Detection method based on dual-label time-resolved fluoroimmunoassay technique

Abstract

The application relates to the technical field of immune analysis detection, and discloses a detection method based on double-label time-resolved fluorescence immune analysis technology.The method comprises the following steps: performing double-label time-resolved fluorescence signal collection, obtaining a historical first-label fluorescence signal data set and a historical second-label fluorescence signal data set of a target analyte in a preset historical time window; then performing data stability analysis on the two historical data sets to determine a first-label fluorescence stability factor and a second-label fluorescence stability factor; then taking the two stability factors as indexes to perform concentrated search in a fluorescence detection configuration space, and determining a target detection integration time; finally, performing double-label time-resolved fluorescence immune analysis detection on the target analyte based on the target detection integration time, obtaining a periodic detection data set, and identifying the data set by using a fluorescence detection identifier to obtain a target analyte detection result.

Description

Technical Field

[0001] This invention relates to the field of immunoassay detection technology, specifically a detection method based on dual-label time-resolved fluorescence immunoassay technology. Background Technology

[0002] In modern medical diagnostics, environmental monitoring, and food safety testing, immunoassay technology has become an indispensable detection method due to its high specificity and sensitivity. Among them, time-resolved fluorescence immunoassay technology, which can effectively eliminate non-specific fluorescence interference and further improve detection accuracy, has been widely used in the quantitative and qualitative detection of various target analytes. As detection requirements continue to upgrade, single-label time-resolved fluorescence immunoassay technology is gradually showing its limitations. For example, in complex sample matrices, the single-label signal is easily affected by background interference, leading to a decrease in detection accuracy. At the same time, it is difficult to achieve simultaneous detection of multiple target analytes, limiting the improvement of detection efficiency.

[0003] To overcome this limitation, dual-label time-resolved fluorescence immunoassay (DRIA) technology has emerged. This technology, by introducing two different labels, theoretically enables the simultaneous detection of two target analytes and reduces the interference risk from a single label signal. However, in practical applications, dual-label time-resolved fluorescence immunoassay still faces several challenges. First, the fluorescence signals generated by the two labels are prone to overlap or instability during acquisition, especially over long detection periods. Fluorescence signals are affected by factors such as temperature, sample concentration changes, and instrument drift, leading to significant differences in the stability of historical fluorescence signal datasets. Directly setting detection parameters based on this data can easily introduce detection errors. Second, current processing of dual-labeled fluorescence signals often uses fixed integration times, failing to consider the stability differences of different label fluorescence signals. This prevents dynamic adjustment of the integration time based on actual signal characteristics, resulting in insufficient fluorescence signal acquisition in some time periods and over-acquisition in others, further affecting the reliability of the detection data. Furthermore, when identifying periodic detection datasets, existing identification methods often fail to fully integrate the synergistic features of dual-labeled signals, analyzing only a single signal. This makes it difficult to effectively distinguish between specific and non-specific interference signals, resulting in accuracy and repeatability of target analyte detection results that do not meet practical detection needs. These problems severely restrict the promotion and application of dual-labeled time-resolved fluorescence immunoassay technology, necessitating a detection method that can solve these issues. Summary of the Invention

[0004] The purpose of this invention is to provide a detection method based on dual-label time-resolved fluorescence immunoassay technology to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a detection method based on dual-label time-resolved fluorescence immunoassay technology, the method comprising:

[0006] Perform dual-label time-resolved fluorescence signal acquisition to obtain a set of historical first-label fluorescence signal data and a set of historical second-label fluorescence signal data for monitoring the target analyte within a preset historical time window;

[0007] Data stability analysis is performed by traversing the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set to determine the first-label fluorescence stability factor and the second-label fluorescence stability factor.

[0008] Using the first and second labeled fluorescence stability factors as indexes, a centralized search is performed in the fluorescence detection configuration space to determine the target detection integration time;

[0009] Based on the target detection integration time, the target analyte is detected by dual-label time-resolved fluorescence immunoassay to obtain a periodic detection data set. The periodic detection data set is then identified using a fluorescence detection identifier to obtain the target analyte detection result.

[0010] Preferably, the process of performing dual-label time-resolved fluorescence signal acquisition further includes:

[0011] Obtain the set of background interference signal data of the target analyte within the preset historical time window;

[0012] The historical first-label fluorescence signal data set, the historical second-label fluorescence signal data set, and the background interference signal data set are simultaneously collected to construct a dual-label fluorescence interference dataset.

[0013] Preferably, the step of performing data stability analysis by traversing the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set further includes:

[0014] Time-frequency analysis was performed on the dual-labeled fluorescence interference dataset to extract the dynamic features of the first label and the dynamic features of the second label.

