interference value at any location.

[0057] The solution of this application achieves precise positioning and subtraction of systematic common-mode interference signals by introducing the physical coordinate information of the reaction channels and utilizing spatial surface fitting technology. Specifically, first, the physical coordinates of each channel are obtained, and the common-mode interference values ​​of each channel at the same time point are paired with the coordinates to form a set of spatial-temporal data points. Second, a surface fitting (such as a quadratic polynomial or kriging interpolation) is performed on this point set on a two-dimensional plane to construct a mathematical model f(X,Y) that reflects the spatial distribution of systematic interference within the entire detection area. Finally, the coordinates (Xi,Yi) of each channel are substituted into f(X,Y) to obtain its specific interference value f(Xi,Yi), which is then subtracted point by point from the original fluorescence signal to reconstruct a more accurate true reaction process. This approach not only takes into account spatial heterogeneity but also ensures the integrity and feasibility of the solution.

[0058] By acquiring the physical coordinates of the reaction channels and constructing a spatial distribution model of the interference signal using spatial surface fitting techniques, this technical solution allows for the determination of a specific interference value for each reaction channel, taking into account the influence of spatial location. This refined interference subtraction method effectively avoids the residual errors that can result from simple global subtraction, significantly improving the accuracy of reconstructing the true reaction progress of each reaction channel from the fluorescence signal, and thus providing a more reliable data foundation for subsequent biomarker quantification.

[0059] In some of the aforementioned embodiments of this application, only the true reaction progress is reconstructed. If the target biomarker expression level is quantified directly based on this uncorrected reaction progress, the true reaction progress will vary in signal amplitude and time due to matrix effects in different samples, which will introduce bias and reduce the accuracy and reliability of the detection. In response to this, this application further proposes S5, specific steps for quantifying the target biomarker expression level based on the true reaction progress: S51. determining a characteristic signal level for each real reaction process; S52, performing amplitude normalization processing on the signal value of each real reaction process according to the characteristic signal level; S53, performing nonlinear adjustment on the time axis of each real reaction process so that the characteristic time point of the real reaction process is aligned with the preset reference time point; S54, extracting quantitative parameters from the actual response process after amplitude normalization and time axis nonlinear adjustment; S55. Quantify the target biomarker expression level based on the quantitative parameters.

[0060] The characteristic signal level refers to the representative signal intensity value during a reaction process. It is determined by identifying the maximum signal value, average signal value, or signal value at a specific time point during the reaction process. Amplitude normalization refers to adjusting the signal amplitudes of different reaction processes to a uniform scale or range. This is achieved through linear scaling, Z-score normalization, or normalization based on a specific reference point. Time axis nonlinearity adjustment refers to correcting for temporal differences between different reaction processes through nonlinear transformations, such as reaction rate or offset of the reaction starting point, to ensure temporal consistency of key reaction events. Time axis nonlinearity adjustment is achieved through dynamic time warping (DTW) algorithms, piecewise linear interpolation, or stretching or compression algorithms based on key time point matching. Quantitative parameters are numerical indicators extracted from the corrected reaction process curve that directly or indirectly reflect the expression level of the target biomarker. Quantitative parameters include the initial reaction rate, maximum reaction rate, time required to reach a specific signal level, area under the curve, or endpoint signal value.

[0061] In some preferred embodiments, the present application is implemented as follows: When determining the characteristic signal level of each true reaction process, the maximum signal value in each true reaction process can be identified and used as the characteristic signal level of each true reaction process. For example, for a typical S-shaped reaction curve, the maximum signal value generally represents the signal intensity when the reaction reaches saturation or near saturation, and can be used as a representative indicator of the signal amplitude of the reaction process. When performing amplitude normalization on the signal values ​​of each true reaction process, the baseline signal level and saturation signal level of each true reaction process can be first determined. The baseline signal level can be calculated by averaging the signal value before the reaction starts or in the early stage of the reaction, while the saturation signal level can be calculated by averaging the signal value in the late stage of the reaction or the maximum value. Then, based on these baseline signal levels and saturation signal levels, the signal values ​​of each true reaction process are interval-normalized, for example, by linearly mapping the signal values ​​to a range of 0 to 1. In addition, to compensate for the nonlinear relationship between the signal response and the expression level of the target biomarker, the signal values ​​after interval normalization can be subjected to nonlinear correction, for example, by applying a logarithmic transformation, a power function transformation, or a calibration curve-based lookup table correction. When performing nonlinear adjustment on the time axis of each real reaction process, multiple key time points can be identified for each real reaction process. These key time points can include the reaction start point, half-maximal signal point, maximum reaction rate point, or reaction end point. Specifically, a sliding window average process can be performed on the signal values ​​of the real reaction process to obtain smoothed signal values ​​to reduce noise interference. Next, the rate of change of the smoothed signal values ​​over consecutive time periods is analyzed to obtain signal change rate information. Based on the signal change rate information, the time points at which the change rate reaches a preset threshold or a significant change in the change trend are identified as multiple key time points. For example, the time point at which the signal first exceeds the baseline threshold can be identified as the start point, and the time point at which the signal change rate reaches a peak value can be identified as the maximum reaction rate point. Then, a correspondence is determined between these key time points and preset reference time points. The preset reference time point can be the average key time points extracted from the reaction process of a standard sample or a sample of known concentration. Finally, based on this correspondence, the time axis of each real reaction process is segmented and stretched or compressed, for example, using linear interpolation or spline interpolation, so that the characteristic time points of the real reaction process are aligned with the preset reference time points. When extracting quantitative parameters from the actual reaction process after amplitude normalization and timeline nonlinear adjustment, one can extract, for example, the corrected maximum signal value, the corrected reaction rate (e.g., the slope within a specific signal interval), or the corrected area under the curve. Ultimately, when quantifying the expression level of a target biomarker based on the quantitative parameters, a pre-established calibration curve can be used to convert the extracted quantitative parameters into the corresponding biomarker concentration or expression level.For example, a regression model between a quantitative parameter and a known miRNA concentration can be established, and then the quantitative parameter of an unknown sample can be substituted into the model for calculation.

