Inflammatory bowel disease recurrence early warning method and system based on excrement calprotectin

Abstract

The invention relates to the technical field of biomedical detection, in particular to an inflammatory bowel disease recurrence early warning method and system based on excrement calprotectin, and the method comprises the following steps: collecting three excrement samples of the same defecation event, splitting and extracting, carrying out immunodetection under different dilution rates to obtain point concentration, and calculating the concentration of the excrement calprotectin; forming an observation data packet containing a dilution consistency index, immunoreaction kinetics uncertainty and spatial heterogeneity uncertainty; synthesizing a measurement variance, calculating an observation gain, updating an inflammatory activity hidden state, and generating a trend slope, an accumulated offset and an individual dynamic threshold value; generating an event code according to the state data to trigger anti-certification retest so as to output an anti-certification result and a detection design instruction; and recursively accumulating evidences based on the state and the anti-evidence result, outputting low / medium / high risk early warning according to the boundary, and updating individual parameters after obtaining a recurrence outcome label. According to the method, the early warning stability and interpretability are improved, false alarm and missing alarm are reduced, and the retest cost is optimized.

Description

Technical Field

[0001] This invention relates to the field of biomedical detection technology, and in particular to a method and system for early warning of recurrence of inflammatory bowel disease based on fecal calprotectin. Background Technology

[0002] Inflammatory bowel disease (IBD) is characterized by its persistent and recurrent nature, with relapses driving cumulative intestinal mucosal damage. Delayed intervention after relapse can easily lead to worsening of the condition, increased hospitalization and surgical risks, and significantly increased follow-up and medication costs. The industry's demand for "home-based, high-frequency, and quantifiable" monitoring continues to rise to support adjustments in treatment strategies, follow-up appointments, and adherence management. However, current monitoring methods largely rely on single test results or fixed thresholds, making it difficult to adequately address uncertainties arising from spatial heterogeneity of fecal samples, differences in dilution factors, fluctuations in the immune response process, and noise from the testing platform. Furthermore, the lack of verification mechanisms for abnormal results and individualized dynamic thresholds and evidence-based reasoning mechanisms results in insufficient stability and interpretability of early warnings, leading to a coexistence of false positives, false negatives, and the cost of repeated testing. Summary of the Invention

[0003] This invention provides a method and system for early warning of relapse of inflammatory bowel disease based on fecal calprotectin, which is used to at least solve the problem of how to achieve verifiable, interpretable and adaptive relapse grading early warning in the context of high-frequency fecal testing at home, under the conditions of sample heterogeneity and detection uncertainty.

[0004] In a first aspect, the present invention provides a method for early warning of recurrence of inflammatory bowel disease based on fecal calprotectin, comprising the following steps: Three fecal samples were collected from the same defecation event and lysed to obtain extracts. Immunoassay was performed on the extracts at different dilution ratios to obtain the concentration of fecal calprotectin at the site, and an observation data package containing dilution consistency index, immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty was generated. Based on the observation data packet, the measurement variance is synthesized and the observation gain is calculated. Based on the observation gain, the latent state of inflammatory activity is updated and the trend slope, cumulative offset and individual dynamic threshold are generated to form a state data packet. Event codes are generated based on state data packets, and the disproving process is executed based on the event codes to generate disproving result data packets and detection design instructions; Evidence is accumulated based on state data packets and counter-evidence result data packets to output graded early warning results, and the individual parameter set is updated after obtaining the relapse outcome label.

[0005] Secondly, the present invention provides an early warning system for recurrence of inflammatory bowel disease based on fecal calprotectin, for implementing an early warning method for recurrence of inflammatory bowel disease based on fecal calprotectin, the system comprising: The observation generation module is used to collect fecal samples from three locations during the same defecation event, lyse and extract the extract, perform immunoassay on the extract at different dilution ratios to obtain the concentration of fecal calprotectin at the site, and generate an observation data package containing dilution consistency index, immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty. The state update module is used to synthesize measurement variance and calculate observation gain based on observation data packets, update the latent state of inflammatory activity based on observation gain, and generate trend slope, cumulative offset and individual dynamic threshold to form state data packets. The disproving trigger module is used to generate event codes based on the status data packet and execute the disproving process based on the event codes to generate a disproving result data packet and detection design instructions. The evidence warning module is used to accumulate evidence based on status data packets and counter-evidence result data packets to output graded warning results, and to update the individual parameter set after obtaining the relapse outcome label.

