Diarrhea virus multiple detection centrifugal microfluidic chip detection method and system

Abstract

This invention belongs to the field of biological detection technology and is used to solve the problem that existing technologies cannot identify specific distortion patterns caused by shared resource competition from multi-channel fluorescence data streams. Specifically, it is a method and system for detecting diarrhea virus multiplex centrifugal microfluidic chips, including: acquiring the growth coefficients of multiple detection channels during thermal cycling, recording the event cycle number of events that cause curve morphology abrupt changes; constructing a feature vector set containing event temporal features, spatial features, and correlation features, and calculating a competition fingerprint index; determining whether to enter a resource competition calibration mode based on the matching relationship between the competition fingerprint index of the feature vector set and a preset range; this invention maps curve morphology abrupt changes of different causes to differentiated quantitative index ranges; and effectively separates distortions caused by resource competition from distortions caused by global system anomalies, local physical anomalies, and temperature disturbances.

Description

Technical Field

[0001] This invention belongs to the field of biological detection technology, specifically a method and system for multiplex detection of diarrhea viruses using a centrifugal microfluidic chip. Background Technology

[0002] Centrifugal microfluidic chip technology integrates sample distribution, nucleic acid amplification, and fluorescence detection into a single chip, enabling the automation and miniaturization of multiplex detection of diarrhea viruses. In this type of detection system, multiple target amplification reaction chambers are typically arranged on the same chip substrate. The liquid is distributed by centrifugal force, and the polymerase chain reaction is completed with the help of a thermal cycling module. Finally, the virus is qualitatively or quantitatively interpreted based on the fluorescence amplification curves of each chamber.

[0003] To simplify chip structure and fluid control, existing technologies often employ an engineering scheme where each reaction chamber shares a premix (containing polymerase, primers, probes, and free nucleotides) from the same source. Under this architecture, existing analytical methods typically process the fluorescence data of each chamber independently, calculating the cycle threshold based on a preset fluorescence threshold or a standard S-curve model. This approach implicitly assumes that the amplification reactions in each chamber are independent at the resource level and do not interfere with each other. However, when the sample is infected with multiple viruses and the load of each target is high, different chambers may compete for resources due to the limited resources (especially polymerase and free nucleotides) shared in the premix. The highly competitive target that enters the exponential amplification phase first will rapidly consume the shared resources, causing the amplification reactions of other targets to be inhibited in the mid-to-late stages, resulting in an atypical "early plateau" or "downward curve" fluorescence curve.

[0004] In this situation, if the traditional method of independently analyzing each curve is still used, the cycle threshold of the suppressed target will be severely overestimated, and may even be misjudged as negative in low-load infection scenarios. More complicatedly, other interfering factors such as bubble rupture, local temperature fluctuations, and non-specific amplification may occur during chip operation, which can also cause similar morphological distortions in the fluorescence curve. Existing technologies lack effective means to distinguish the curve distortions caused by the above different reasons, resulting in the inability to accurately identify the source of distortion and to reasonably compensate for the detection results of the suppressed channel when facing special distortions caused by resource competition, thus affecting the accuracy and reliability of multiple detection results. Therefore, how to identify specific distortion patterns caused by shared resource competition from multi-channel fluorescence data streams and thereby achieve effective calibration of the distortion curves is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for detecting diarrhea virus multiplex centrifugation microfluidic chips, which solves the problem that the existing technology cannot identify specific distortion patterns caused by competition for shared resources from multi-channel fluorescence data streams, and thereby achieve effective calibration of distortion curves. The technical problem to be solved by the present invention is: how to provide a method and system for detecting multiplex centrifugal microfluidic chips for diarrheal viruses that can identify specific distortion patterns caused by competition for shared resources in multi-channel fluorescence data streams and thereby achieve effective calibration of distortion curves.

[0006] The objective of this invention can be achieved through the following technical solutions: On one hand, the present invention provides a method for detecting diarrhea virus multiplex centrifugation microfluidic chip, comprising: The fluorescence intensity sequences of multiple detection channels are acquired in real time during thermal cycling, and the fluorescence intensity ratio of adjacent cycles is calculated based on the fluorescence intensity sequences as the growth coefficient of the corresponding detection channel. Based on the numerical value of the growth coefficient, events that cause sudden changes in curve shape are identified, and the event cycle number of the event is recorded. Based on the event cycle number of all events that cause abrupt changes in curve shape, the preset physical location code of the detection channel corresponding to each event, and the number of associated channels of each event within the preset cycle window, a feature vector set containing event temporal features, spatial features, and associated features is constructed. The competitive fingerprint index is calculated based on the temporal interval consistency among the feature vectors in the feature vector set. Based on the matching relationship between the competitive fingerprint index of the feature vector set and the preset range, it is determined whether to enter the resource competition calibration mode. When entering the resource competition calibration mode, the detection channel is marked as a strong competitor group or a suppressed group based on the spatial-temporal correlation index of the feature vectors in the feature vector set. The detection results of the suppressed group were calibrated by calculating the degree of linear correlation and extrapolating linear regression.

[0007] On the other hand, the present invention provides a diarrhea virus multiplex detection centrifugal microfluidic chip detection system, including a processor, which executes the above-described diarrhea virus multiplex detection centrifugal microfluidic chip detection method when the processor is running.

[0008] The present invention has the following beneficial effects: By constructing a four-dimensional feature vector that includes normalized temporal features, physical location encoding, number of associated channels, and fluorescence reduction, and calculating a competition fingerprint index based on the spatial-temporal correlation index of the event sequence, the abrupt changes in curve morphology caused by different reasons are mapped to differentiated quantitative index ranges. Compared with the shortcomings of existing technologies that independently analyze each channel curve and cannot distinguish the root cause of the mutation, this invention can effectively separate the distortion caused by resource competition (manifested as uniform temporal intervals of the event sequence and a competition fingerprint index falling in the range of 0.15 to 0.45) from the distortion caused by global system anomalies, local physical anomalies, and temperature perturbations. This avoids triggering incorrect calibration or missing the real competition scenario due to misjudgment, and provides a reliable decision basis for subsequent targeted processing. For identified resource competition scenarios, linear regression extrapolation calibration is performed using the suppressed channel from the peak cycle to the undisturbed exponential growth segment before distortion. The quality of exponential growth is verified by calculating the linear correlation coefficient between the natural logarithm of fluorescence intensity in this segment and the cycle number. Calibration is performed only when the correlation coefficient meets preset conditions, avoiding the direct use of atypical curves after distortion for cycle threshold calculation. Compared with the shortcomings of existing technologies, which cannot compensate for competition interference and thus cause the cycle threshold of weak competitors to be seriously overestimated or even missed, this invention can reconstruct the true amplification curve of the suppressed channel, significantly improving the quantitative accuracy and negative / positive interpretation reliability of multiplex detection in mixed infection high-load samples. By employing a temperature screening step and orthogonal verification between the heating unit temperature sequence and fluorescence events, false events caused by temperature control fluctuations are eliminated. Furthermore, through a temperature disturbance source identification step, based on the spatial distribution pattern of temperature-related events and competitive background information, temperature anomalies are attributed to local contact thermal resistance disturbances, temperature disturbances under resource competition, temperature control system oscillation disturbances, or environmental temperature disturbances, and corresponding refined diagnostic labels are output. Compared to existing technologies that can only detect temperature exceedances but cannot pinpoint the specific cause, this invention provides operators with a clear basis for troubleshooting system status and distinguishing between hardware faults and sample-related anomalies, enhancing the system's robustness and maintainability under complex operating conditions. In the identification of abrupt change events in curve morphology, the adaptive trigger coefficient is dynamically selected by storing the competition fingerprint index of the previous batch of detections. When the historical competition fingerprint index is within the competition index range, a coefficient of 1.5 is used to improve sensitivity; otherwise, a coefficient of 2.0 is used to maintain stability. Compared with existing technologies that use fixed thresholds or rely solely on the data of the current batch, this invention uses the historical competition status as a correction factor for the current batch detection, realizing an adaptive feedback closed loop between batches. This allows the sensitivity of abrupt change event identification to be dynamically adjusted according to the system's operating history, avoiding missed detections in scenarios with frequent resource competition and reducing false triggers in normal scenarios, thus improving the system's adaptability to different application environments. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is the main flowchart of Embodiment 1 of the present invention; Figure 2 This is a system block diagram of Embodiment 2 of the present invention. Detailed Implementation