[0015] Based on the first label dynamic characteristics and the second label dynamic characteristics, the mean absolute deviation is calculated to obtain the first label fluorescence stability factor and the second label fluorescence stability factor.

[0016] Preferably, the step of performing a centralized search in the fluorescence detection configuration space using the first labeled fluorescence stability factor and the second labeled fluorescence stability factor as indexes further includes:

[0017] The fluorescence stability factors of the first label and the fluorescence stability factors of the second label of multiple samples, as well as the corresponding detection integration times of multiple samples, were obtained as the construction data.

[0018] A pre-constructed two-dimensional space is provided, wherein the origin of the coordinate system of the two-dimensional space is a reference point, the horizontal axis is the first label fluorescence stability factor, and the vertical axis is the second label fluorescence stability factor.

[0019] The constructed data is input into the two-dimensional space to obtain multiple sample space points, and the multiple sample space points are marked using the multiple sample detection integration time to obtain the fluorescence detection configuration space.

[0020] Preferably, the step of performing a centralized search in the fluorescence detection configuration space to determine the target detection integration time further includes:

[0021] The line in the fluorescence detection configuration space that passes through the first labeled fluorescence stability factor and is parallel to the vertical axis is extracted as the first reference line;

[0022] The line in the fluorescence detection configuration space that has been marked with the second fluorescence stability factor and is parallel to the horizontal axis is extracted as the second reference line;

[0023] The intersection of the first reference line and the second reference line is taken as the center reference point. Starting from the center reference point, a neighborhood space is constructed according to a preset search radius. The neighborhood space includes multiple neighborhood sample space points.

[0024] A concentrated search is performed on the multiple neighborhood sample spatial points to determine the target sample spatial point, and the sample detection integration time corresponding to the target sample spatial point is used as the target detection integration time.

[0025] Preferably, the centralized search of the plurality of neighborhood sample spatial points further includes:

[0026] Calculate the neighborhood density of the neighborhood space;

[0027] Randomly select a neighborhood sample space point from the edge of the neighborhood space as the current search point, and calculate the search density of the current search point;

[0028] Determine whether the search density is greater than or equal to the neighborhood density. If so, update the current search point to the starting point and continue the concentrated search until the preset number of searches is met. Use the search point obtained in the last search as the target sample space point.

[0029] Preferably, determining whether the search density is greater than or equal to the neighborhood density further includes:

[0030] If not, the failure count, which was initially zero, is updated to one, and a neighborhood sample space point is randomly selected from the edge of the neighborhood space as the current search point for search analysis. When the failure count is greater than the preset maximum failure count, the central reference point is used as the target sample space point.

[0031] Preferably, the step of performing dual-label time-resolved fluorescence immunoassay on the target analyte based on the target detection integration time further includes:

[0032] Start the dual-label time-resolved fluorescence detection device to collect real-time fluorescence signal data of the first label and real-time fluorescence signal data of the second label;

[0033] Based on the real-time first label fluorescence signal data and the real-time second label fluorescence signal data, the background noise interference amount is determined;

[0034] The robustness of fluorescence detection is determined by the amount of background noise interference and the correlation strength between the real-time first-label fluorescence signal data and the real-time second-label fluorescence signal data.

[0035] Preferably, the step of using a fluorescence detection identifier to identify the periodic detection data set further includes:

[0036] Signal drift bias is obtained by predicting the signal drift using the background noise interference and the fluorescence detection robustness of the periodic detection dataset.

[0037] The recognition threshold of the fluorescence detector is adjusted based on the signal drift deviation to obtain the adjusted recognition threshold.

[0038] The adjusted identification threshold is used to identify the periodic detection data set to obtain the detection results of the target analyte.

[0039] Preferably, the step of predicting signal drift in the periodic detection dataset based on the background noise interference and the fluorescence detection robustness further includes:

[0040] By combining the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set, a dynamic signal trajectory is constructed;

[0041] Based on the gradient characteristics of the dynamic signal trajectory, multiple signal hysteresis features are filtered out, and the multiple signal hysteresis features are converted into signal hysteresis quantities;

[0042] By combining the signal hysteresis and the fluorescence detection robustness, the signal drift bias is predicted.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] This detection method based on dual-label time-resolved fluorescence immunoassay acquires historical fluorescence signal data of two labels within a preset historical time window by performing dual-label time-resolved fluorescence signal acquisition, providing a comprehensive signal basis for subsequent optimization of detection parameters. Compared with traditional methods that rely on only a small amount of historical data or fixed parameters, this method can make full use of long-term historical signal data, more comprehensively reflecting the changing characteristics of the fluorescence signals of the two labels, and avoiding parameter setting deviations caused by insufficient data sample size.