[0062] Through the above technical solution, the present application can effectively eliminate the differences in signal amplitude and time dimensions of the actual reaction process caused by the matrix effect of different samples: through amplitude normalization processing, the signal intensity scale of different samples can be unified, avoiding the quantification bias introduced by signal amplitude differences; through nonlinear adjustment of the time axis, the problem of inconsistent reaction kinetic rates of different samples can be corrected, ensuring comparison under a unified time reference. This enables the quantitative parameters extracted from the processed actual reaction process to more accurately reflect the actual expression level of the target biomarker, thereby significantly improving the accuracy and reliability of the quantification of the target biomarker expression level, overcoming the bias problem caused by quantification directly based on uncorrected reaction processes.

[0063] In some of the aforementioned embodiments of this application, the problem of accurately and efficiently determining the characteristic signal level to avoid deviations in the characteristic signal level due to signal fluctuations or noise interference, which in turn affects the subsequent amplitude normalization processing and the quantification accuracy of the target biomarker expression level, is a problem that needs to be solved. To this end, this application further proposes S51, the specific steps of determining the characteristic signal level of each actual reaction process include: S511, identifying the maximum signal value in each real reaction process; S512. The maximum signal value is used as the characteristic signal level of each real reaction process.

[0064] Among them, the maximum signal value is achieved by traversing all signal data points and selecting the maximum value, and can also be identified by a peak detection algorithm. The purpose is to capture the peak response of the reaction process as a direct indicator of the reaction intensity; the characteristic signal level is determined by selecting specific points in the reaction process (such as the maximum value, the average value of the plateau period, or a specific percentage point). The purpose is to provide a unified benchmark for subsequent signal standardization processing.

[0065] The above scheme provides a direct method for determining characteristic signal levels. In the method for detecting patient serum miRNA expression levels, after reconstructing the true reaction process for each reaction channel, these processes need to be quantified to obtain the target biomarker expression level. Determining the characteristic signal level is a key step, directly impacting the accuracy of the subsequent amplitude normalization process. By using the maximum signal value as the characteristic signal level, this scheme avoids complex curve fitting or multi-point averaging calculations, simplifying the processing flow and reducing computational resource consumption. Furthermore, the maximum signal value typically represents the peak value at which the signal intensity reaches or nearsaturation during the reaction process. This value is relatively stable and has a high signal-to-noise ratio, effectively resisting the influence of random noise and signal fluctuations. Therefore, using the maximum signal value as the characteristic signal level provides a reliable and representative benchmark for subsequent amplitude normalization, allowing the signal value after amplitude normalization to more accurately reflect the true concentration of the target biomarker, thereby improving the accuracy of extracting quantitative parameters from the processed true reaction process and ultimately enhancing the precision and reliability of quantifying the target biomarker expression level. This method not only ensures efficiency but also improves the robustness of data processing, ensuring accurate quantification of miRNA expression levels in complex biological sample detection.

[0066] In some preferred embodiments, determining the characteristic signal level of each real reaction process can be specifically implemented as follows: Assume that the data of a real reaction process is recorded as a series of fluorescence signal values ​​at time points, for example, time points t1, t2, ..., tn correspond to signal values ​​V1, V2, ..., Vn, respectively. In order to identify the maximum signal value in each real reaction process, the data processing unit can traverse these signal values, from V1 to Vn, compare them one by one, and record the signal value with the largest value. For example, if the signal value sequence of a reaction process is [100, 120, 150, 180, 170, 160], then the maximum signal value is 180. Once this maximum signal value is identified, the data processing unit directly uses this value as the characteristic signal level of the real reaction process. For example, in the above example, 180 is determined to be the characteristic signal level of the reaction process. This characteristic signal level can then be used for subsequent amplitude normalization processing, for example, dividing all signal values ​​by 180, or performing other forms of normalization to eliminate amplitude deviations between different reaction processes due to differences in initial signal intensities, thereby making the reaction processes of different samples comparable in amplitude.

[0067] By the technical solution, the maximum signal value in each real reaction process is directly identified and taken as the characteristic signal level, which can effectively avoid the deviation of the characteristic signal level caused by signal fluctuation or noise interference, thereby improving the accuracy of the characteristic signal level, effectively simplifying the determination process of the characteristic signal level, reducing the calculation complexity and time cost, and further improving the reliability of the subsequent amplitude normalization processing, and finally helping to improve the quantization accuracy of the target biomarker expression level.

[0068] In some embodiments of the present application, the signal value of the real reaction process is subjected to amplitude normalization processing, which can be specifically performed by identifying the maximum signal value in each real reaction process and taking it as the characteristic signal level for signal value normalization. In this way, the quantization deviation caused by the difference in signal intensity between different reaction processes can be eliminated. However, in the implementation process, only the maximum signal value is used for normalization, which may not be sufficient to solve the problem of inconsistent amplitude range caused by signal baseline drift and saturation effect. In addition, the non-linear relationship between signal response and target biomarker expression level is not effectively compensated, which affects the accuracy of subsequent quantization parameters. To this end, the present application further proposes that the specific steps of step S52 include: S521, determining the baseline signal level and the saturation signal level of each real reaction process; S522, performing interval normalization processing on the signal value of each real reaction process according to the baseline signal level and the saturation signal level; S523, performing non-linear correction on the signal value after interval normalization processing to compensate for the non-linear relationship between signal response and target biomarker expression level.