[0011] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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.

[0012] In the field of multiplex detection of diarrheal viruses, centrifugal microfluidic chip technology integrates sample distribution, nucleic acid amplification, and fluorescence detection into a single chip, achieving automated and high-throughput simultaneous detection of multiple targets. To simplify chip structure and fluid control, existing technologies typically employ an engineering scheme where each reaction chamber shares a premixed solution (containing polymerase, primers, probes, and free nucleotides) from the same source. From the perspective of PCR amplification kinetics, the underlying physical logic of this scheme lies in the fact that the amplification reactions in each chamber are pre-programmed as independent chemical processes that do not interfere with each other.

[0013] However, when the sample is infected with multiple viruses and the viral load of each target is high, this premise deviates significantly from the actual physical process. With limited shared resources, the highly competitive targets that enter the exponential amplification phase first rapidly consume polymerase and free nucleotides, causing the amplification reaction of other targets to be inhibited in the mid-to-late stages. Their fluorescence curves exhibit atypical "early plateau" or "downward curve" morphologies. More complexly, other interfering factors such as bubble rupture, localized temperature fluctuations, and non-specific amplification may occur during chip operation, also causing similar morphological distortions in the fluorescence curves. Current technologies lack a mechanism to distinguish between these different causes of curve distortion, making it impossible to identify special distortion patterns caused by resource competition, and even more impossible to effectively compensate for the detection results of inhibited channels.

[0014] For example, in multiplex viral testing of clinical diarrhea samples, when a patient is simultaneously infected with rotavirus and norovirus and both viral loads are at high levels, rotavirus, due to its higher amplification efficiency, enters the exponential phase first, consuming a large amount of polymerase resources in the shared premix. This causes the norovirus amplification curve to bend abnormally downwards in the middle of the exponential growth phase. If the method of analyzing each curve independently is still used, the system will misjudge norovirus as having a low viral load or being negative, resulting in a missed detection. Furthermore, when contact thermal resistance fluctuations occur at the edge of the chip, causing a sudden drop in temperature, the fluorescence curve of the relevant channel will also undergo morphological changes. Existing technology cannot distinguish this change from changes caused by resource competition, thus incorrectly attributing temperature disturbance events to competition or vice versa, leading to false or missed triggering of calibration logic.

[0015] If the above problems are not addressed, the multiplex detection system will continue to lose its ability to objectively identify resource competition scenarios. Specifically, the failure to identify distortions caused by resource competition will lead to a significant overestimation of the cyclic threshold of the suppressed channel, resulting in the missed detection of weak competitors in mixed infection samples, thus affecting the accuracy of clinical diagnosis. At the same time, if curve distortions caused by other factors such as temperature fluctuations are misjudged as resource competition, it will trigger an incorrect calibration process, further distorting the test results. As a result, the reliability of the system's multiplex detection results cannot be guaranteed, directly affecting the accuracy of diarrhea virus differential diagnosis.

[0016] Example 1: As Figure 1 As shown, a method for detecting diarrhea virus using a centrifugal microfluidic chip includes: Step S1: Obtain the fluorescence intensity sequence of multiple detection channels in real time during thermal cycling, and calculate the fluorescence intensity ratio of adjacent cycles based on the fluorescence intensity sequence as the growth coefficient of the corresponding detection channel. Based on the value of the growth coefficient, identify the event of abrupt change in curve shape and record the event cycle number. During thermal cycling detection, the system first acquires fluorescence intensity data for each detection channel in real time. Specifically, at the end of each thermal cycle, the optical detection module sequentially reads the fluorescence intensity values ​​of all detection channels on the chip, forming a fluorescence intensity sequence corresponding one-to-one with the cycle number. Let there be M detection channels in total, and the original fluorescence intensity acquired by the i-th channel in the n-th cycle be denoted as... The value of n ranges from 1 to N, where N is the preset maximum number of cycles, usually set to 40 or 45.

[0017] To eliminate differences in optical background between chips and baseline drift between channels, the system performs baseline correction processing on the fluorescence intensity sequence of each channel. Specifically, the fluorescence intensity values ​​of the first 5 cycles for each channel are taken, and their arithmetic mean is calculated as the baseline value for that channel. Then, baseline subtraction is performed on the fluorescence intensity for each cycle to obtain the corrected fluorescence intensity. If the calculation result is negative, it will be set to zero to avoid abnormal ratios in subsequent calculations.

[0018] Based on the corrected fluorescence intensity sequence, the system calculates the instantaneous growth coefficient of each channel in each cycle. Growth coefficient Defined as the ratio of corrected fluorescence intensity between adjacent cycles, i.e., for cycle n (n starting from 2), the ratio is calculated. If the denominator If it is zero, then the growth factor of the cycle will be zero. Defined as 1, indicating no growth; when the corrected fluorescence intensity sequence is zero for multiple consecutive cycles, this channel is marked as having no amplification signal, and processing is terminated directly to avoid invalid calculations; this channel does not participate in subsequent curve morphology mutation event identification and directly outputs a negative interpretation result. This growth coefficient During the exponential amplification period, the value stabilizes in the range of 1+E, where E is the amplification efficiency, typically ranging from 0.8 to 1.0; during the plateau period, it approaches 1; when the amplification curve exhibits a morphological abrupt change, the growth coefficient will show a significant decrease.

[0019] The system monitors the growth coefficient sequence of each channel in real time to identify abrupt changes in curve shape. To achieve adaptive detection, the system maintains a sliding window for each channel, with a window length of 5 cycles, storing the growth coefficient values ​​of the current cycle and the previous 4 cycles. For each cycle, the system calculates the statistical characteristics of the data within that window: the median... and interquartile range The interquartile range is calculated by sorting the five values ​​in the window from smallest to largest and taking the difference between the third quartile and the first quartile.

[0020] To adapt to noise fluctuations in different detection scenarios, the system introduces an adaptive triggering mechanism. The system maintains a historical contention fingerprint index variable in memory. This variable stores the competitive fingerprint index calculated after the previous batch of tests, and is initialized to zero upon first run. Based on the numerical range of this historical competitive fingerprint index, the system selects an adaptive trigger coefficient from preset candidate coefficient values. :when When the value is between 0.15 and 0.45, it indicates that there was resource contention in the previous batch of testing, and in this case... Set to 1.5 to improve mutation detection sensitivity; otherwise, Set to 2.0 to maintain detection stability in normal scenarios.