[0006] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: By employing three-point sampling and multi-dilution immunoassay, and constructing dilution consistency indices, immune response kinetic uncertainty, and spatial heterogeneity uncertainty, explicit quantification of sample and detection process uncertainties was achieved. By synthesizing measurement variance and mapping it to observation gain, adaptive updates of latent inflammatory activity and generation of individual dynamic thresholds were realized. A rebuttal process involving dilution retesting, kinetic retesting, occult blood retesting, spiked recovery, and site expansion triggered by anomaly type event codes enabled closed-loop verification of suspected anomalies and output of detection design instructions. Through evidence recursion of state data and rebuttal results, and boundary grading, stable and interpretable low / medium / high risk early warnings were achieved. Continuous self-learning for individuals was realized through relapse outcome label calibration threshold rules, boundary and platform noise. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the execution flow of the method of the present invention; Figure 2 This is a schematic diagram illustrating the changes in accumulated evidence and hierarchical boundaries over time in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the changes in measurement variance and observation gain over time in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the changes over time in event-level fecal calprotectin observations, latent inflammatory activity, and individual dynamic thresholds in an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the changes in accumulated evidence and hierarchical boundaries over time in an embodiment of the present invention; Figure 6 This is a structural block diagram of the method system of the present invention. Detailed Implementation

[0008] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation.

[0009] Fecal calprotectin, a neutrophil-associated protein, enters the intestinal lumen with inflammatory exudate and is excreted in feces during intestinal mucosal inflammatory infiltration, thus reflecting the level of intestinal inflammatory activity to some extent. Compared to assessment methods that rely solely on symptom complaints or occasional re-examinations, fecal calprotectin offers advantages such as relatively convenient sampling, suitability for repeated testing, and ease of generating individual longitudinal sequences, allowing for the quantification of changes in intestinal inflammation as "time-varying observations." However, in practical applications, the reading of this indicator can be affected by factors such as uneven sample location, fecal matrix interference, dilution ratio selection, and fluctuations in the immune response process, making it difficult to directly equate a single result with a stable risk conclusion. Therefore, this invention constructs a mechanism for observational uncertainty and counter-evidence verification based on multi-site sampling and multi-dilution immunoassay of the same defecation event, and performs state updates and evidence accumulation on the individual longitudinal sequence, thereby forming an interpretable and calibrable relapse grading and early warning process.

[0010] like Figure 1 As shown, a method for early warning of relapse of inflammatory bowel disease based on fecal calprotectin includes the following steps: Three fecal samples were collected from the same defecation event and lysed to obtain extracts. Immunoassay was performed on the extracts at different dilution ratios to obtain the concentration of fecal calprotectin at the site, and an observation data package containing dilution consistency index, immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty was generated. Following the occurrence of the same defecation event, three fecal sampling sites were selected and numbered according to a pre-defined sampling rule. Fecal samples from each site were obtained using a quantitative sampler and placed into corresponding lysis containers. Lysis buffer was added to each lysis container, and the samples were mixed, incubated for an incubation period, and centrifuged. The supernatant was used as the extraction solution. At least two dilution ratios were set for the extraction solution, and immunoassays were performed on each. During the immunoassay process, the detection device acquired the immunoreaction time-series signal, and the signal was converted into fecal calprotectin concentration at each site based on a calibration curve. The dilution consistency index was used to characterize the consistency of fecal calprotectin concentration at different dilution ratios, and was calculated using the following expression: ; in, To dilute the consistency index, This represents the fecal calprotectin concentration at the location corresponding to the first dilution ratio. The concentration of fecal calprotectin at the site corresponding to the second dilution factor is given. The uncertainty in immune response kinetics is obtained from the dispersion of the fitting residuals of the immune response time series signal, and the spatial heterogeneity uncertainty is obtained from the dispersion of fecal calprotectin concentrations at the three sites. The site fecal calprotectin concentrations, dilution consistency index, immune response kinetic uncertainty, and spatial heterogeneity uncertainty are written into the observation data package for subsequent measurement variance synthesis and state updates.