[0069] The baseline signal level is achieved by averaging, fitting or statistical analysis of the signal value at the initial stage of the reaction process curve. The saturation signal level is achieved by averaging, fitting or identifying the maximum value of the signal value at the end of the reaction process curve or the signal plateau period. Interval normalization processing refers to linearly mapping the original signal value to a predetermined specific numerical interval, for example, 0 to 1, which aims to eliminate the signal amplitude difference caused by baseline drift and saturation effect in different reaction processes, so that the signal is comparable. Non-linear correction refers to adjusting the signal value to establish a more accurate mapping relationship between the signal value and the target biomarker expression level. Non-linear correction is achieved by using a pre-set correction curve, a lookup table, a polynomial fitting or a model based on machine learning.

[0070] The baseline signal and the saturation signal of each real reaction process are first identified, and interval normalization is performed on the signals with the baseline signal and the saturation signal as the upper and lower limits to eliminate amplitude differences; then, based on the nonlinear relationship between the signals and the miRNA concentration, nonlinear correction is performed on the normalized data to compensate for the distortion of the response curve. The quantitative parameters extracted after this processing can accurately reflect the target miRNA expression level, greatly improving the accuracy and reliability of the detection.

[0071] In some preferred embodiments, the amplitude standardization processing of the signal values of the real reaction processes can be implemented as follows: first, the baseline signal level and the saturation signal level of each real reaction process are determined. For example, the baseline signal level can be obtained by averaging the signal values at the first 10 time points of the reaction process curve; and the saturation signal level can be obtained by averaging the signal values at the last 20 time points of the reaction process curve, or by identifying the maximum stable signal value after the curve reaches the plateau. Then, according to the determined baseline signal level and saturation signal level, the signal values of each real reaction process are interval normalized to map all signal values to the range of 0 to 1, where the baseline signal level corresponds to 0 and the saturation signal level corresponds to 1. Subsequently, the interval normalized signal values are nonlinearly corrected. For example, a correction lookup table can be pre-established, which records the correction factors of the real biomarker expression levels corresponding to different normalized signal values, or a pre-trained polynomial model (such as a quadratic or cubic polynomial) is used to fit and correct the normalized signal values to compensate for the nonlinear relationship between the signal response and the target biomarker expression level. In this way, even in the case of nonlinear signal response, the final quantification result can accurately reflect the real concentration of the biomarker.

[0072] Through the above technical solutions, the present application can effectively solve the problem of inconsistent amplitude range caused by baseline drift and saturation effect in the real reaction process, as well as the nonlinear relationship between the signal response and the target biomarker expression level, thereby significantly improving the accuracy of extracting quantitative parameters from the real reaction process, and further improving the quantification precision and reliability of the target biomarker expression level.

[0073] In some embodiments of the present application, the time axis of each real reaction process is nonlinearly adjusted so that the characteristic time points of the real reaction process are aligned with the preset reference time points. The nonlinear adjustment of the time axis can be achieved by simple linear interpolation or global stretching to attempt alignment. This can compensate for the time offset to some extent. However, in the implementation process, due to the complex nonlinear differences in the reaction rates of different reaction channels, simple linear adjustment cannot accurately capture and correct the offset of all time periods, resulting in local misalignment of the reaction process on the time axis. If the quantitative parameters are directly extracted, errors will be introduced due to the misalignment in time, affecting the accuracy of the final quantitative results. To this end, the present application further proposes that the specific steps of nonlinearly adjusting the time axis of each real reaction process so that the characteristic time points of the real reaction process are aligned with the preset reference time points include: S531, identifying a plurality of key time points on each real reaction process; S532, determining the correspondence between the plurality of key time points and the preset reference time points; S533, according to the correspondence, performing segmented stretching or compression processing on the time axis of each real reaction process, so that the characteristic time points of the real reaction process are aligned with the preset reference time points.

[0074] The key time points refer to time points with specific biological or kinetic significance in the reaction process, such as the starting point of the reaction, the turning point of rapid signal growth, the maximum signal value, the peak point of the plateau period, and the specific point of signal decay. The key time points can be identified using signal rate analysis, curve fitting, or pattern recognition techniques. The preset reference time points are calibrated and set according to known standards or historical data. The segmented stretching or compression processing refers to using spline interpolation, dynamic time warping (DTW), or local linear transformation algorithms to ensure that the characteristic time points in each time period are accurately aligned with the corresponding reference time points.

[0075] In some preferred embodiments, specifically, when performing segmented stretching or compression processing on the time axis of each real reaction process, first, the inflection point detection algorithm based on second derivative or curvature analysis can be used to identify key time nodes such as reaction starting point, maximum reaction rate point, and signal saturation point; second, the optimal matching path between the real process and the standard process is calculated by means of the dynamic time warping algorithm with these key nodes as anchor points, so as to determine the correspondence of each key time node; finally, according to the matching result, the time axis of the real process is stretched or compressed in each node segment by cubic spline or linear interpolation, and the time sequence is kept consistent, so that all key time nodes are accurately aligned with the reference time points.

[0076] Through the above technical solution, the present application can effectively solve the problem of misalignment of reaction processes on the time axis due to differences in reaction rates of different reaction channels. This enables the subsequent extraction of quantitative parameters from the calibrated reaction process to eliminate errors introduced by time offset, thereby significantly improving the accuracy of the quantitative results and providing reliable guarantees for the accurate evaluation of the expression levels of target biomarkers.

[0077] In some of the aforementioned embodiments of this application, the actual reaction process may be subject to various noise interferences, resulting in signal fluctuations, making it difficult to accurately identify key time points, thereby affecting the accuracy of subsequent nonlinear adjustments to the time axis. To address this issue, this application further proposes that the step of identifying multiple key time points in each actual reaction process (S531) includes: S5311, performing sliding window averaging processing on the signal value of the actual reaction process to obtain a smoothed signal value; S5312. Analyze the change rate of the smoothed signal value in a continuous time period to obtain signal change rate information; S5313. Based on the signal change rate information, identify the time points when the change rate reaches a preset threshold or the change trend undergoes a significant change, as multiple key time points.