[0021] The mutation event detection mechanism employs a two-cycle triggering method. When the system detects that a channel meets the condition in two consecutive cycles... If the condition is met, then the channel is determined to have experienced a sudden change in curve shape during the nth cycle. Using a two-cycle determination condition here effectively avoids false triggering caused by a single noise spike.

[0022] Once an event is detected, the system immediately records its key information. Event loop count. The sequence number of the second triggering loop is retrieved. Simultaneously, the system calculates the fluorescence decrease corresponding to this event. The value is taken as the difference between the mean corrected fluorescence intensity of the three cycles before the mutation and the corrected fluorescence intensity at the time of the mutation, i.e. This decrease reflects the degree of change in fluorescence intensity before and after the curve abruptly, and is one of the important dimensions for subsequent feature vector construction.

[0023] In addition, the system also acquires auxiliary features related to the event. Physical location encoding. For pre-configured channel attributes, values ​​are assigned based on the physical region of the detection channel on the chip: channel codes for the central region are 0, for the middle region are 1, and for the edge region are 2. This encoding reflects the spatial distribution of channels on the chip, helping to distinguish abnormal patterns caused by different physical locations. Number of associated channels. The statistics are based on a window of 5 cycles before and after the event occurrence cycle, and the total number of events in which the curve shape of other detection channels changes abruptly within these 11 cycle windows is counted.

[0024] Through the above processing, the system has completed all the steps in step S1 and output the corrected fluorescence intensity sequence for each detection channel. Records of all events that cause abrupt changes in curve shape (including event loop counts) fluorescence reduction Physical location coding Number of associated channels and the growth coefficient sequence of each channel. These data form the basis for feature vector construction and competitive fingerprint index calculation in subsequent steps.

[0025] Step S2: Based on the event cycle number of all events that cause abrupt changes in curve shape, the preset physical location code of the detection channel corresponding to each event, and the number of associated channels of each event within the preset cycle window, construct a feature vector set containing event temporal features, spatial features, and associated features. After identifying and recording the abrupt changes in curve morphology, the system enters the feature vector set construction stage. The core objective of this stage is to transform the temporal, spatial, and correlational information contained in each abrupt change event into a unified data format that can be used for subsequent quantitative analysis.

[0026] Specifically, for each curve morphology abrupt change event identified in step S1, the system first extracts its event cycle number. And calculate the normalized time series characteristics of the event. The normalized timing characteristic is defined as the ratio of the number of event loops to the preset maximum number of loops, i.e. Where N is the preset maximum number of cycles, typically set to 40 or 45 according to the testing procedure. This feature maps the absolute number of cycles in which an event occurs to the range of 0 to 1, eliminating the dimensional influence caused by differences in the total number of cycles in different testing batches, thus making subsequent time interval analysis comparable.

[0027] At the same time, the system reads the physical location code of the detection channel corresponding to the event. This encoding is determined and pre-stored in the system configuration file during the chip design phase. Its assignment rules strictly correspond to the physical layout of each channel on the chip: channels located in the central region of the chip are encoded as 0, channels located in the middle region are encoded as 1, and channels located in the edge region are encoded as 2. This encoding provides the basis for subsequently distinguishing anomalies under different spatial distribution patterns.

[0028] Furthermore, the system counts the number of associated channels for each event. The statistical window is defined as a time window of 11 cycles, consisting of 5 cycles before and after the current event cycle. Within this window, the system searches other detection channels to see if curve shape abrupt changes have also occurred, and accumulates the total number of these events. If the same channel undergoes multiple mutations within a window, it is counted only once to avoid duplicate counting. This feature reflects the spatiotemporal clustering of mutation events and is an important basis for distinguishing between resource competition (usually involving multiple channels) and isolated physical anomalies (usually involving only a single channel).

[0029] In addition, the system also directly obtains the fluorescence reduction corresponding to each event from step S1. This feature has already been calculated in the previous steps and does not need to be recalculated here.

[0030] Based on the above four dimensions, the system constructs a four-dimensional feature vector for each event. To facilitate subsequent processing, the system sorts the feature vectors of all events by event loop number. The features are arranged in ascending order to form an ordered sequence of feature vectors. For example, suppose three events are identified in a certain detection, occurring in the 18th, 23rd, and 30th cycles respectively. The corresponding normalized temporal features are 0.45, 0.575, and 0.75, the physical location codes are 0, 2, and 1, the number of associated channels are 1, 2, and 1, and the fluorescence reduction amplitudes are 150, 220, and 180, respectively. Then the system will generate three four-dimensional vectors in sequence: (0.45, 0, 1, 150), (0.575, 2, 2, 220), and (0.75, 1, 1, 180), and save them as a sequence according to the order of event occurrence.

[0031] It is worth noting that during the feature vector construction process, the system does not perform any normalization or scaling on the feature values. This is because the subsequent spatial-temporal correlation index calculation will handle the temporal features and location encoding separately, while the number of correlation channels and fluorescence reduction are used to assist in analysis and decision-making. All features are preserved in their original numerical form to retain the physical meaning of the original data to the greatest extent possible.

[0032] Through the above processing, the system has completed the construction of the feature vector set, forming a structured dataset containing temporal, spatial, correlation, and magnitude information of each mutation event. This set will serve as the direct input for subsequent calculations of the competitive fingerprint index and resource contention mode determination.

[0033] After identifying abrupt changes in curve morphology and constructing feature vectors, the system performs a temperature screening step to eliminate spurious events caused by temperature fluctuations. The core of this step lies in using temperature data from the heating unit to perform orthogonal verification of the events, ensuring that the event sequence used in subsequent competitive fingerprint index calculations truly reflects the kinetic characteristics of the amplification reaction, rather than instantaneous disturbances in the temperature control system.

[0034] At the end of each thermal cycle, the system collects the temperature value of the heating unit through a temperature sensor, forming a temperature sequence that corresponds one-to-one with the cycle number of the fluorescence intensity sequence. Where n is the cycle number, and its value range is consistent with that of fluorescence acquisition. The temperature sensor is installed at the center of the contact surface between the heating unit and the chip, and its sampling frequency is synchronized with the thermal cycle cycle to ensure that a reading representing the current thermal cycle temperature state can be obtained at the end of each cycle.

[0035] For each curve morphology mutation event identified in step S1, the system extracts the cycle number of that event. The temperature values ​​from two cycles before and after, for a total of five cycles, constitute a temperature subsequence. .like Near the boundary of the cyclic sequence (e.g.) or If the actual temperature value is extracted, only the actual temperature value will be extracted. However, to ensure statistical stability, the system requires that there be no fewer than three valid temperature values. Otherwise, the event will be retained directly. In this case, the temperature fluctuation cannot be reliably estimated, so the event is retained to avoid accidental deletion.

[0036] The system calculates the sample variance of the temperature subsequence as the temperature fluctuation of the event. The calculation formula is: Where m is the number of temperature values ​​actually used in the calculation. This is the arithmetic mean of these temperature values. When m=1, the variance is defined as 0. This variance reflects the degree of temperature fluctuation within a short time window before and after the event.