[0011] To obtain the concentration of fecal calprotectin at different dilution ratios, the following steps were taken: immunoreaction time-series signals were acquired and the kinetic uncertainty of the immunoreaction was obtained based on these signals; internal standard channel immunoreaction was performed on the extract to obtain the internal standard recovery rate; the fecal calprotectin concentration at different dilution ratios was corrected for recovery to obtain the corrected fecal calprotectin concentration; spatial heterogeneity uncertainty was obtained based on the dispersion of the corrected fecal calprotectin concentrations at three different dilution ratios; and dilution consistency index was obtained based on the consistency of the corrected fecal calprotectin concentrations at different dilution ratios.

[0012] Three fecal samples from the same defecation event were lysed and extracted to obtain extracts, which were then used as the detection targets for three sites. To reduce the interference of fecal matrix viscosity, ionic strength, and non-specific binding on the immune response, at least two diluted extracts were prepared for each extract using a pre-defined dilution process. The dilution process was completed by an automated pipetting mechanism or a quantitative pipette. The dilution medium used was a dilution buffer compatible with the immunoassay, and an internal standard was pre-added to the dilution buffer to form an internal standard channel input. The immunoassay used an immunoassay analyzer with time-resolved acquisition capability and a disposable reaction carrier. The reaction carrier contained a solid-phase immobilized calprotectin capture antibody and a labeled detection antibody. After the reaction started, the immunoassay analyzer acquired the immunoreaction time-series signal at a fixed sampling period. The immunoreaction time-series signal included calprotectin channel signals and internal standard channel signals. The concentration of fecal calprotectin at the sampling site was obtained by converting the calprotectin channel signal according to the pre-established standard curve. At the same time, the immune response time series signal was curve-fitted and the dispersion of the fitting residual was calculated. The dispersion of the fitting residual was written into the observation data packet as the uncertainty of immune response kinetics to reflect the reading instability caused by abnormal response kinetics, insufficient mixing or matrix inhibition.

[0013] The internal standard pathway immunoassay is used to obtain the internal standard recovery rate, which characterizes the systematic loss or amplification of quantitative results due to sample processing and detection procedures. The internal standard recovery rate and the site-corrected fecal calprotectin concentration are calculated using the following expressions: ； ; in, Internal standard recovery rate, The concentration of the internal standard obtained by converting the internal standard channel is the measured concentration. The concentration added is for internal standard. To correct the concentration of fecal calprotectin at the sampling point, The concentration of fecal calprotectin at the sampling site. Through the above recovery correction, the differences in sample lysis efficiency, pipetting errors, and the impact of reaction inhibition on quantification can be converted into traceable correction factors, thereby improving the comparability of test results from different sites and different batches.

[0014] Spatial heterogeneity uncertainty is obtained by the dispersion of fecal calprotectin concentrations at three sites. Specifically, it is achieved by averaging the fecal calprotectin concentrations at the three sites and calculating the standard deviation. The ratio of the standard deviation to the mean is used as the spatial heterogeneity uncertainty to characterize the patchy differences in the spatial distribution of the same defecation event. The dilution consistency index is used to characterize the consistency of fecal calprotectin concentrations at different dilution ratios. The calculation method of the dilution consistency index is described in the aforementioned calculation expression. By performing consistency calculations on fecal calprotectin concentrations at at least two dilution ratios, measurement results that exceed the linear range or exhibit matrix effects can be identified. The dilution consistency index, along with the immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty, are written into the observation data package to provide a reliable basis for subsequent measurement variance synthesis, observation gain calculation, and state updates.

[0015] Based on the observation data packet, the measurement variance is synthesized and the observation gain is calculated. Based on the observation gain, the latent state of inflammatory activity is updated and the trend slope, cumulative offset and individual dynamic threshold are generated to form a state data packet. After the observation data package is generated, the fecal calprotectin concentrations at each location are fused to obtain event-level fecal calprotectin observations. Dilution consistency indices, immune response kinetic uncertainty, and spatial heterogeneity uncertainty are extracted from the observation data package as sources of uncertainty, and the measurement variance is synthesized to characterize the reliability of the current observation. The measurement variance and observation gain can be determined using the following calculation expressions: ； ; in, To measure variance, These are the uncertainty components derived from the dilution consistency index. This represents the uncertainty component corresponding to the kinetic uncertainty of the immune response. These are the uncertainty components corresponding to spatial heterogeneity uncertainty. To detect the basic noise of the platform, The observation gain is used to recursively update the latent state of inflammatory activity. The update method uses the following calculation expression: ; in, This represents the latent state of inflammatory activity corresponding to the current defecation event. This represents the latent state of inflammatory activity corresponding to the previous defecation event. This represents the event-level fecal calprotectin observation value corresponding to the current defecation event. The trend slope is calculated by trending the most recent latent inflammatory activity states in chronological order; the individual dynamic threshold is determined by the central trend and dispersion of historical latent inflammatory activity states; the cumulative offset is recursively accumulated for the portion of latent inflammatory activity states exceeding the individual dynamic threshold, used to characterize the risk of persistently increased risk. All of the above results are written into the state data packet for subsequent event code generation and evidence accumulation.