[0078] Sliding window averaging refers to sliding a fixed-size window and taking the mean within the window to smooth random noise; the signal change rate is the rate of increase or decrease of the signal over time, which is used to quantify dynamics and capture inflection points; the preset threshold is a reference value for determining whether the rate of change is significant, which is used to eliminate small fluctuations; a significant trend turning point is the moment when the size or sign of the rate of change shows a significant jump (such as from rising to falling or from slow to fast).

[0079] In some preferred embodiments, when the signal value of the real reaction process is subjected to sliding window averaging, a window with a fixed length of 5 to 15 data points can be used for averaging. For example, for a signal collected once per second, data points of 2 seconds before and after can be taken for averaging to eliminate short-term noise. When analyzing the rate of change of the smoothed signal value in a continuous time period, the signal rate of change information can be calculated by using a difference method or a least squares method to fit the local slope. For example, the difference between two adjacent smoothed signal values ​​is used as the instantaneous rate of change. When identifying key time points according to the signal rate of change information, an absolute value threshold of the rate of change can be set. For example, when the absolute value of the rate of change exceeds 0.05 signal units / second, or when the rate of change signs of three consecutive data points change, they are marked as key time points. These key time points can include the starting point of the reaction, the half-maximum signal point, the maximum signal point, and the reaction end point, etc., providing accurate anchor points for subsequent time axis nonlinear adjustment.

[0080] Through the above technical solution, the present application can effectively overcome the problem of signal fluctuations caused by noise interference in the actual reaction process and achieve accurate identification of key time points. By smoothing the signal, the influence of noise on the judgment of signal change trend is eliminated; by analyzing the signal change rate, the dynamic characteristics of the signal are quantified; by setting thresholds and identifying trend changes, important turning points in the reaction process are accurately locked. This enables subsequent nonlinear adjustments to the time axis to be based on more reliable key time points, thereby improving the accuracy of time axis alignment and ultimately improving the accuracy of quantification of target biomarker expression levels.

[0081] A specific example of this application is as follows: To verify this application, this embodiment uses the data from Table 1 (Multi-disease Performance Verification), Table 2 (Multi-channel Parallel Efficiency), and Table 3 (Postoperative Dynamic Follow-up) in combination under the same processing flow to provide comprehensive examples from the three dimensions of accuracy / stability, efficiency / throughput, and clinical consistency.

[0082] 1. Accuracy and stability (Table 1) The samples were extracted using a commercially available kit using a fluorescent real-time quantitative PCR instrument. The samples were then tested for the presence of iRNA124 and miRNA210 using the RNA fluorescent real-time quantitative PCR method, and their Ct values ​​were determined.

[0083] Table 1: Performance Validation of Multi-Disease Serum MiRNA Detection in the "Multi-Disease Serum MiRNA Detection Performance Validation" dataset (n=1551, 41 subgroups). After common pattern separation and spatial subtraction in steps S3–S4, the mean Ct standard deviation decreased from 4.26 to 0.51, an average decrease of 87.97%. In representative subgroups with high matrix complexity, the "lung cancer patient group" decreased from 6.2 to 0.6 (a decrease of 90.32%), and the "liver cirrhosis group" decreased from 5.5 to 0.7 (a decrease of 87.27%). After amplitude interval normalization, nonlinear correction, and time axis alignment in S5, the intra-assay CV mean was 2.92%, and the inter-assay CV mean was 5.40%. The results were expressed as ΔCt. The estimated central mean accuracy was approximately 0.55, and the mean concordance with the clinical gold standard was 95.96%. These results demonstrate that, even in high-throughput settings with significant matrix effects and instrument drift, this method can significantly reduce channel / well-site variations and maintain cross-batch reproducibility (see Table 1).