[0037] The system presets a temperature fluctuation threshold. This threshold was obtained through engineering experiments. The specific calibration method was as follows: under standard operating conditions, temperature data were continuously collected for 50 thermal cycles. The temperature variance within two cycle windows before and after each cycle was calculated, and the 95th percentile of these variances was taken as the threshold. Through experimental calibration, this threshold is typically set to 0.5 degrees Celsius squared. When the temperature fluctuation exceeds this threshold, it indicates a significant temperature disturbance at the time of the event, suggesting that the event is likely triggered by instantaneous fluctuations in the temperature control system rather than by kinetic changes in the amplification reaction itself.

[0038] The system will measure the temperature fluctuation for each event. With the preset temperature fluctuation threshold Compare. If If the event occurs, the feature vector corresponding to that event is removed from the feature vector set constructed in step S2; if If an event is not related to temperature fluctuations, its feature vector is retained. After filtering all events one by one, the system obtains a set of filtered feature vectors, which contains only events unrelated to temperature fluctuations.

[0039] For example, suppose an event occurring in cycle 20 is identified in a certain detection. The temperature values ​​extracted from the two cycles before and after are as follows: cycle 18: 95.2 degrees Celsius; cycle 19: 95.3 degrees Celsius; cycle 20: 95.1 degrees Celsius; cycle 21: 95.4 degrees Celsius; cycle 22: 95.2 degrees Celsius. The average of these five temperature values ​​is calculated to be 95.24 degrees Celsius, and the sum of squares of the deviations is (95.2 - 95.24). 2 =0.0016, (95.3-95.24) 2 =0.0036, (95.1-95.24) 2 =0.0196, (95.4-95.24) 2 =0.0256, (95.2-95.24) 2=0.0016, the sum is 0.052, and the sample variance is 0.052 / 4 = 0.013. This value is much less than 0.5, so the event is retained.

[0040] For example, another event occurring in cycle 25 has the following temperature subsequence: cycle 23, 95.8 degrees; cycle 24, 94.2 degrees; cycle 25, 93.5 degrees; cycle 26, 94.0 degrees; cycle 27, 95.5 degrees. The average is calculated to be 94.6 degrees, and the sum of squares of the deviations is (95.8 - 94.6). 2 =1.44, (94.2-94.6) 2 =0.16, (93.5-94.6) 2 =1.21, (94.0-94.6) 2 =0.36, (95.5-94.6) 2 =0.81, the sum is 3.98, and the variance is 3.98 / 4=0.995, which is greater than 0.5. Therefore, the feature vector corresponding to this event is removed.

[0041] Through the above processing, the system effectively eliminates spurious events caused by temperature fluctuations, enabling the subsequent calculation of the competitive fingerprint index to be based on event sequences that truly reflect the dynamic characteristics of the amplification reaction, thereby improving the accuracy and robustness of resource competition pattern recognition. The filtered feature vector set will serve as the direct input for calculating the competitive fingerprint index in step S3.

[0042] Step S3: Calculate the competitive fingerprint index based on the temporal interval consistency among the feature vectors in the feature vector set; After constructing the feature vector set, the system enters the competitive fingerprint index calculation stage. The core objective of this stage is to quantify the regularity of the event sequence by analyzing the temporal consistency of abrupt changes in the curve morphology, thereby providing a quantitative basis for distinguishing resource competition patterns from other abnormal patterns.

[0043] Based on the filtered feature vector set, the system begins to calculate the competitive fingerprint index. The system first sorts the feature vectors in the feature vector set according to their corresponding event loop numbers in ascending order, obtaining a sorted feature vector sequence, denoted as […]. , where K is the total number of events after filtering.

[0044] For each sorted feature vector, the system calculates its spatial-temporal correlation index. The design intent of this index is to couple the temporal information of events with their spatial location information, thereby highlighting the uniformity of event intervals in resource competition scenarios, while disrupting this uniformity in location-related anomaly scenarios. The specific calculation formula is as follows: ,in This represents the normalized time-series feature of the feature vector. Encoding the physical location, The preset spatial weighting coefficient, calibrated experimentally, typically has a value of 0.3. The physical meaning of this formula is: for events in the edge region ( Its spatial-temporal correlation index is amplified. Times; for events in the central area ( The index remains unchanged; for events in the middle region ( The index was amplified. This results in a time-event uniformity that is independent of spatial location when the event sequence is caused by resource contention. After the above transformation, The sequence remains evenly spaced; however, when the event sequence is caused by spatially correlated anomalies (such as edge region preferential failure), the edge events... Magnified, leading to The intervals in the sequence are disordered.

[0045] The system then calculates the sequence of differences between the spatial-temporal correlation indices of adjacent events. For the sorted sequence of feature vectors, the difference between the spatial-temporal correlation indices of two adjacent events is calculated. ,in This difference sequence reflects the interval distribution of events in the coupling space.

[0046] Competitive fingerprint index The coefficient of variation is defined as the ratio of the standard deviation to the mean of the difference series. ,in Difference sequence standard deviation This is the arithmetic mean of the sequence. If all differences in the difference sequence are zero, that is, the sum of all events... If they are equal, then define =0. The coefficient of variation is a dimensionless measure of dispersion, which can eliminate the influence of the mean on the judgment of dispersion, making event sequences of different orders of magnitude comparable; when K=2, the two events are insufficient to assess the consistency of the time interval, the system conservatively judges it as an anomaly, and no calibration is performed.

[0047] For example, suppose a detection, after temperature screening, retains three events with normalized temporal features of 0.45, 0.575, and 0.75, respectively, and physical location codes of 0, 2, and 1, respectively, and spatial weight coefficients... =0.3. Then calculate the spatial-temporal correlation index: =0.45×(1+0.3×0 / 2)=0.45; =0.575×(1+0.3×2 / 2)=0.7475, =0.75×(1+0.3×1 / 2)=0.8625. The difference sequence is... =0.7475-0.45=0.2975; =0.8625 - 0.7475 = 0.1150. Standard deviation of the difference sequence. =0.09125, mean =0.20625, Competitive Fingerprint Index =0.09125 / 0.20625≈0.442.

[0048] Through the above calculations, the system obtains a quantitative indicator that can reflect the regularity of event sequences—the competitive fingerprint index. The smaller the index value, the more uniform the event intervals and the stronger the regularity of the event sequence; the larger the index value, the more chaotic the event intervals and the stronger the randomness of the event sequence. This quantitative indicator will serve as the core basis for subsequent resource competition calibration mode determination.

[0049] After completing temperature screening and calculating the competition fingerprint index, the system further identifies the disturbance sources of events where temperature fluctuations exceed a preset temperature fluctuation threshold, in order to refine the diagnosis of the specific causes of temperature anomalies. This identification process utilizes the spatial distribution characteristics of temperature-related events and competition background information to attribute temperature disturbances to local contact thermal resistance problems or temperature fluctuations under resource competition, providing a more accurate diagnostic basis for system status monitoring.