[0016] The synthesis of measurement variance and calculation of observation gain based on observation data includes: generating event-level fecal calprotectin observations based on the fecal calprotectin concentrations at three locations. The event-level fecal calprotectin observations are obtained by weighted fusion of the fecal calprotectin concentrations at the three locations according to the weights determined by the dilution consistency index, the uncertainty of immune response kinetics, and the uncertainty of spatial heterogeneity, and are used to update the latent state of inflammatory activity.

[0017] In this embodiment, the observation data package includes fecal calprotectin concentrations at three locations, along with the corresponding dilution consistency index, immune response kinetic uncertainty, and spatial heterogeneity uncertainty. To ensure that subsequent updates to the latent state of inflammatory activity can preferentially utilize readings from locations with higher reliability, this embodiment merges the fecal calprotectin concentrations at the three locations into an event-level fecal calprotectin observation. Specifically, a location composite uncertainty is first constructed for each location. The location composite uncertainty is jointly determined by the dilution consistency index, the immune response kinetic uncertainty, and the spatial heterogeneity uncertainty of that location. The spatial heterogeneity uncertainty reflects the degree of deviation of the reading at that location from the overall distribution of the three locations. The spatial heterogeneity uncertainty is obtained by using the median of the fecal calprotectin concentrations at the three locations as a reference value, calculating the relative deviation between the fecal calprotectin concentration at that location and the reference value, and then writing the relative deviation into the spatial heterogeneity uncertainty of that location.

[0018] After obtaining the combined uncertainty of the three locations, the fusion weights of the three locations are calculated and weighted fusion is performed. The event-level fecal calprotectin observation value is calculated using the following expression: ； ; in, This represents the event-level fecal calprotectin observation value. For the first The concentration of fecal calcium protectant at the sampling site. For the first The fusion weight of the points, For the first The combined uncertainty of the points at the location, To avoid positive numbers with a denominator of zero. Point combination uncertainty. It is obtained by combining the dilution consistency index, the kinetic uncertainty of the immune response at the site, and the spatial heterogeneity uncertainty at the site. The combination method is to add the three as uncertainty components or weighted summation while maintaining the same dimension.

[0019] Through the aforementioned fusion mechanism, when a dilution consistency bias or increased dispersion of the residual fitting of the immune response time series signal occurs at a certain location, the uncertainty of the location synthesis increases and the fusion weight decreases, thereby reducing the impact of outliers on the event-level fecal calprotectin observation. When there are significant spatial distribution differences among the three locations, spatial heterogeneity uncertainty will impose a weight penalty on locations deviating from the median, making the event-level fecal calprotectin observation closer to the representative level of the same defecation event. The event-level fecal calprotectin observation is written into the observation data package and used for subsequent observation gain calculation and inflammatory activity latent state updates.

[0020] The measurement variance synthesized based on the observation data package includes: synthesizing the uncertainty components corresponding to the dilution consistency index, the uncertainty of immune response kinetics, the uncertainty of spatial heterogeneity, and the basic noise of the detection platform to obtain the measurement variance.

[0021] In this embodiment, the measurement variance is used to uniformly quantify the reliability of event-level fecal calprotectin observations. A larger measurement variance indicates that the current observation is more significantly affected by dilution linear deviation, immune response instability, or spatial differences, thereby reducing the update strength of this observation on the latent state of inflammatory activity in subsequent observation gain calculations. The measurement variance is synthesized from the uncertainty component corresponding to the dilution consistency index in the observation data package, the uncertainty of immune response kinetics, the uncertainty of spatial heterogeneity, and the basic noise of the detection platform. The basic noise of the detection platform reflects the reading fluctuations and intra-batch differences of the immunoassay analyzer itself. It is a fixed configuration item or estimated offline from the platform quality control data and can be updated with each detection batch.