[0084] Table 1: Performance verification of serum miRNA detection for multiple diseases Experiment ID Sample type Target miRNA Number of samples <![CDATA[基质复杂度评分 † ]]> Ct value fluctuation range before treatment Fluctuation range of Ct value after treatment <![CDATA[检测准确性(ΔCt) ‡ ]]> Intra-assay CV% Inter-assay CV% Compliance rate with clinical gold standard 1 Healthy control group hsa-miR-16-5p 30 1.2±0.3 24.8±3.5 25.1±0.4 0.3±0.2 2.10% 4.80% 98.30% 2 Lung cancer patient group hsa-miR-21-5p 45 3.8±0.9 (hyperlipidemia) 19.5±6.2 20.7±0.6 1.2±0.3 (↓82% fluctuation) 3.50% 6.00% 94.70% 3 Breast cancer patient group hsa-miR-155-5p 38 2.5±0.7 22.3±4.1 23.0±0.5 0.7±0.2 2.80% 5.20% 96.20% 4 Colorectal cancer group hsa-miR-92a-3p 42 3.1±0.6 21.5±4.8 22.0±0.5 0.5±0.2 3.10% 5.50% 95.50% 5 Prostate cancer group hsa-miR-141-3p 35 2.8±0.5 23.2±3.9 23.6±0.4 0.4±0.1 2.60% 4.90% 96.80% 6 Diabetes group hsa-miR-126-3p 50 2.0±0.4 25.1±2.8 25.3±0.3 0.2±0.1 1.80% 4.20% 97.80% 7 Cirrhosis group hsa-miR-122-5p 40 4.2±0.8 18.9±5.5 19.8±0.7 0.9±0.3 3.80% 6.50% 94.20% 8 Alzheimer's group hsa-miR-132-3p 28 1.8±0.3 26.3±3.2 26.5±0.3 0.2±0.1 1.90% 4.40% 98.10% 9 Heart failure group hsa-miR-499a-5p 36 3.5±0.7 20.1±4.5 20.8±0.6 0.7±0.2 3.30% 5.80% 95.10% 10 Rheumatoid arthritis group hsa-miR-155-5p 33 2.9±0.6 22.8±3.7 23.3±0.5 0.5±0.2 2.70% 5.10% 96.30% 11 Pancreatic cancer group hsa-miR-196a-5p 30 4.0±0.9 19.8±5.8 20.6±0.7 0.8±0.3 3.60% 6.20% 94.50% 12 Thyroid cancer group hsa-miR-222-3p 32 2.7±0.5 23.5±3.6 23.9±0.4 0.4±0.1 2.50% 4.80% 97.00% 13 Stroke group hsa-miR-124-3p 38 3.2±0.6 21.2±4.3 21.7±0.5 0.5±0.2 2.90% 5.30% 95.70% 14 Gastric cancer group hsa-miR-21-5p 41 3.6±0.7 20.3±4.7 21.0±0.6 0.7±0.2 3.40% 5.90% 94.90% 15 Kidney cancer group hsa-miR-210-3p 34 2.6±0.5 23.8±3.8 24.2±0.4 0.4±0.1 2.40% 4.70% 97.20% 16 Ovarian cancer group hsa-miR-200c-3p 37 3.3±0.6 21.0±4.4 21.5±0.5 0.5±0.2 3.00% 5.40% 95.90% 17 Leukemia group hsa-miR-191-5p 39 3.7±0.7 20.0±4.6 20.7±0.6 0.7±0.2 3.50% 6.00% 94.80% 18 Asthma group hsa-miR-146a-5p 44 2.4±0.5 24.2±3.3 24.5±0.4 0.3±0.1 2.20% 4.60% 97.50% 19 Multiple sclerosis group hsa-miR-let-7d-5p 29 2.1±0.4 25.5±3.0 25.7±0.3 0.2±0.1 2.00% 4.50% 98.00% 20 Hepatitis group hsa-miR-122-5p 46 4.1±0.8 19.2±5.4 20.0±0.7 0.8±0.3 3.70% 6.40% 94.30% 21 Osteoporosis group hsa-miR-21-5p 31 2.3±0.5 24.5±3.4 24.8±0.4 0.3±0.1 2.30% 4.70% 97.70% 22 Psoriasis group hsa-miR-31-5p 43 3.0±0.6 22.0±4.0 22.5±0.5 0.5±0.2 2.80% 5.20% 96.10% 23 Pancreatitis group hsa-miR-216a-5p 47 3.9±0.8 19.6±5.2 20.4±0.7 0.8±0.3 3.60% 6.30% 94.40% 24 Brain tumor group hsa-miR-10b-5p 33 3.4±0.7 20.8±4.5 21.4±0.6 0.6±0.2 3.20% 5.70% 95.30% 25 Lupus erythematosus group hsa-miR-let-7a-5p 40 2.8±0.5 23.0±3.7 23.4±0.4 0.4±0.1 2.60% 5.00% 96.90% 26 Coronary heart disease group hsa-miR-1-3p 48 3.1±0.6 21.8±4.2 22.3±0.5 0.5±0.2 2.90% 5.40% 95.80% 27 Tuberculosis Group hsa-miR-29a-3p 45 3.8±0.7 20.2±4.8 20.9±0.6 0.7±0.2 3.40% 5.90% 95.00% 28 Bladder cancer group hsa-miR-96-5p 36 2.9±0.6 22.5±3.9 23.0±0.5 0.5±0.2 2.70% 5.10% 96.40% 29 Nasopharyngeal carcinoma group hsa-miR-17-5p 39 3.5±0.7 20.5±4.6 21.2±0.6 0.7±0.2 3.30% 5.80% 95.20% 30 Gallbladder cancer group hsa-miR-182-5p 34 3.2±0.6 21.4±4.3 21.9±0.5 0.5±0.2 3.00% 5.50% 95.60% 31 Esophageal cancer group hsa-miR-25-3p 41 3.7±0.7 20.1±4.7 20.8±0.6 0.7±0.2 3.50% 6.00% 94.70% 32 Melanoma group hsa-miR-221-3p 37 3.0±0.6 22.2±4.0 22.7±0.5 0.5±0.2 2.80% 5.20% 96.20% 33 Laryngeal cancer group hsa-miR-205-5p 35 3.3±0.6 21.1±4.4 21.6±0.5 0.5±0.2 3.10% 5.60% 95.50% 34 Myeloma Group hsa-miR-15a-5p 38 3.6±0.7 20.4±4.5 21.1±0.6 0.7±0.2 3.40% 5.90% 95.10% 35 Lymphoma group hsa-miR-155-5p 42 3.4±0.7 20.7±4.6 21.3±0.6 0.6±0.2 3.20% 5.70% 95.40% 36 Sarcoma group hsa-miR-9-5p 40 3.8±0.8 19.9±5.0 20.6±0.7 0.7±0.3 3.60% 6.20% 94.60% 37 Neuroblastoma Group hsa-miR-34a-5p 33 3.1±0.6 21.6±4.2 22.1±0.5 0.5±0.2 2.90% 5.40% 95.70% 38 Retinoblastoma Group hsa-miR-183-5p 30 2.7±0.5 23.6±3.6 24.0±0.4 0.4±0.1 2.50% 4.90% 97.10% 39 Testicular cancer group hsa-miR-371a-3p 32 2.8±0.5 23.3±3.7 23.7±0.4 0.4±0.1 2.60% 5.00% 96.80% 40 Endometrial cancer group hsa-miR-200a-3p 36 3.0±0.6 22.4±3.9 22.9±0.5 0.5±0.2 2.70% 5.10% 96.30% 41 Cervical cancer group hsa-miR-21-5p 44 3.2±0.6 21.3±4.3 21.8±0.5 0.5±0.2 3.00% 5.50% 95.60% Note: Matrix complexity score: 1-5 (5 = highest interference); ΔCt = |Ct after treatment - Ct of standard reference| Core conclusions: The system significantly compressed the fluctuation range of Ct values ​​(fluctuation in the lung cancer group ↓82%) and maintained intra-batch CV <4% under high matrix interference, demonstrating the effectiveness of eliminating matrix effects and suppressing instrument drift.