[0050] The system first acquires the events that were rejected in the temperature screening step, i.e., events whose temperature fluctuations exceed a preset temperature fluctuation threshold, and constructs these events into a temperature-related event set. For each event in this set, the system reads the physical location code of its corresponding detection channel. The system also counts the frequency of occurrence of codes for different physical locations within the set of temperature-related events. Simultaneously, it records the competitive fingerprint index of events that were not eliminated from the original feature vector set constructed in step S2; this index was calculated in step S3.

[0051] The system analyzes the spatial distribution patterns of temperature-related events. A spatially concentrated distribution pattern is defined as follows: in a set of temperature-related events, events exceeding a preset proportion threshold are concentrated within a channel region corresponding to a specific physical location code. Specifically, the system calculates edge region codes ( The system determines the spatially concentrated distribution pattern by the proportion of events in each physical location encoding. If this proportion is greater than a preset concentration ratio threshold, such as 80%, the distribution is determined to be spatially concentrated. Conversely, if the proportion of events in each physical location encoding does not exceed the threshold and the distribution is relatively uniform, the distribution is determined to be spatially uniform. It should be noted that when the total number of temperature-related events is less than the preset minimum number of statistics (e.g., 3), the system will skip the temperature disturbance source identification step and directly retain all events for subsequent processing because the sample size is too small to perform reliable distribution pattern analysis. In this embodiment, the concentration ratio threshold is 80%, the minimum number of statistics is 3, and the synchronization window threshold is 5 cycles.

[0052] When the temperature disturbance is determined to be a spatially concentrated distribution pattern, it indicates that the temperature disturbance is mainly concentrated in a specific physical area of ​​the chip. This is consistent with the characteristics of local contact thermal resistance anomalies, because contact thermal resistance is usually affected by the uniformity of the bonding between the chip and the heating unit, and is often more prominent in the edge areas. In this case, the system determines it as a local contact thermal resistance disturbance and outputs a local temperature disturbance label.

[0053] When the distribution pattern is determined to be spatially uniform, it indicates that temperature disturbances are prevalent across all areas of the chip, requiring further analysis based on the competitive context. The system acquires the competitive fingerprint index of the set of events that were not eliminated. And determine whether the index is within the preset competition index range, i.e., 0.15 to 0.45. If If the temperature disturbance is within this range, it indicates that there is a resource contention background in the current testing batch. At this time, temperature disturbance and resource contention occur simultaneously. The system determines it as a temperature disturbance under resource contention background and outputs a label indicating that the disturbance is accompanied by contention. If the temperature is not within this range, it is determined to be another type of global temperature disturbance, and the process proceeds to the subsequent global temperature disturbance type subdivision step.

[0054] After initially determining the spatially uniform distribution pattern, the system performs a global temperature disturbance type subdivision step to distinguish between temperature control system oscillation disturbances and ambient temperature disturbances. The core of this step lies in analyzing the temporal synchronicity characteristics of temperature-related events and the periodic characteristics of temperature fluctuations.

[0055] The system acquires the temperature sequence corresponding to each event in the set of temperature-related events. For each event, it extracts the temperature values ​​of five cycles, two cycles before and two cycles after its event cycle number, forming the temperature subsequence for that event. The system calculates the temporal synchronization index of all temperature-related events. Specifically, it constructs a sequence from the event cycle numbers of all temperature-related events and calculates the difference between the cycle numbers of adjacent events in this sequence. If the maximum value of these differences is less than a preset synchronization window threshold, typically set to five cycles, it indicates that the temperature-related events occur concentratedly within a short cycle window, and this is determined to be a global synchronization mode.

[0056] When the system is determined to be in global synchronization mode, it indicates that temperature disturbances affect multiple cycles within a short period of time. This may be caused by periodic oscillations of the temperature control system or sudden changes in ambient temperature. The system further analyzes the periodic characteristics of temperature fluctuations, obtains the temperature subsequence corresponding to the earliest occurrence in the set of temperature-related events, and performs fluctuation period analysis on this sequence. The specific method is as follows: calculate the sign change of the difference between adjacent temperature values ​​in the temperature subsequence, and count the length of continuous unidirectional changes. If the temperature value shows a regular alternation of rising and falling and the period length matches the control cycle of the heating unit (if the absolute difference between the period length of the temperature fluctuation and the preset control cycle of the heating unit (e.g., 2 cycles) is less than 0.5 cycles, it is considered a match), the control cycle of the heating unit is usually an integer fraction of the thermal cycle cycle. If it is two or three cycles, it is determined to be a temperature control system oscillation disturbance, and a temperature control oscillation label is output. If the temperature fluctuation has no obvious periodicity or the period length does not match the control cycle of the heating unit, it is determined to be an ambient temperature disturbance, and an ambient disturbance label is output.

[0057] For example, suppose that in a certain detection process, the temperature screening step removes three events, occurring in loops 12, 14, and 16, with corresponding channel physical location codes of 2, 2, and 1, respectively. The calculated proportion of events in the edge region is two-thirds, approximately 66.7%, which does not reach the 80% concentration threshold, but the distribution is still biased towards the edges, requiring further consideration of specific threshold settings. If the preset concentration threshold is 70%, then this distribution does not meet the concentration pattern and is determined to be a uniform distribution pattern. The system obtains a competition fingerprint index of 0.32 for the events not removed, which is within the competition index range; therefore, it outputs a competition-accompanied temperature perturbation label.

[0058] For example, in another scenario, the temperature screening step removes four events, occurring in cycles 20, 21, 22, and 23, respectively. The physical location codes for the corresponding channels are 0, 1, 1, and 2, respectively, showing a uniform distribution. The maximum difference in cycle number for these events is 3, which is less than the synchronization window threshold of 5, thus indicating a global synchronization mode. Taking the temperature subsequence corresponding to the earliest event, cycle 20, the temperature values ​​are 95.2, 95.3, 94.9, 94.8, and 95.1, exhibiting a fluctuation of first slightly increasing, then decreasing, and then increasing again. The fluctuation period is approximately two cycles, consistent with the PWM control cycle of the heating unit. Therefore, it is determined to be an oscillation disturbance in the temperature control system, and a temperature control oscillation label is output.

[0059] Through the above steps of identifying temperature disturbance sources and subdividing global temperature disturbance types, the system further attributes the originally simple temperature anomaly events and outputs refined diagnostic labels, providing operators with a clear basis for troubleshooting system status or subsequent adaptive compensation.

[0060] Step S4: Based on the matching relationship between the competitive fingerprint index of the feature vector set and the preset range, determine whether to enter the resource competition calibration mode. When entering the resource competition calibration mode, mark the detection channel as a strong competitor group or a suppressed group based on the spatial-temporal correlation index of the feature vectors in the feature vector set. After calculating the competitive fingerprint index, the system enters the stage of determining the resource competition calibration mode and grouping channels. This stage comprehensively evaluates the statistical characteristics of the event sequence based on the filtered feature vector set and its competitive fingerprint index to determine whether resource competition exists in the current detection batch, and groups and marks the channels participating in the competition.

[0061] The system first obtains key information from the filtered feature vector set, including the spatial-temporal correlation index of each feature vector. and the competitive fingerprint index calculated through step S3 At the same time, the system counts the total number of feature vectors after filtering, denoted as K.