[0022] In practice, the dilution consistency index is first segmented and mapped to obtain the corresponding uncertainty component. The segmented mapping is based on the linear interval characteristics of immunoassay: when the consistency of fecal calprotectin concentration at different dilution ratios is high, the dilution consistency deviation is considered small, and the uncertainty component is set to a smaller value; when the consistency is poor, it is determined that the dilution ratio exceeds the linear interval, there is matrix effect, or pipetting deviation, and the uncertainty component is set to a larger value. This segmented mapping can be achieved through a lookup table. The input to the lookup table is the dilution consistency index, and the output is the uncertainty component corresponding to the dilution consistency index. The lookup table entries are preset during the system's factory release or laboratory validation phase and can be updated based on quality control samples.

[0023] Subsequently, the uncertainty of immune response kinetics was obtained from the dispersion of the residuals fitted to the immune response time series signal. This uncertainty was directly used as an uncertainty component in the measurement variance synthesis to reflect short-term fluctuations caused by abnormal immune response curves, insufficient mixing, or response inhibition. Spatial heterogeneity uncertainty was obtained by correcting the dispersion of fecal calprotectin concentrations at three locations. This uncertainty reflected the impact of uneven distribution of the same defecation event across fecal sample locations on representative readings. This spatial heterogeneity uncertainty was also used as an uncertainty component in the measurement variance synthesis.

[0024] After obtaining each uncertainty component, the uncertainty components are combined with the basic noise of the detection platform to obtain the measurement variance. The measurement variance can be calculated using the following expression: ; in, To measure variance, This is to dilute the uncertainty component corresponding to the consistency index. This represents the uncertainty component corresponding to the kinetic uncertainty of the immune response. These are the uncertainty components corresponding to spatial heterogeneity uncertainty. To detect the platform's underlying noise, the above synthesis method can simultaneously reflect the combined impact of dilution consistency bias, immune response instability, site spatial differences, and inherent platform noise on measurement reliability within the same quantification framework, providing a unified input for subsequent observation gain calculations. Measurement variance is written into the status data package or stored in association with event-level fecal calprotectin observations, so that this reliability information can be reused during the disproving process triggering and evidence accumulation stages.

[0025] Calculating the observation gain includes determining the observation gain based on the measurement variance, and decreasing the observation gain as the measurement variance increases; updating the latent state of inflammatory activity includes weighted fusion updating of the latent state of inflammatory activity corresponding to the previous defecation event and the event-level fecal calprotectin observation value corresponding to the current defecation event based on the observation gain; generating the trend slope includes calculating the trend based on the latent state of inflammatory activity corresponding to multiple defecation events in chronological order to obtain the trend slope; generating the cumulative offset includes recursively accumulating the difference between the updated latent state of inflammatory activity and the individual dynamic threshold and limiting the cumulative offset to zero or a positive value; generating the individual dynamic threshold includes determining the individual dynamic threshold based on the central trend and dispersion of the latent state of inflammatory activity corresponding to historical defecation events.

[0026] Measurement variance characterizes the reliability of event-level fecal calprotectin observations for the current defecation event. A larger measurement variance indicates a more significant impact from dilution consistency bias, immune response instability, or spatial heterogeneity. Observation gain is determined by the measurement variance using the previously described expression for observation gain calculation, and decreases as measurement variance increases. The recursive update of the latent inflammatory state is performed using the previously described expression for updating the latent inflammatory state, weighting and fusing the observation gain with the latent inflammatory state corresponding to the previous defecation event and the event-level fecal calprotectin observation for the current defecation event. This update mechanism allows for a larger correction of the latent inflammatory state for defecation events with smaller measurement variances, and a smaller correction for defecation events with larger measurement variances, thereby suppressing false triggers caused by outliers, dilution nonlinearity, and instability in the immune response time series signal.