[0045] In the data processing stage, this method performs data stability analysis by traversing historical datasets of first-label fluorescence signals and second-label fluorescence signals to determine the fluorescence stability factors of each label, accurately capturing the stability differences of fluorescence signals from different labels. This approach of evaluating stability separately for different labels overcomes the limitation of traditional methods that use a uniform stability standard for dual-label signals, allowing for more targeted parameter adjustments. For example, when there is a significant difference in fluorescence stability between the two labels, this difference can be clearly identified through the stability factor, providing a clear basis for the dynamic adjustment of the integration time. Traditional methods, lacking this distinction, often employ averaging, failing to consider the signal characteristics of both labels and potentially leading to poor signal acquisition for one label.

[0046] Regarding the determination of the integration time, this method uses the fluorescence stability factors of the two markers as an index to perform a centralized search within the fluorescence detection configuration space to determine the target integration time. This method, which dynamically adjusts the integration time based on signal stability, can match the optimal integration time to the actual stability of the fluorescence signals of the two markers. For example, for markers with high stability, the integration time can be appropriately shortened to improve detection efficiency while ensuring sufficient signal acquisition; for markers with low stability, the integration time can be appropriately extended to ensure sufficient effective signal acquisition and reduce interference caused by signal fluctuations. Compared to the fixed integration time mode used in traditional methods, this method can achieve adaptive optimization of the integration time under different detection scenarios. It avoids incomplete signal acquisition due to excessively short integration time, and also prevents low detection efficiency and signal redundancy due to excessively long integration time, effectively balancing detection efficiency and detection data quality.

[0047] In the detection execution and result recognition stages, this method uses dual-label time-resolved fluorescence immunoassay based on the target detection integration time to obtain a periodic detection dataset, which is then identified using a fluorescence detection identifier. Since the target detection integration time is optimized based on the stability characteristics of the two labels, the collected periodic detection dataset more accurately reflects the true information of the target analyte, reducing the influence of non-specific interference signals. Simultaneously, the fluorescence detection identifier fully integrates the synergistic features of the dual-labeled signals, rather than single-signal features, during the recognition process, enabling more precise differentiation between specific and interference signals and avoiding misjudgments caused by neglecting the correlation between the dual-labeled signals in traditional identification methods. Furthermore, by periodically collecting detection data, this method can monitor the dynamic changes of the target analyte in real time. Compared to traditional single-detection modes, it provides richer detection information, helping to comprehensively understand the state and concentration trends of the target analyte, further improving the reliability and practicality of the detection results. Whether for the detection of trace biomarkers in medical diagnosis or the screening of harmful residues in food safety testing, this method, with its superior performance, provides more effective technical support for practical detection work. Attached Figure Description

[0048] Figure 1 This is a schematic diagram illustrating the working principle of the detection method based on dual-label time-resolved fluorescence immunoassay technology described in this invention.

[0049] Figure 2 A flowchart for dual-label time-resolved fluorescence signal acquisition;

[0050] Figure 3 A flowchart for configuring space for fluorescence detection;

[0051] Figure 4 This is a flowchart for searching a set of spatial points in the neighborhood sample area. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Please see Figure 1 This invention provides a detection method based on dual-label time-resolved fluorescence immunoassay technology, the method comprising:

[0054] A dual-label time-resolved fluorescence signal acquisition process is performed to obtain historical datasets of first-label fluorescence signals and second-label fluorescence signals monitored within a preset historical time window. Data stability analysis is then performed on both datasets to determine the first-label fluorescence stability factor and the second-label fluorescence stability factor. Using these factors as indices, a focused search is conducted within the fluorescence detection configuration space to determine the target detection integration time. Based on this integration time, dual-label time-resolved fluorescence immunoassay is performed on the target analyte to obtain a periodic detection dataset. This periodic dataset is then identified using a fluorescence detection identifier to obtain the target analyte detection results.

[0055] Example 1: See Figure 2 When performing dual-label time-resolved fluorescence signal acquisition, setting a preset historical time window is fundamental. This window must cover a sufficient number of signal periods to capture the fluorescence response characteristics of the target analyte under potentially varying conditions. For example, for an in vitro diagnostic test, this window might be set to a monitoring period lasting several hours, acquiring data once per second, thus accumulating tens of thousands of data points to form a historical first-label fluorescence signal dataset and a historical second-label fluorescence signal dataset. These two datasets correspond to the specific fluorescence signals produced by antibodies or antigens labeled with two different rare-earth elements, respectively. Their time-resolved characteristics allow for measurement after delayed excitation to eliminate interference from short-lived background fluorescence. Acquiring the background interference signal dataset of the target analyte is a crucial step performed simultaneously. Background interference signals do not originate from the fluorescence of the target analyte itself, but rather from factors such as the sample matrix, non-specific binding, instrument dark current, or ambient stray light. This dataset is typically acquired within the same time window using a blank control sample without the target analyte or a detection channel shielded from a specific fluorescent label. The collected background signal is strictly aligned with the two aforementioned marked signal sets in the time dimension, and each marked signal data point corresponds to a background interference data point with the same timestamp.