[0062] The system classifies the current detection state according to preset judgment rules. These preset rules contain multiple sets of conditional branches, each corresponding to a different detection scenario: When the total number K of the filtered feature vectors meets the preset quantity condition, the system determines that it will not enter the resource contention calibration mode. Specifically, this quantity condition is set to K less than 2. That is, when the total number of filtered events is less than 2, it is impossible to form a valid event sequence for time interval consistency analysis. In this case, it is directly determined to be in the no-event mode, and the subsequent calibration process is not executed.

[0063] When the total number of filtered feature vectors K is greater than or equal to 2, the system further uses the competitive fingerprint index. The system determines the index based on its numerical range. Three index ranges are preset: the competition index range, the anomaly index range, and the physical anomaly index range. The competition index range corresponds to the typical value range of the competition fingerprint index in resource contention scenarios. Calibrated through extensive experimental data, this range has a lower limit of 0.15 and an upper limit of 0.45. When the event falls within this range, it indicates that the temporal interval consistency of the event sequence conforms to the typical characteristics of resource contention, and the system determines that it has entered the resource contention calibration mode.

[0064] when When the intervals of the event sequence are less than the lower limit of the competition index range (i.e., less than 0.15) and the total number of feature vectors K after screening is greater than or equal to 2, it indicates that the intervals of the event sequence are too uniform and exceed the normal fluctuation range of resource competition. This may be related to global system anomalies (such as the overall failure of the temperature control system). At this time, the system is judged to be in a global system anomaly mode and does not enter the resource competition calibration mode.

[0065] when When the value is greater than the upper limit of the competition index range (i.e., greater than 0.45) or the total number of filtered feature vectors K equals 1, it indicates that the interval of the event sequence is severely disordered or there are only isolated events, which may be related to local physical anomalies (such as bubbles or leaks in a single channel). In this case, the system determines it to be a local physical anomaly mode and does not enter the resource competition calibration mode.

[0066] After determining that the system has entered the resource contention calibration mode, it performs channel grouping. The system obtains the spatial-temporal correlation index of each feature vector in the filtered feature vector set. and sort all feature vectors according to The sequence is sorted from smallest to largest. In the sorted sequence, the feature vectors at the beginning correspond to channels where the event occurred earlier or where the spatial location is more central, while the feature vectors at the end correspond to channels where the event occurred later or where the spatial location is more peripheral.

[0067] The system divides the sorted feature vectors into two groups according to a preset grouping ratio. This preset ratio is set to 50% for each group; that is, the detection channels corresponding to feature vectors in the first half of the sorted sequence are labeled as the strong competitor group, and the detection channels corresponding to feature vectors in the second half are labeled as the suppressed group. When the total number of feature vectors K is odd, the first half is selected. As a group of strong competitors, the later One is designated as the suppressed group, and the remaining middle one can be assigned to one of the groups or used as a transition group depending on the specific design. In this embodiment, to simplify the process, the middle one is assigned to the strong competitor group.

[0068] For example, continuing from the example in step S3, suppose that after temperature filtering, three events are retained, with spatial-temporal correlation indices of 0.45, 0.7475, and 0.8625, respectively, corresponding to channel numbers A, B, and C. The competition fingerprint index... The value is approximately 0.442, falling within the competition index range of 0.15 to 0.45. Therefore, the system determines that it has entered the resource competition calibration mode. Subsequently, the system sorts the three events according to their spatial-temporal correlation index from smallest to largest, resulting in the sorted event order as channel A (0.45), channel B (0.7475), and channel C (0.8625). Following a 50% grouping ratio, the first half (channel A corresponding to the first event) is marked as the strong competitor group, the second half (channel C corresponding to the third event) is marked as the suppressed group, and the second event, channel B, is placed in the middle and included in the strong competitor group in this embodiment. Ultimately, the strong competitor group includes channels A and B, and the suppressed group includes channel C.

[0069] In another scenario, if the total number of events after filtering is K=4, the spatial-temporal correlation indices are 0.42, 0.56, 0.71, and 0.89 respectively, and the competitive fingerprint index... =0.32, which is within the competition index range, so the system determines that it has entered the resource competition calibration mode. After sorting, the channels corresponding to the first two events are marked as the strong competitor group, and the channels corresponding to the last two events are marked as the suppressed group.

[0070] Through the above-mentioned judgment and grouping process, the system completed the identification of resource competition patterns and the division of strong competitors and suppressed groups, laying the foundation for channel grouping for subsequent calibration of detection results for suppressed group channels.

[0071] Step S5: The detection results of the suppressed group are calibrated by calculating the degree of linear correlation and extrapolating linear regression.

[0072] After determining the resource contention calibration mode and grouping the channels, the system enters the calibration phase for the suppressed group detection results. The core objective of this phase is to reconstruct the complete amplification curve of the suppressed channel using linear regression extrapolation, based on the undisturbed exponential growth segment before distortion, thereby obtaining the cyclic threshold after eliminating contention interference.

[0073] For each detection channel in the suppressed group, the system first acquires the baseline-corrected fluorescence intensity sequence for that channel. and the number of event loops recorded by the channel in step S1. The system calculates the baseline noise threshold based on the fluorescence intensity of the first 10 cycles of this channel. The specific method is as follows: calculate the arithmetic mean of the corrected fluorescence intensity for the first 10 cycles. and sample standard deviation The baseline noise threshold is defined as follows: This threshold reflects the upper limit of normal fluctuations in system optical noise and chip background fluorescence before the amplification reaction has begun.

[0074] The system iterates through the fluorescence intensity sequence of this channel starting from the second cycle to determine the peak cycle number. The specific decision logic is as follows: find the first condition that satisfies... Furthermore, the fluorescence intensity in the subsequent three consecutive cycles was greater than [a certain value]. The cycle number is determined as the peak-starting cycle number. If no cycle satisfying the above conditions is found in the entire cycle sequence, it indicates that the channel has not shown effective amplification during the entire thermal cycling process. The system abandons the calibration of the channel, directly outputs a negative interpretation result, and adds a "no peak" label.

[0075] After successfully locating the peak cycle number, the system extracts the fluorescence intensity sequence from the peak cycle number to the cycle preceding the event cycle number, i.e., the cycle number interval. All inside Value. Let the length of the sequence be L. If L < 3, it indicates that there are too few data points available for fitting, the linear regression result is unreliable, the system abandons calibration and marks it as "insufficient data"; when L equals 3, the three data points are completely collinear, and the correlation coefficient must be 1. At this time, although the fitted line is mathematically valid, due to the low degree of freedom, the extrapolation result may have a large uncertainty. The system marks the calibration result as 'low confidence calibration' to prompt the user; the 'low confidence' label is added to the output result for clinical decision reference.

[0076] The system then calculates the linear correlation between the fluorescence intensity sequence and the cycle number, specifically using the Pearson correlation coefficient. Let the extracted cycle number sequence be... Where L is the sequence length of the cycle number sequence, and the corresponding natural logarithm sequence of fluorescence intensity is... The formula for calculating the Pearson correlation coefficient r is: ;in and These are the arithmetic mean of the cycle number sequence and the natural logarithm sequence, respectively. This correlation coefficient reflects the strength of the linear relationship between the logarithmic growth trend of fluorescence intensity and the number of cycles, ranging from -1 to 1. In the ideal exponential amplification phase, this value should be close to 1. The system presets a correlation coefficient threshold of 0.95. If r < 0.95, it indicates that the exponential growth pattern of this sequence segment is not ideal, and there may be other interfering factors. The system abandons calibration and marks it as "unideal exponential growth".