[0027] The individual dynamic threshold is used to characterize the boundary between the superposition of long-term levels and short-term fluctuations in an individual. It is determined based on the central tendency and dispersion of the latent inflammatory activity states corresponding to historical defecation events. The central tendency is obtained from the median of the latent inflammatory activity states within a preset window, and the dispersion is obtained from the median of the absolute deviations of the latent inflammatory activity states relative to the central tendency within the preset window. The individual dynamic threshold is calculated using the following expression: ; in, For individual dynamic thresholds, As the central trend, For the degree of dispersion, This is the preset scaling factor.

[0028] The trend slope is used to characterize the changing trend of latent inflammatory activity with the defecation event sequence. The latent inflammatory activity within a preset window is linearly fitted in chronological order, and the fitted slope is used as the trend slope and written into the state data packet. The cumulative offset is used to characterize the persistence of latent inflammatory activity exceeding the individual dynamic threshold. The difference between the updated latent inflammatory activity and the individual dynamic threshold is recursively accumulated, and the cumulative offset is limited to zero or a positive value. The recursive method uses the following calculation expression: ; in, This is the cumulative offset corresponding to the current defecation event. This is the cumulative offset corresponding to the previous defecation event. This represents the latent state of inflammatory activity corresponding to the current defecation event. The latent state of inflammatory activity, trend slope, cumulative offset, individual dynamic threshold, and observation gain are written into the state data packet for subsequent event code generation and evidence accumulation.

[0029] Event codes are generated based on state data packets, and the disproving process is executed based on the event codes to generate disproving result data packets and detection design instructions; After the status data packet is generated, the system reads the latent state of inflammatory activity, trend slope, cumulative offset, individual dynamic threshold, measurement variance, and observation gain. Based on a pre-defined rule table, it determines the anomaly type and encodes it as an event code. The rule table marks an increase in measurement variance or a decrease in observation gain as an anomaly in observation confidence, and marks an increase in latent inflammatory activity exceeding the individual dynamic threshold, a persistently positive trend slope, or an increasing cumulative offset as an anomaly in inflammatory activity risk. The anomaly type combination is then mapped to an event code field.

[0030] The system selects and executes the disproving process based on event codes, outputting detection design instructions and a disproving result data package. The detection design instructions indicate the additional dilution ratio immunoassay, immunoassay retest, fecal occult blood test, spiked recovery verification, or site expansion sampling required. The disproving result data package records the dilution consistency index, immunoreaction kinetic uncertainty, spatial heterogeneity uncertainty, and various verification results after retesting, for credibility adjustment and early warning correction in the subsequent evidence accumulation stage.

[0031] The event codes generated based on the state data packet include: determining the anomaly type based on the dilution consistency index, immune response kinetic uncertainty, spatial heterogeneity uncertainty, measurement variance and observation gain, and mapping the anomaly type to event codes used to indicate dilution consistency retest, immune response kinetic retest, hemorrhage retest, spike recovery and site expansion in the proof-of-conformity process.

[0032] In this embodiment, event codes are used to convert the anomaly clues reflected in the state data packet into executable disproving task selection results. This enables the disproving process to trigger dilution consistency retesting, immune response kinetics retesting, hemorrhage retesting, spike recovery, and site expansion in a deterministic manner, thereby forming a targeted verification and outputting a disproving result data packet. The state data packet at least contains measurement variance and observation gain, and is associated with the dilution consistency index, immune response kinetics uncertainty, and spatial heterogeneity uncertainty in the observation data packet. The system reads these indices as decision inputs.

[0033] Anomaly type determination is achieved using a pre-defined rule table, which includes rules for dilution consistency, immunodynamics, spatial heterogeneity, platform noise dominance, and composite anomalies. The dilution consistency rule identifies a decrease in the consistency of fecal calprotectin concentration at different dilution ratios. An anomaly is identified when the uncertainty component corresponding to the dilution consistency index increases, leading to an increase in measurement variance. The immunodynamics rule identifies an increase in the dispersion of the residuals from the fitting of the immunoreaction time series signal. An anomaly is identified when the uncertainty of immunodynamics increases, leading to an increase in measurement variance, or when the observation gain decreases to a pre-defined lower limit. The spatial heterogeneity rule identifies an increase in the dispersion of fecal calprotectin concentration at three different locations. An anomaly is identified when the spatial heterogeneity uncertainty increases, accompanied by an increase in measurement variance. The platform noise dominance rule identifies situations where the measurement variance is mainly contributed by the base noise of the detection platform, and the uncertainty of dilution consistency, immunodynamics, and spatial heterogeneity have not increased significantly, indicating batch or equipment status factors. Composite anomaly rules are used to identify situations where multiple anomalies are simultaneously satisfied, and to set priority order for subsequent proof-of-contrast tasks.