[0056] Three datasets—historical first-label fluorescence signal data set, historical second-label fluorescence signal data set, and background interference signal data set—were simultaneously acquired to construct a dual-label fluorescence interference dataset. Synchronization was ensured through a unified system clock and triggering mechanism, guaranteeing complete consistency of the timelines of all data streams. This dataset is a multi-dimensional time series, where each time point contains three values: the first-label signal intensity, the second-label signal intensity, and the background noise level. The construction of this composite dataset provides a complete information foundation for subsequent analysis of signal behavior in real-world noisy environments.

[0057] After construction, time-frequency analysis is performed on the dual-labeled fluorescence interference dataset. Time-frequency analysis is a tool that can simultaneously reveal how the frequency components of a signal change over time. In this implementation, a sliding window short-time Fourier transform is applied to the entire dataset. The analysis process first divides the long-term series data into a series of overlapping shorter time segments (windows). For the first labeled signal data subset within each window, its Fourier transform is calculated to obtain the frequency spectrum of the signal within that time period. The same operation is applied to the second labeled signal data subset and the background interference data subset. Features characterizing the dynamic behavior of the signal are extracted from these spectra. For the first labeled signal, the first labeled dynamic features may include the time sequence of the amplitude of its main peak, the offset trajectory of the center frequency, and the fluctuation of specific frequency band energy (such as the signal's characteristic transmission frequency band). The extraction of the second labeled dynamic features follows the same logic but focuses on its own unique frequency characteristics. The spectral features of the background interference signal are also extracted, mainly to understand the frequency domain characteristics of the noise.

[0058] Based on the extracted first-labeled and second-labeled dynamic features, their mean absolute deviation is calculated to obtain the stability factor. Taking the calculation of the first-labeled fluorescence stability factor as an example: the first-labeled dynamic feature obtained from time-frequency analysis is, for example, a sequence containing amplitude values ​​at N time points. First, the arithmetic mean of all amplitude values ​​in the sequence is calculated. Then, the absolute difference between each amplitude value in the sequence and this mean is calculated. Finally, these absolute differences are summed and divided by N to obtain the mean absolute deviation of the feature sequence. This value intuitively reflects the average degree of fluctuation of the signal amplitude around its average level. The smaller the mean absolute deviation value, the more stable the signal is in that feature dimension. To obtain the final first-labeled fluorescence stability factor, the reciprocal of the mean absolute deviation is usually taken, or a normalization to a specific interval is used, so that a higher value represents better stability. The second-labeled fluorescence stability factor is obtained through the same calculation process, but its calculation basis is the second-labeled dynamic feature sequence.

[0059] The construction of the dual-labeled fluorescence interference dataset places the labeled signal and background noise within the same analytical framework. Time-frequency analysis goes beyond simple time-domain or frequency-domain analysis, revealing the evolution of signal frequency components over time. This is crucial for capturing transient changes or slow drifts that might be overlooked in fixed-time-point or fixed-frequency analysis. Mean absolute deviation (MAD), as a robust measure of dispersion, is less sensitive to extreme values ​​in the data and better reflects the central tendency of most data points compared to variance or standard deviation. This allows the calculated first-labeled fluorescence stability factor and second-labeled fluorescence stability factor to more reliably characterize the overall stability of the signal over the monitoring period. These factors serve as key input parameters for configuring the search space in subsequent steps; their accuracy and representativeness directly affect the rationality of the final determined target detection integration time, thus impacting the performance of the entire detection method.

[0060] Example 2: See Figure 3 The data originates from a long-term accumulated experimental database, containing records of a large number of historical detection samples. Each record consists of three key data elements: a fluorescence stability factor for the first label of a sample, a fluorescence stability factor for the second label of a sample, and a detection integration time that has been proven effective under the given stability factor conditions. For example, a record might show that when the calculated stability factor for the first label is 0.85 and the stability factor for the second label is 0.72, the corresponding optimal detection integration time is 350 milliseconds. This historical data underwent preprocessing before being entered into the database, including removing obvious outliers, normalizing the stability factors to ensure all values ​​fall within the range of 0 to 1, and validating the integration time to ensure it matches the corresponding signal quality. These cleaned and standardized data elements form the foundation for subsequent spatial construction.