[0085] 2. Parallel Efficiency and Scalability (Table 2) In Table 2: Analysis of Multi-channel Parallel Detection Efficiency (18 conditions, 8–144 reaction channels, 1–12 Hz sampling frequency), the overall mean consistency (§) of the reconstruction process was 0.901 (range 0.82–0.98), the mean interference signal separation time was 153.1 ms, and the mean single-sample processing time was 11.2 s. Compared with the traditional process, the average error reduction was 64.8% (↓45%–↓85%). Under typical high-throughput conditions of 96 channels / 5 Hz, the consistency was 0.92±0.03, the separation time was 120±20 ms, the single-sample processing time was 8.5 s, and the error of the traditional method was ↓63%. These results demonstrate that the implementation of DSP / FPGA real-time separation and host-based reconstruction can improve throughput while maintaining high consistency and low latency (see Table 2).

[0086] Table 2: Multi-channel parallel detection efficiency analysis

[0087] Description: Reconstruction process consistency: correlation coefficient of kinetic curves between channels (1 = perfect consistency) Core conclusion: 96-channel parallel processing still maintains high consistency (>0.9) and takes only 37% of the time of traditional methods, demonstrating the high efficiency of multi-channel signal synchronous processing and real-time interference removal.

[0088] III. Clinical Dynamic Follow-up Consistency (Table 3) In "Table 3: Validation of Quantitative Dynamic Monitoring of Postoperative Liver Cancer (Example)", 13 patients (P01–P13) were included and followed up at baseline, 7 days after surgery, and 30 days after surgery. After normalization and timeline alignment, the post-treatment expression level (ΔΔCt) decreased by an average of 4.54 units from preoperative to 30 days after surgery, with an average decrease of 71.64%; the mean kinetic alignment deviation (¶) was 0.032. Stratified by 30-day radiographic outcome, the average decrease was 72.32% for complete remission (n=4) and 70.74% for partial remission (n=3). At the individual level, a monotonic downward trend of "preoperative > 7 days after surgery > 30 days after surgery" was observed. This indicates that the quantitative changes obtained by this method are well consistent with radiographic responses and can be used for efficacy assessment and recurrence monitoring (see Table 3).

[0089] Table 3: Quantitative verification of postoperative dynamic monitoring (taking liver cancer as an example)

[0090] Description: Kinetic alignment deviation: characteristic time point error after time axis nonlinear correction (unit: number of cycles) Core conclusion: The expression fluctuation range in dynamic monitoring was compressed to ±0.3ΔΔCt, and the time axis alignment error was <0.05 cycles, demonstrating the reliability of kinetic alignment quantification for therapeutic response assessment.

[0091] Specifically, the key dimension definitions are: Matrix complexity score: A ranking of the concentration of protein / lipid interfering substances in serum samples (1 = pure, 5 = very turbid).

[0092] Reconstruction process consistency: the correlation coefficient of the real reaction process curves between multiple channels (measurement of the system's anti-interference ability).

[0093] Dynamic alignment deviation: The offset between a characteristic time point (such as an inflection point) and the standard time axis (reflecting the accuracy of time correction).

[0094] ΔΔCt: relative expression of the reference gene calibrated to the baseline (smaller value = higher expression).

[0095] IV. Summary Based on the results in Tables 1, 2, and 3, this application achieved its goals in terms of accuracy (significant error convergence and cross-batch stability), efficiency (low latency and high throughput), and clinical consistency (consistent with the direction and magnitude of imaging outcomes). This validates the usability and robustness of the end-to-end "acquisition-separation-reconstruction-quantification" process in complex matrices and high-throughput application scenarios.

[0096] Example 2: Relying solely on the method flow cannot guarantee the stability and efficiency of the detection. It also needs to rely on a specific system structure to implement these method steps. Especially in a high-throughput, automated detection environment, an integrated detection system can better coordinate various functional modules, reduce manual intervention, and improve detection efficiency and accuracy. Figure 2 ,This embodiment constructs a patient serum miRNA expression level detection system.

[0097] The patient serum miRNA expression level detection system includes: The signal acquisition module 100 is used to synchronously acquire fluorescence signals from multiple reaction channels; Interference signal separation module 200, used to perform statistical processing on the fluorescence signal to obtain common pattern information, and perform signal processing on the common pattern information to separate the systematic common pattern interference signal; A reaction progress reconstruction module 300 is used to deduct the systematic common mode interference signal from the fluorescence signal of each reaction channel and reconstruct the true reaction progress of each reaction channel; The expression level quantification module 400 is used to quantify the expression level of the target biomarker based on the actual reaction process.