[0034] After obtaining the anomaly type, it is mapped to an event code. The event code consists of a task indication field and a priority field. The task indication field indicates the specific set of tasks to be executed in the disproving process, and the priority field indicates the task execution order and whether additional tasks are needed. The mapping relationships are as follows: dilution consistency anomaly corresponds to dilution consistency retest; immune response kinetics anomaly corresponds to immune response kinetics retest; spatial heterogeneity anomaly corresponds to site expansion; platform noise-dominated anomaly corresponds to spiked recovery. When dilution consistency anomaly and immune response kinetics anomaly coexist, immune response kinetics retest is triggered first, followed by dilution consistency retest. When spatial heterogeneity anomaly and latent inflammatory activity exceeding the individual dynamic threshold occur simultaneously, bleeding retest is triggered concurrently with site expansion to reduce the risk of misjudgment due to blood component interference. The event code is written into the status data packet and used as the basis for disproving process scheduling, thus ensuring that the selection of disproving tasks for each defecation event has traceable input basis and a consistent execution path.

[0035] The event-code-based disprovenance process to generate a disprovenance result data package and detection design instructions includes: adding an immunoassay at a different dilution ratio to the existing immunoassays at different dilution ratios and recalculating the dilution consistency index; performing immunoassays again on the extract and collecting the immunoreaction time series signal to re-obtain the immunoreaction kinetic uncertainty; performing fecal occult blood testing on fecal samples from the same defecation event to generate bleeding retest results; dividing the extract and adding calprotectin standard to the spiked sample, obtaining the spiked recovery rate based on the immunoassay results before and after spiking; expanding the fecal samples from the same defecation event from three sites to five sites and lysing and extracting the expanded extracts separately, mixing the five expanded extracts to obtain a combined extract, and selecting one of the expanded extracts as the fixed site extract, recalculating the spatial heterogeneity uncertainty based on the immunoassay results of the combined extract and the fixed site extract; generating a disprovenance result data package and detection design instructions based on each retest result and verification result.

[0036] In this embodiment, the disproving process is triggered by an event code and driven by detection design instructions. These instructions specify the retesting target, add dilution ratios, and define the verification items and sampling point expansion methods, enabling the disproving process to form a reproducible retesting and verification loop within the same defecation event. After receiving the event code, the system parses the corresponding task set and sequentially completes one or more of the following according to a preset execution order: dilution consistency retesting, immune response kinetics retesting, bleeding retesting, spike recovery verification, and sampling point expansion verification. Subsequently, a disproving result data packet is generated.

[0037] The dilution consistency retest was performed based on the existing immunoassays at different dilution ratios. The system added a dilution ratio to the test design instructions and prepared an additional diluted extract at the added dilution ratio using the same extract. Immunoassays were then performed using the same immunoassay analyzer and reaction carrier as before. The system used both the site-corrected fecal calprotectin concentration obtained at the added dilution ratio and the site-corrected fecal calprotectin concentration at the original dilution ratio to recalculate the dilution consistency index, distinguishing between systematic bias caused by readings outside the linear range and random bias caused by occasional fluctuations. The recalculated dilution consistency index was written into the evidence-of-contrast data package and used to adjust the observation confidence during the subsequent evidence accumulation phase.

[0038] Immunoreactivity kinetics retesting is used to confirm whether the instability of the immunoreactivity time-series signal is a one-time anomaly. The system performs immunoassay again on the same extract, acquiring immunoreactivity time-series signals at a fixed sampling period after the reaction starts. The system calculates the dispersion of the fitting residuals using the same fitting method as before to re-obtain the immunoreactivity kinetics uncertainty. The re-obtained immunoreactivity kinetics uncertainty is written into the proof-of-contrast data packet and compared with the previous immunoreactivity kinetics uncertainty to determine whether the anomaly is reproducible.