[0061] The core step is to pre-construct a two-dimensional space, an abstract mathematical representation used to establish the mapping between stability factors and integration time. The origin of the coordinate system is set as a reference point, typically representing the theoretically lowest stable state. The horizontal axis is defined as the first labeled fluorescence stability factor, with its scale linearly extending from 0 (completely unstable) to 1 (completely stable). Similarly, the vertical axis is defined as the second labeled fluorescence stability factor, with a scale consistent with the horizontal axis. Each point in this two-dimensional plane corresponds to a pair of possible stability factor values. The creation of this space provides a framework for visualizing and mathematically processing multi-parameter optimization problems.

[0062] All the construction data is input into this two-dimensional space. The first two values ​​of each historical data triple (first factor, second factor, integration time), i.e., a pair of stability factors, determine the planar coordinates of a point. This point is called a sample space point. The third value in the triple, the sample detection integration time, is associated with this spatial point as a label attribute or "value." After all historical data records are mapped into the space, a large number of sample space points with integration time labels are formed. The distribution of these points in the space may be uneven; some areas have dense point sets, representing common stability combinations; others are sparse, representing more special operating conditions. This set composed of sample points and their labels is the fluorescence detection configuration space. It is essentially a discrete lookup table based on historical experience, but its structure allows for proximity search and interpolation, rather than simple exact matching.

[0063] When a new target analyte needs to be detected, the system first calculates the current first-label fluorescence stability factor and the second-label fluorescence stability factor based on its real-time acquired historical signal data. Using these two calculated factor values ​​as indexes, the process of searching the configuration space begins. First, points in the two-dimensional space with an x-coordinate equal to the current first-label fluorescence stability factor are found; all these points form a straight line perpendicular to the x-axis, which is extracted as the first reference line. Similarly, all points with an y-coordinate equal to the current second-label fluorescence stability factor are found; these points form a straight line perpendicular to the y-axis, which is extracted as the second reference line. The intersection of these two reference lines in space, i.e., the coordinate point determined by the current two factor values, is defined as the central reference point. This point represents the theoretical position of the current system stability state in the configuration space.

[0064] Since the configuration space is discrete, the probability that a historical sample point happens to exist at the central reference point is very low. Therefore, the search needs to be carried out around this central point. A circular neighborhood space is constructed with the central reference point as the center and a preset search radius. The size of this radius can be adaptively adjusted according to the overall density of sample points in the configuration space, with the goal of including a certain number of neighborhood sample space points. The integration time information carried by these sample points falling within the circle constitutes the candidate set for the current search operation. A concentrated search is performed on multiple neighborhood sample space points to find the most suitable point from these candidate points and determine its label value, i.e., the sample detection integration time, as the final target detection integration time. Search strategies can be diverse, such as calculating the average or median of the integration times of all points in the neighborhood, or using a more complex weighted average, with weights inversely proportional to the distance between the point and the central point. Sample points closer to the central point have a more similar stability state to the current system, and therefore their integration time values ​​have a higher reference weight.

[0065] Example 3: See Figure 4 This space is a circular region defined by a preset search radius R, with the central reference point as its geometric center. This radius is not fixed but dynamically adjusted based on the global distribution density of sample points throughout the fluorescence detection configuration space. The basic principle is: in sparsely distributed macroscopic regions, a larger search radius is used to encompass a sufficient number of candidate points; in densely distributed regions, a smaller search radius is used to achieve a more refined local search. All sample space points falling within this circular region constitute the candidate set for this search, called the neighborhood sample space points.

[0066] The neighborhood density of this neighborhood space is calculated. This density is a macroscopic indicator that reflects the degree of clustering of candidate points within this local region. The calculation formula is as follows: in: This represents the calculated neighborhood density. This represents the total number of sample space points that fall within the current circular neighborhood space. This indicates the currently used preset search radius. This represents the area of ​​the circular neighborhood. The physical meaning of this density is the number of candidate points per unit area. The higher the value, the richer the historical experience data in the area, and theoretically, the higher the probability of finding high-quality candidate points.

[0067] The search iteration begins randomly. A sample point is randomly selected from the edge of the current neighborhood space as the current search point. This randomization avoids the bias that might arise from always starting the search from a fixed direction, increasing the diversity of the exploration. Using this newly selected current search point as the center, a new sub-neighborhood space is constructed with the exact same search radius R as before. The density of this sub-neighborhood space, i.e., the search density, is then calculated. Calculation method and calculation Completely identical: Count the number of points falling into the sub-neighborhood and divide by the area of ​​the circle.