[0098] Among them, the signal acquisition module 100 refers to a device for converting the fluorescent signal generated in the biological reaction process into an electrical signal or digital signal that can be processed. It can be implemented by a photomultiplier tube array with high sensitivity, a CCD camera with low noise or a CMOS sensor array, and its purpose is to obtain real-time fluorescence intensity data of the reaction channel; statistical processing refers to the mathematical or algorithmic analysis of the collected multi-channel fluorescence signal data to identify the general rules or trends in the data, which can be achieved by calculating the average, median, standard deviation of multiple channel signals or performing principal component analysis, independent component analysis and other methods, and its purpose is to extract the common patterns in the signal; common pattern information refers to the data pattern that is prevalent in the fluorescence signals of multiple reaction channels and has similar change trends or characteristics, and can refer to non-biological signal components superimposed on all channels caused by systematic factors (such as light source drift, detector noise), and its purpose is to characterize the characteristics of systematic interference; signal processing refers to filtering, transforming, decomposing or modeling the common pattern information. Distinguishing the interference components from the effective information can be achieved through Fourier transform, wavelet analysis, adaptive filtering or machine learning algorithms, with the aim of accurately identifying and separating systematic common-mode interference signals; systematic common-mode interference signals refer to non-target signals with similar change patterns that are caused by the detection system itself or environmental factors and affect all or most reaction channels at the same time, with the aim of eliminating errors introduced by the equipment or environment; subtraction refers to subtracting or removing the identified interference signal components from the original fluorescence signal to obtain a processed biological signal, with the aim of removing interference and restoring the authenticity of the signal; the true reaction process refers to the signal curve that can accurately reflect the changes in the biological reaction kinetics in each reaction channel after correction for the systematic common-mode interference signal, with the aim of providing interference-free biological reaction data; quantifying the expression level of the target biomarker refers to calculating or deriving a numerical value related to the concentration or activity of the target biomarker (such as miRNA) based on the reconstructed true reaction process, with the aim of obtaining accurate biomarker concentration information.

[0099] The present invention utilizes a modular system design to automate and accurately execute a method for detecting patient serum miRNA expression levels. First, the signal acquisition module 100 synchronously acquires fluorescence signals from multiple reaction channels, ensuring temporal consistency across all channel data and providing a foundation for subsequent signal analysis. Next, the interference signal separation module 200 statistically processes these synchronously acquired fluorescence signals to identify and extract common pattern information common across all channels. This information typically originates from systematic drift in the detection equipment or environment. The module then further processes this common pattern information to accurately isolate the systematic common pattern interference signal. This separation mechanism effectively identifies and removes non-biologically derived noise that pervasively affects all channels, thereby avoiding signal distortion caused by systematic interference in traditional methods. After the systematic common pattern interference signal is separated, the reaction progress reconstruction module 300 accurately deducts this interference signal from the original fluorescence signal of each reaction channel, thereby reconstructing the true reaction progress of each reaction channel. This critical step ensures that the signal curve of each channel accurately reflects the kinetic characteristics of the biological reaction, eliminating confounding from external factors such as equipment drift. Finally, the expression level quantification module 400 accurately quantifies the expression level of the target biomarker based on the real reaction process after these reconstructions by analyzing its characteristic parameters. Through this collaborative work, the system can overcome the reaction kinetic differences caused by the sample matrix effect in the prior art and the influence of the systemic common mode interference signal on the detection accuracy. The synchronization of signal acquisition, the accurate separation of the interference signal and the real reconstruction of the reaction process jointly ensure the reliability of the subsequent quantitative results. This integrated system structure not only improves the automation and efficiency of the detection, but also enhances the accuracy and stability of the quantification of miRNA expression levels in a high-throughput detection environment.

[0100] In some preferred embodiments, the present application is implemented as follows: The patient serum miRNA expression level detection system can be integrated into an automated detection platform. The signal acquisition module 100 can be configured to include a highly sensitive photomultiplier tube array or a low-noise scientific-grade CMOS image sensor, which is used to synchronously scan and record fluorescence intensity data from all reaction wells on a microplate at preset time intervals, such as every few seconds. The interference signal separation module 200 can be implemented by a high-performance digital signal processor (DSP) or a field-programmable gate array (FPGA), running a specific algorithm. For example, this module can receive the raw fluorescence signal data stream transmitted by the signal acquisition module 100 in real time and perform a sliding average or principal component analysis on the signals of all channels to identify common fluctuation patterns exhibiting high correlation between different channels, i.e., common pattern information. Subsequently, this module can employ an adaptive filtering algorithm, such as the least mean square (LMS) algorithm, to dynamically adjust filter parameters based on the common pattern information, thereby separating systematic common pattern interference signals from the original signal. The reaction progress reconstruction module 300 can be a software module running on the main control computer. This module receives the raw fluorescence signal data from the signal acquisition module 100 and the systematic common mode interference signal from the interference signal separation module 200. This software module can perform a point-by-point subtraction operation, deducting the corresponding interference signal value at each time point from the raw fluorescence signal of each reaction channel, thereby generating a series of processed fluorescence intensity time curves reflecting the actual biological reaction process. The expression level quantification module 400 can also be another software module on the main control computer. This module receives the reconstructed actual reaction progress data. This module can perform curve fitting on each actual reaction progress curve, for example, using a four-parameter logistic curve fit, and extract quantitative parameters related to miRNA expression level, such as maximum fluorescence intensity, time required to reach half-maximal fluorescence intensity, or reaction rate. Ultimately, these quantitative parameters can be converted into specific miRNA expression level values ​​and output to the user or stored in a database.

[0101] The system can effectively solve the problem of inconsistent reaction kinetics of each channel caused by sample matrix effect in the prior art, and the problem of being affected by systematic common mode interference signals changing with time. Through synchronous acquisition of the signal acquisition module 100, accurate identification and stripping of the interference signal separation module 200, and restoration of the reaction process reconstruction module 300, the system can obtain processed biological reaction process data. On this basis, the expression level quantification module 400 can accurately quantify the expression level of the target biomarker based on these real data. This improves the accuracy and stability of patient serum miRNA expression level detection, especially in high-throughput, automated detection scenarios, reducing human intervention and improving overall detection efficiency.