[0077] When the correlation coefficient meets the preset conditions, the system performs linear regression fitting on the sequence segment. Let the fitting equation be... Where b is the slope, reflecting amplification efficiency; and a is the intercept, reflecting initial fluorescence intensity. The regression coefficients a and b are calculated using the least squares method: , The system then presets a calibration fluorescence threshold. The value is taken as 1.5 times the baseline noise threshold, that is... Substitute this threshold into the fitting equation to solve for the calibrated cyclic threshold. : .

[0078] To ensure the physical validity of the calculation results, the system... Perform a validity check. The preset validity conditions include: It must be within a reasonable number of cycles, that is , where N is the preset maximum number of loops. If If the value exceeds this range, abandon the calibration and mark it as "calibration value abnormal"; otherwise, proceed with the calibration. As the final test result for this channel, a "competitive calibration" label is added to the output.

[0079] To facilitate understanding, a specific numerical example will be provided below. Assume the event loop number of a certain suppressed group channel. =28, the mean fluorescence intensity of the first 10 cycles is 120, and the standard deviation is 15. Therefore, the baseline noise threshold is... =120 + 3 × 15 = 165. A review of the fluorescence sequence revealed that the fluorescence intensity first exceeded 165 in the 18th cycle and remained above 165 for the next three consecutive cycles. Therefore, the peak cycle number is... =18. Ten fluorescence intensity values ​​were extracted from cycles 18 to 27. The correlation coefficient (r) between their natural logarithm and the cycle number was calculated to be 0.98, which is greater than 0.95, satisfying the condition. Linear regression yielded a = 2.1 and b = 0.25. The fluorescence threshold was then calibrated. =1.5×165=247.5, substitute and solve. = (ln(247.5)-2.1) / 0.25≈(5.51-2.1) / 0.25=13.64. This value is valid in the range of 0 to 40, so the system outputs the calibrated loop threshold of this channel as 13.64.

[0080] Through the above processing, the system completed the calibration of the detection results of each channel in the suppressed group, effectively eliminating the distortion effect of shared resource competition on the amplification curve, and improving the quantitative accuracy of multiplex detection in mixed infection scenarios.

[0081] Through the above steps, this invention achieves accurate identification and effective calibration of shared resource competition scenarios in multiplex detection of diarrhea viruses. Specifically, the system first adaptively identifies curve morphology mutation events based on a continuous two-cycle triggering mechanism of the growth coefficient, and constructs a four-dimensional feature vector by combining physical location encoding and the number of associated channels, transforming the temporal, spatial, and correlation information of mutation events into structured data. Then, a temperature screening step is used to eliminate false events caused by temperature control fluctuations, and a competition fingerprint index is calculated based on the spatial-temporal correlation index of the event sequence to quantify the consistency of event intervals, thereby addressing resource competition, global system anomalies, and local physical... A clear distinction boundary is established between anomalies. Based on this, the system only performs linear regression extrapolation calibration on the suppressed group channels identified as being in resource competition scenarios, using the exponential growth segment that was not disturbed before distortion, thus eliminating the impact of competition interference on the cyclic threshold. Therefore, this invention solves the technical problem that the prior art cannot distinguish between different causal curve distortions and cannot effectively compensate for resource competition scenarios, significantly improving the quantitative accuracy of multiplex detection in mixed infection high-load samples. At the same time, through temperature disturbance source identification and global disturbance type subdivision, it provides a refined basis for system status monitoring and fault diagnosis, ensuring the reliability and interpretability of the detection results.

[0082] Example 2: Figure 2 As shown, a multiplex detection system for diarrhea viruses using a centrifugal microfluidic chip includes a processor, which is communicatively connected to: Mutation event detection module: acquires fluorescence intensity sequences collected in real time from multiple detection channels during thermal cycling, calculates the fluorescence intensity ratio of adjacent cycles based on the fluorescence intensity sequences as the growth coefficient of the corresponding detection channel, identifies events that cause abrupt changes in curve shape based on the value of the growth coefficient, and records the event cycle number of the event; Feature Analysis Module: Based on the event cycle number of all events that cause abrupt changes in curve shape, the preset physical location code of the detection channel corresponding to each event, and the number of associated channels of each event within the preset cycle window, a feature vector set containing event temporal features, spatial features, and associated features is constructed. Temperature screening module: Uses temperature data from the heating unit to perform orthogonal verification of events, ensuring that the event sequence used in subsequent competitive fingerprint index calculations truly reflects the dynamic characteristics of the amplification reaction, rather than the instantaneous disturbances of the temperature control system.

[0083] Competition analysis module: Calculates the competition fingerprint index based on the temporal interval consistency among the feature vectors in the feature vector set; Global Disturbance Classification Module: Utilizing the spatial distribution characteristics of temperature-related events and competitive background information, temperature disturbances are attributed to local contact thermal resistance problems or temperature fluctuations under the background of resource competition.

[0084] Pattern discrimination module: Based on the matching relationship between the competitive fingerprint index of the feature vector set and the preset range, determine whether to enter the resource competition calibration mode. When entering the resource competition calibration mode, the detection channel is marked as a strong competitor group or a suppressed group based on the spatial-temporal correlation index of the feature vectors in the feature vector set. Calibration processing module: The detection results of the suppressed group are calibrated by calculating the degree of linear correlation and extrapolating linear regression.

[0085] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0086] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0087] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for detecting diarrhea virus using a centrifugal microfluidic chip, characterized in that, include: The fluorescence intensity sequences of multiple detection channels are acquired in real time during thermal cycling, and the fluorescence intensity ratio of adjacent cycles is calculated based on the fluorescence intensity sequences as the growth coefficient of the corresponding detection channel. Based on the value of the growth coefficient, events that cause sudden changes in curve shape are identified, and the event cycle number of the event is recorded. Based on the event cycle number of all events that cause abrupt changes in curve shape, the preset physical location code of the detection channel corresponding to each event, and the number of associated channels of each event within the preset cycle window, a feature vector set containing event temporal features, spatial features, and associated features is constructed. Based on the consistency of the temporal intervals among the feature vectors in the feature vector set, the competitive fingerprint index is calculated. Based on the matching relationship between the competitive fingerprint index of the feature vector set and the preset range, it is determined whether to enter the resource competition calibration mode. When entering the resource competition calibration mode, the detection channel is marked as a strong competitor group or a suppressed group based on the spatial-temporal correlation index of the feature vectors in the feature vector set. The detection results of the suppressed group were calibrated by calculating the degree of linear correlation and extrapolating linear regression.

2. The method for detecting diarrhea virus using a centrifugal microfluidic chip according to claim 1, characterized in that, Based on the numerical identification of the growth coefficient, events causing abrupt changes in the curve shape are specifically included: For each detection channel, the growth coefficients of the current cycle and the preset number of consecutive cycles before the current cycle are obtained to form a sliding window data set, and the median and interquartile range of this data set are calculated. Obtain the historical competitive fingerprint index stored after the previous batch of testing is completed, and select one from multiple preset candidate coefficient values ​​as the adaptive trigger coefficient based on the preset value range in which the historical competitive fingerprint index is located. When the growth coefficient of two consecutive cycles is less than the median minus the product of the adaptive triggering coefficient and the interquartile range, the channel is determined to have experienced a sudden change in curve shape in the current cycle, and the sequence number of the cycle is recorded as the event cycle number. At the same time, the difference between the average fluorescence intensity of the preset number of consecutive cycles before the current cycle and the fluorescence intensity of the current cycle is recorded as the fluorescence decrease.