[0068] The system compares these two density values ​​and determines whether the density of the sub-region where the current search point is located is higher than that of the sub-region where the search point is located. ) is greater than or equal to the average density of the entire original neighborhood ( This typically means we've randomly jumped to a "golden" region with richer data points, potentially containing more valuable information. At this point, the search is considered successful. The system updates the center point of this new, denser region—the current search point—as the starting center point for the next iteration. Simultaneously, the failure count is reset to zero. The search process then uses this new starting point as the center to reconstruct the neighborhood, calculate neighborhood density, and begin a new round of random selection and density comparison. This iterative process continues until a preset limit on the number of searches is reached. At this point, the sample space point reached in the last successful iteration is selected as the target sample space point. This mechanism gives the search a tendency to move towards denser, more information-rich regions.

[0069] If the calculated search density Smaller than the original neighborhood density If the random jump reaches a relatively sparse region with weak historical data support, the search attempt is considered a failure. The system does not move the search center but increments a failure counter specifically for recording consecutive failures. Subsequently, the system randomly selects another sample space point from the edge of the original neighborhood space as the new current search point, repeating the search density calculation and comparison process. If such consecutive failures continue, causing the failure counter to accumulate beyond a preset threshold—the maximum failure count—it indicates that within the current initial neighborhood space, there may not be a significantly better dense region than the location of the central reference point, or that the random search is failing to effectively hit the target for some reason. To avoid endless invalid attempts, the search algorithm will proactively terminate the iteration and adopt a conservative strategy: directly determining the initial central reference point itself as the target sample space point. This means that the system's final integration time scheme is based on the direct mapping of the current real-time stability factor in the configuration space, and fails to find a better alternative validated by historical data in its neighboring region.

[0070] Example 4: Starting the dual-label time-resolved fluorescence detection device is not a simple power-on operation, but involves a series of precise initialization steps. The device's optical module first performs a self-test and calibration. The intensity of the excitation source is set to a preset operating point, and the gain and bias of the photomultiplier tube or other detectors are set according to the expected signal intensity of the target analyte. Most importantly, the target detection integration time, previously determined through a search, is loaded into the timing control unit, which strictly defines the duration of each fluorescence signal acquisition. After the device starts, it begins real-time signal acquisition of the current sample. The system synchronously captures fluorescence signals from two different wavelength channels according to the set integration time window, denoted as real-time first-label fluorescence signal data and real-time second-label fluorescence signal data, respectively. These data are discrete intensity value sequences that vary over time, reflecting the fluorescence decay process of the label within a specific integration time.

[0071] The background noise interference is determined based on these real-time signal data. This interference is not directly measured but derived through calculation and analysis. The analysis typically selects the tail end of the signal attenuation curve, where the specific fluorescence of the target analyte has largely decayed, and the remaining fluctuations mainly originate from background interference. The system calculates the standard deviation of the intensity within this signal range, or uses more complex algorithms such as fitting a noise model to estimate its amplitude. This calculated value is quantified as the background noise interference, which objectively reflects the overall level of non-specific signals in the current detection environment.

[0072] The system needs to evaluate the relationship between two signal channels by calculating the correlation strength between real-time first-label fluorescence signal data and real-time second-label fluorescence signal data. Correlation strength describes the degree of consistency in the changing trends of the two signal sequences. The calculation is not a simple comparison of absolute intensities, but rather an analysis of their relative fluctuation patterns. A common method is to calculate the correlation coefficient of the intensity values ​​of the two signal sequences at the same time point. A high positive correlation coefficient indicates that the two labeled signals are affected by similar interference factors, and their changes are coordinated; while a low coefficient may indicate that the interference source is independent or that an anomaly has occurred in a certain channel. Ultimately, the robustness of fluorescence detection is determined by the amount of background noise interference and the correlation strength. Robustness is a comprehensive index, and its calculation logic is: under the same background noise level, the more coordinated the performance of the two signals (the higher the correlation strength), the better the overall detection system's robustness against random interference and in identifying true signals. Therefore, robustness is usually positively correlated with correlation strength and negatively correlated with the amount of background noise interference. A simplified quantification method is to divide the correlation strength value by the amount of background noise interference and then perform appropriate normalization to obtain a dimensionless scalar value to characterize robustness.