[0102] System-level performance evaluation: under the batch operation of the automated platform, based on "Table 1: Performance verification of serum miRNA detection for multiple diseases" (n=1551, 41 subgroups), the processed curve output by the system end to end is used for 4PL regression, and the average batch CV / inter-batch CV is 2.92% / 5.40%, and the average consistency with the clinical gold standard is 95.96% (Table 1). The results verify the stability and accuracy of the integrated process of "collection-separation-reconstruction-quantification" in the high-throughput scene.

[0103] While the application has been described above with reference to various embodiments, it will be understood that many changes and modifications can be made and equivalents can be substituted for elements thereof without departing from the scope of the application. That is, the methods, systems and apparatus discussed above are examples. Various configurations can omit, substitute, or add various procedures or components as appropriate. For instance, the methods can be performed in an order different from that described, and / or various steps can be added, omitted, and / or combined. Also, features described with respect to certain configurations can be combined in various other configurations, for instance, different aspects and elements of the configurations can be combined in a similar manner.

Claims

1. A method for detecting the expression level of miRNA in patient serum, characterized in that: The following steps are involved: S1, synchronously collect fluorescence signals from multiple reaction channels; S2. performing statistical processing on the fluorescence signal to obtain common pattern information; S3. performing signal processing on the common pattern information to separate a systematic common pattern interference signal; S4. subtracting the systematic common mode interference signal from the fluorescence signal of each reaction channel to reconstruct the true reaction process of each reaction channel; S5. quantifying the expression level of the target biomarker based on the actual reaction progress.

2. The method for detecting miRNA expression levels in patient serum according to claim 1, wherein: The specific steps of step S2 include: S21, obtaining the grouping information of the reaction channel; S22. Selecting the fluorescence signal of the control group channel according to the grouping information; S23. Perform statistical processing on the fluorescence signals of the control group channel to obtain common pattern information.

3. The method for detecting the expression level of miRNA in patient serum according to claim 1, wherein: The specific steps of step S3 include: S31, analyzing the time fluctuation characteristics of the common pattern information in real time; S32. Adjusting adjustment parameters for separating the systematic common mode interference signal according to the time fluctuation characteristics; S33: Process the common mode information using the adjustment parameter to separate the systematic common mode interference signal.

4. The method for detecting miRNA expression levels in patient serum according to claim 1, wherein: The specific steps of step S4 include: S41, obtaining the physical coordinates of the reaction channel; S42, forming a data point set by combining the fluorescence signal and the physical coordinates of the corresponding reaction channel; S43, performing spatial surface fitting on the data point set to obtain the spatial distribution of the systematic common mode interference signal; S44. Determine, according to the physical coordinates of the reaction channel, a specific interference value corresponding to the reaction channel from the spatial distribution of the systematic common mode interference signal; S45. Deduct the specific interference value from the fluorescence signal of each reaction channel to reconstruct the actual reaction process of each reaction channel.

5. The method for detecting the expression level of miRNA in patient serum according to claim 1, wherein: The specific steps of step S5 include: S51, determining the characteristic signal level of each of the real reaction processes; S52, performing amplitude normalization processing on the signal value of each of the true reaction processes according to the characteristic signal level; S53, performing nonlinear adjustment on the time axis of each of the real reaction processes so that the characteristic time point of the real reaction process is aligned with a preset reference time point; S54, extracting quantitative parameters from the real reaction process after amplitude normalization and time axis nonlinear adjustment; S55. Quantify the target biomarker expression level based on the quantification parameter.

6. The method for detecting the expression level of miRNA in patient serum according to claim 5, characterized in that: The specific steps of step S51 include: S511, identifying the maximum signal value in each of the real reaction processes; S512: Using the maximum signal value as the characteristic signal level of each of the real reaction processes.

7. The method for detecting the expression level of miRNA in patient serum according to claim 5, wherein: The specific steps of step S52 include: S521, determining the baseline signal level and saturation signal level of each of the real reaction processes; S522, performing interval normalization processing on the signal value of each of the true reaction processes according to the baseline signal level and the saturation signal level; S523 , performing nonlinear correction on the signal value after the interval normalization process to compensate for the nonlinear relationship between the signal response and the expression level of the target biomarker.

8. The method for detecting the expression level of miRNA in patient serum according to claim 5, wherein: The specific steps of step S53 include: S531, identifying multiple key time points in each of the real reaction processes; S532: Determine the correspondence between the multiple key time points and the preset reference time points; S533: According to the corresponding relationship, perform segmented stretching or compression processing on the time axis of each of the real reaction processes, so that the characteristic time points of the real reaction processes are aligned with the preset reference time points.

9. The method for detecting the expression level of miRNA in patient serum according to claim 8, characterized in that: The specific steps of step S531 include: S5311, performing sliding window averaging processing on the signal value of the real reaction process to obtain a smoothed signal value; S5312. Analyze the change rate of the smoothed signal value in a continuous time period to obtain signal change rate information; S5313. Based on the signal change rate information, identify the time points at which the change rate reaches a preset threshold or the change trend undergoes a significant change, as the multiple key time points.

10. A system for detecting the expression level of miRNA in patient serum, for detecting the expression level of miRNA in patient serum, characterized in that: The system includes a signal acquisition module, an interference signal separation module, a reaction process reconstruction module and an expression level quantification module; A signal acquisition module, used to synchronously acquire fluorescence signals from multiple reaction channels; an interference signal separation module, configured to perform statistical processing on the fluorescence signal to obtain common pattern information, and perform signal processing on the common pattern information to separate systematic common pattern interference signals; a reaction progress reconstruction module, configured to deduct the systematic common mode interference signal from the fluorescence signal of each reaction channel to reconstruct the true reaction progress of each reaction channel; The expression level quantification module is used to quantify the expression level of the target biomarker based on the actual reaction process.

Images