3. The method for detecting diarrhea virus using a centrifugal microfluidic chip according to claim 1, characterized in that, Construct a feature vector set that includes event temporal features, spatial features, and correlation features, specifically including: The ratio of the event loop number to the preset maximum loop number for each event is used as the normalized temporal feature of that event. Obtain the physical location code of the detection channel corresponding to the event. This physical location code is pre-set according to the physical area of ​​the detection channel on the chip, and different physical areas correspond to different code values. The total number of events in which other detection channels exhibit abrupt changes in curve shape within each preset number of loop windows before and after the occurrence of the event is counted, and this number is taken as the number of associated channels for the event. The normalized temporal features, physical location encoding, number of associated channels, and fluorescence reduction of the event are combined into a four-dimensional feature vector to form a feature vector set.

4. The method for detecting diarrhea virus using a centrifugal microfluidic chip according to claim 1, characterized in that, A temperature screening step is included before calculating the competitive fingerprint index: The temperature sequence collected in real time during the thermal cycle of the heating unit is obtained, and the temperature sequence corresponds one-to-one with the cycle number of the fluorescence intensity sequence; For each event in which the curve shape changes abruptly, extract the temperature values ​​corresponding to a preset number of consecutive cycles before and after the event cycle number, and calculate the variance of these temperature values ​​as the temperature fluctuation of the event. The temperature fluctuation amount is compared with a preset temperature fluctuation threshold. When the temperature fluctuation amount is greater than the preset temperature fluctuation threshold, the feature vector corresponding to the event is removed from the feature vector set to obtain a filtered feature vector set.

5. The method for detecting diarrhea virus multiplex centrifugation microfluidic chip according to claim 4, characterized in that, Based on the temporal interval consistency among the feature vectors in the aforementioned feature vector set, a competitive fingerprint index is calculated, specifically including: The feature vectors in the filtered feature vector set are sorted in ascending order according to their corresponding event loop number to obtain the sorted feature vector sequence. For each feature vector in the sorted feature vector sequence, the spatial-temporal correlation index is calculated based on its normalized temporal features and physical location code. The calculation method is as follows: divide the physical location code by two, multiply by the preset spatial weight coefficient, add one to the result, and finally multiply by the normalized temporal features. Calculate the difference between the spatial-temporal correlation indices of two adjacent feature vectors after sorting to obtain the difference sequence; Calculate the standard deviation and mean of all differences in the difference sequence. Divide the standard deviation by the mean and use the ratio as the competitive fingerprint index. When all differences in the difference sequence are zero, set the competitive fingerprint index to zero.

6. The method for detecting diarrhea virus multiplex centrifugation microfluidic chip according to claim 5, characterized in that, The process for determining the resource contention calibration mode and marking the detection channel as either a strong competitor or a suppressed group when entering the resource contention calibration mode specifically includes: Obtain the spatial-temporal correlation index of each feature vector in the filtered feature vector set and the competitive fingerprint index of the filtered feature vector set; When the total number of feature vectors in the filtered feature vector set meets the preset quantity condition, it is determined that the resource competition calibration mode will not be entered. When the competitive fingerprint index of the selected feature vector set is within the preset competitive index range, it is determined to enter the resource competition calibration mode, and the feature vectors are sorted according to the spatial-temporal correlation index. The detection channels corresponding to the sorted feature vectors are marked as strong competitors and suppressed groups according to the preset ratio. When the competitive fingerprint index of the filtered feature vector set is within the preset abnormal index range and the total number of feature vectors meets the preset quantity condition, it is determined to be a global system abnormality and will not enter the resource competition calibration mode. When the competitive fingerprint index of the filtered feature vector set is within the preset physical anomaly index range or the total number of feature vectors meets the preset scarcity condition, it is determined to be a local physical anomaly and will not enter the resource competition calibration mode.

7. The method for detecting diarrhea virus using a centrifugal microfluidic chip according to claim 1, characterized in that, The detection results of the suppressed group were calibrated by calculating the degree of linear correlation and extrapolating linear regression, specifically including: For each detection channel in the suppressed group, a baseline noise threshold is calculated based on the fluorescence intensity of a preset number of consecutive cycles preceding that channel; Starting from the second cycle, find the first cycle that satisfies the condition that the fluorescence intensity is greater than the baseline noise threshold and the fluorescence intensity of subsequent cycles for a preset number of consecutive cycles is greater than the baseline noise threshold, and determine this cycle as the peak cycle number of this channel; Obtain the fluorescence intensity sequence from the peak cycle number to the cycle before the event cycle number, calculate the linear correlation between the fluorescence intensity sequence and the cycle number, and abandon the calibration of the channel if the linear correlation does not meet the preset correlation condition. When the linear correlation meets the preset correlation condition, a linear regression fit is performed on the fluorescence intensity sequence and the cycle number to obtain the linear regression equation. The preset calibration fluorescence threshold is substituted into the linear regression equation to obtain the calibrated cyclic threshold. If the calibrated cyclic threshold does not meet the preset validity condition, the calibration of the channel is abandoned; otherwise, the calibrated cyclic threshold is used as the detection result of the channel.

8. The method for detecting diarrhea virus multiplex centrifugation microfluidic chip according to claim 5, characterized in that, It also includes a step for identifying temperature disturbance sources: The events in the temperature screening step where the temperature fluctuation exceeds a preset temperature fluctuation threshold are collected as a set of temperature-related events. The physical location code of the detection channel corresponding to each event in the set of temperature-related events is then obtained. When the distribution of events in the temperature-related event set conforms to a preset spatial concentrated distribution pattern in terms of physical location encoding, it is determined to be a local contact thermal resistance disturbance, and a local temperature disturbance label is output. When the events in the temperature-related event set conform to a preset spatial uniform distribution pattern in terms of physical location encoding, the competition fingerprint index of the events in the event set that have not been removed is obtained. Based on the matching relationship between the competition fingerprint index and the preset index interval, it is determined whether the temperature disturbance belongs to the background of resource competition.

9. The method for detecting diarrhea virus using a centrifugal microfluidic chip according to claim 8, characterized in that, It also includes a global temperature perturbation type subdivision step: After determining that the events in the temperature-related event set conform to a spatially uniform distribution pattern, the temporal synchronization characteristics of the temperature fluctuation of each event in the temperature-related event set are obtained. When the synchronization feature meets the preset global synchronization mode, the fluctuation period feature of the temperature sequence corresponding to the event in the temperature-related event set is obtained; Based on the matching relationship between the fluctuation cycle characteristics and the preset working mode of the heating unit, the output temperature control system oscillation disturbance label or the ambient temperature disturbance label is distinguished.

10. A multiplex detection system for diarrhea viruses using a centrifugal microfluidic chip, characterized in that, Includes a processor, which executes the diarrhea virus multiplex detection centrifugal microfluidic chip detection method according to any one of claims 1-9 when running.

Images