[0073] Table 1: Real-time signal acquisition and background noise analysis data. Referring to Table 1, a segment of real-time signal data is captured within a target detection integration time (e.g., 1000 ms). The time point column shows the time since the excitation began. The real-time first marker signal intensity and real-time second marker signal intensity columns record intensity information such as photon counts or voltage values ​​detected by the two channels at the corresponding time points. The signal attenuation tail interval identifier column indicates which time points are determined by the system to be the signal attenuation tail and is used to calculate the background noise interference. For example, starting from 400 ms, the signal intensity has dropped to a low and slowly changing level, and the data in this interval is used to calculate the background noise interference. The system may calculate the standard deviation of the first marker signal intensity within this identifier interval as 8.5 counts and the standard deviation of the second marker signal intensity as 7.2 counts, and finally synthesize a background noise interference. At the same time, the system calculates the correlation strength between the two signal sequences throughout the integration time, which may yield a correlation coefficient as high as 0.98, indicating that the two signals are highly coordinated. By combining high correlation strength with low noise interference, the system ultimately calculates a high fluorescence detection robustness value, indicating good environmental quality and more reliable results for subsequent identification steps. The entire process demonstrates a complete closed loop from parameter setting to real-time data acquisition and immediate quality assessment.

[0074] Example 5: Signal drift prediction on a periodic detection dataset, which is a set of intensity readings of first and second marker fluorescence signals obtained over multiple consecutive detection cycles within the target detection integration time. Signal drift prediction is not a simple time-series extrapolation, but a multi-factor fusion analysis process. First, the system retrieves historical first-marker fluorescence signal datasets and historical second-marker fluorescence signal datasets. These historical data provide a baseline of signal behavior over a longer period. The system aligns and overlays the current cycle's data with these historical data to construct a dynamic signal trajectory spanning the past and present. This trajectory clearly demonstrates the overall path and trend of signal intensity changes over time.

[0075] Based on the constructed dynamic signal trajectory, the system further analyzes its gradient characteristics. Gradient calculation reveals the rate and direction of signal change. By identifying key points in the trajectory where the rate of change significantly changes, the system can filter out multiple signal hysteresis features. These features may manifest as: after a certain time point, the rate of increase or decrease in signal strength slows down significantly; or the rates of change of two labeled signals begin to diverge. These phenomena often indicate that the system response may be delayed or desynchronized to some extent. These qualitative features are further converted into quantitative signal hysteresis. This hysteresis is a comprehensive indicator that may quantify the delay time of the rate of change or the percentage decrease in the rate of change.

[0076] The implementation process enters the crucial fusion phase, where the calculated signal hysteresis is fused with the previously determined fluorescence detection robustness. The robustness metric reflects the overall anti-interference capability of the detection system, while the signal hysteresis specifically indicates anomalous behavior in the time dimension. The fusion algorithm assigns different weights to these two metrics. Generally, in cases of low robustness, even small signal hysteresis may be given higher weight because it is more likely to indicate severe drift. Conversely, in a high-robust context, the system's tolerance for hysteresis increases accordingly. Through this weighted fusion calculation, the system ultimately outputs a quantified prediction value, namely the signal drift deviation. This value quantitatively estimates the degree and direction to which the signal may deviate from its expected ideal position.

[0077] Based on the predicted signal drift bias, the system adjusts the recognition threshold of the fluorescence detection identifier. The fluorescence detection identifier is essentially a decision function, and its built-in recognition threshold is the key boundary for distinguishing between positive and negative signals, or quantifying concentration levels. In an ideal drift-free environment, using a fixed threshold is feasible. However, when signal drift bias exists in the prediction, continuing to use a fixed threshold may lead to misjudgments. Therefore, the adjustment mechanism shifts or scales the original threshold accordingly based on the magnitude and sign of the signal drift bias. For example, if the prediction shows an overall positive signal drift (i.e., overall intensity is high), the system will proportionally increase the recognition threshold to avoid misjudging background noise as a weak positive signal. This adjustment process is dynamic and adaptive, ensuring that the decision boundary always maintains a relatively consistent positional relationship with the actual signal distribution. The adjusted recognition threshold is then used to perform the final identification of the periodic detection dataset. The signal intensity reading for each detection cycle is compared with this dynamically adjusted threshold. The identification logic depends on the specific detection type: for qualitative detection, an intensity higher than the threshold is considered positive, and vice versa; for quantitative detection, the specific concentration of the target analyte is calculated based on the difference between the intensity value and the threshold, with reference to the standard curve.

[0078] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0079] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A detection method based on dual-label time-resolved fluorescence immunoassay, characterized in that, Includes the following steps: Perform dual-label time-resolved fluorescence signal acquisition to obtain a set of historical first-label fluorescence signal data and a set of historical second-label fluorescence signal data for monitoring the target analyte within a preset historical time window; Data stability analysis is performed by traversing the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set to determine the first-label fluorescence stability factor and the second-label fluorescence stability factor. Using the first and second labeled fluorescence stability factors as indexes, a centralized search is performed in the fluorescence detection configuration space to determine the target detection integration time; Based on the target detection integral time, the target analyte is detected by dual-label time-resolved fluorescence immunoassay to obtain a periodic detection data set, and the periodic detection data set is identified by a fluorescence detection identifier to obtain the target analyte detection result; The step of performing a centralized search within the fluorescence detection configuration space to determine the target detection integration time also includes: The line in the fluorescence detection configuration space that passes through the first labeled fluorescence stability factor and is parallel to the vertical axis is extracted as the first reference line; The line in the fluorescence detection configuration space that has been marked with the second fluorescence stability factor and is parallel to the horizontal axis is extracted as the second reference line; The intersection of the first reference line and the second reference line is taken as the center reference point. Starting from the center reference point, a neighborhood space is constructed according to a preset search radius. The neighborhood space includes multiple neighborhood sample space points. A concentrated search is performed on the multiple neighborhood sample spatial points to determine the target sample spatial point, and the sample detection integration time corresponding to the target sample spatial point is used as the target detection integration time.

2. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 1, characterized in that, The process of performing dual-label time-resolved fluorescence signal acquisition also includes: Obtain the set of background interference signal data of the target analyte within the preset historical time window; The historical first-label fluorescence signal data set, the historical second-label fluorescence signal data set, and the background interference signal data set are simultaneously collected to construct a dual-label fluorescence interference dataset.

3. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 2, characterized in that, The step of performing data stability analysis by traversing the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set also includes: Time-frequency analysis was performed on the dual-labeled fluorescence interference dataset to extract the dynamic features of the first label and the dynamic features of the second label. Based on the first label dynamic characteristics and the second label dynamic characteristics, the mean absolute deviation is calculated to obtain the first label fluorescence stability factor and the second label fluorescence stability factor.

4. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 3, characterized in that, The step of performing a centralized search in the fluorescence detection configuration space using the first labeled fluorescence stability factor and the second labeled fluorescence stability factor as indexes also includes: The fluorescence stability factors of the first label and the fluorescence stability factors of the second label of multiple samples, as well as the corresponding detection integration times of multiple samples, were obtained as the construction data. A pre-constructed two-dimensional space is provided, wherein the origin of the coordinate system of the two-dimensional space is a reference point, the horizontal axis is the first label fluorescence stability factor, and the vertical axis is the second label fluorescence stability factor. The constructed data is input into the two-dimensional space to obtain multiple sample space points, and the multiple sample space points are marked using the multiple sample detection integration time to obtain the fluorescence detection configuration space.

5. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 1, characterized in that, The centralized search of the multiple neighborhood sample spatial points also includes: Calculate the neighborhood density of the neighborhood space; Randomly select a neighborhood sample space point from the edge of the neighborhood space as the current search point, and calculate the search density of the current search point; Determine whether the search density is greater than or equal to the neighborhood density. If so, update the current search point to the starting point and continue the concentrated search until the preset number of searches is met. Use the search point obtained in the last search as the target sample space point.

6. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 5, characterized in that, The step of determining whether the search density is greater than or equal to the neighborhood density also includes: If not, the failure count, which was initially zero, is updated to one, and a neighborhood sample space point is randomly selected from the edge of the neighborhood space as the current search point for search analysis. When the failure count is greater than the preset maximum failure count, the central reference point is used as the target sample space point.

7. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 1, characterized in that, The dual-label time-resolved fluorescence immunoassay detection of the target analyte based on the target detection integration time further includes: Start the dual-label time-resolved fluorescence detection device to collect real-time fluorescence signal data of the first label and real-time fluorescence signal data of the second label; Based on the real-time first label fluorescence signal data and the real-time second label fluorescence signal data, the background noise interference amount is determined; The robustness of fluorescence detection is determined by the amount of background noise interference and the correlation strength between the real-time first-label fluorescence signal data and the real-time second-label fluorescence signal data.

8. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 7, characterized in that, The step of using a fluorescence detection identifier to identify the periodic detection data set also includes: Signal drift bias is obtained by predicting the signal drift using the background noise interference and the fluorescence detection robustness of the periodic detection dataset. The recognition threshold of the fluorescence detector is adjusted based on the signal drift deviation to obtain the adjusted recognition threshold. The adjusted identification threshold is used to identify the periodic detection data set to obtain the detection results of the target analyte.

9. The detection method based on dual-label time-resolved fluorescence immunoassay as described in claim 8, characterized in that, The method of predicting signal drift in the periodic detection dataset based on the background noise interference and the fluorescence detection robustness further includes: By combining the historical first-label fluorescence signal data set and the historical second-label fluorescence signal data set, a dynamic signal trajectory is constructed; Based on the gradient characteristics of the dynamic signal trajectory, multiple signal hysteresis features are filtered out, and the multiple signal hysteresis features are converted into signal hysteresis quantities; By combining the signal hysteresis and the fluorescence detection robustness, the signal drift bias is predicted.

Images