[0039] Bleeding retesting is used to identify potential interference from blood components on immunoassay signals. The system performs fecal occult blood testing on stool samples from the same defecation event. Fecal occult blood testing can be performed using either a chemical colorimetric method or an immunoassay strip. The negative or positive result output is written into the reconsideration data package as the bleeding retest result. The bleeding retest result is used to adjust the credibility of risk evidence during subsequent evidence accumulation phases, preventing signal increases due to bleeding factors from being directly interpreted as a recurrence of inflammation.

[0040] Spiked recovery validation is used to verify matrix inhibition or quantitative drift. The system aliquots the extract into spiked and unspiked samples. Calprotectin standard is added to the spiked samples and incubated. Immunoassays are then performed on both spiked and unspiked samples, and the spiked recovery rate is obtained based on the results. The spiked recovery rate reflects the quantitative traceability of this detection chain, is written into the counter-evidence data package, and is used to correct the observation reliability in subsequent evidence accumulation stages.

[0041] The expanded sampling site verification is used to confirm whether three sampling sites are sufficient to represent the overall level of the defecation event. In the detection design instructions, the system expands the sampling sites from three to five, and extracts expanded extracts from each of the five fecal samples. The system mixes the five expanded extracts in equal volumes to obtain a composite extract, and selects one of the expanded extracts as the fixed-site extract. The system performs immunoassay on both the composite extract and the fixed-site extract, and recalculates the spatial heterogeneity uncertainty based on the detection results to determine whether spatial distribution differences are the main source of observational instability. The recalculated spatial heterogeneity uncertainty is written into the proof-of-contrast data packet.

[0042] The counter-evidence result data package includes at least the recalculated dilution consistency index, the re-obtained immunoreaction kinetic uncertainty, the hemorrhage retest results, the spiked recovery rate, and the recalculated spatial heterogeneity uncertainty, and is stored in association with the current test design instructions. The test design instructions record the specific retesting and verification content performed in this counter-evidence process, facilitating subsequent traceability and unified retrieval of counter-evidence results from different sources during the evidence accumulation phase. Through the above counter-evidence process, anomalies in the observation and status data packages can be transformed into executable verification tasks and form a counter-evidence result data package, thereby reducing false alarms caused by dilution nonlinearity, immunoreaction fluctuations, hemorrhage interference, and spatial heterogeneity while maintaining early warning sensitivity.

[0043] Evidence is accumulated based on state data packets and counter-evidence result data packets to output graded early warning results, and the individual parameter set is updated after obtaining the relapse outcome label.

[0044] In this embodiment, after obtaining the state data packet and the counter-evidence result data packet, the system enters the evidence accumulation and graded early warning stage. The state data packet provides the latent state of inflammatory activity, trend slope, cumulative offset, individual dynamic threshold, measurement variance, and observation gain; the counter-evidence result data packet provides dilution consistency retest results, immune response kinetics retest results, bleeding retest results, spiked recovery rate, and site expansion verification results. The system first generates a credibility coefficient based on the counter-evidence result data packet, which is used to attenuate or maintain the current evidence increment, thereby preventing the amplification of risk signals when the observation credibility is insufficient.

[0045] The evidence is accumulated using the following calculation expression: ； ; in, To increase the amount of evidence, To accumulate evidence, The credibility coefficient This represents the latent state of inflammatory activity corresponding to the current defecation event. This represents the individual dynamic threshold corresponding to the current defecation event. The trend slope For cumulative offset, , , The system uses pre-defined weights. It compares the accumulated evidence with the accumulated evidence boundary to output a graded early warning result. If the accumulated evidence is below the first boundary, it outputs a low-risk level; if the accumulated evidence is between the first and second boundaries, it outputs a medium-risk level; and if the accumulated evidence is above the second boundary, it outputs a high-risk level.

[0046] After obtaining the relapse outcome label, the system updates the individual parameter set. The individual parameter set includes at least the central tendency and dispersion used to generate the individual dynamic threshold, the cumulative evidence boundary, and the observation gain mapping rule. The central tendency and dispersion can be updated using exponential smoothing. ； ; in, As the central trend, For the degree of dispersion, This is the smoothing coefficient. The system then calibrates the cumulative evidence boundary and confidence coefficient generation rules according to a pre-set update table based on the matching relationship between the relapse outcome label and the results of the most recent graded early warnings, in order to maintain the consistency and traceability of individualized early warnings.

[0047] The cumulative evidence based on state data packets and counter-evidence result data packets to output graded early warning results includes: generating evidence increments based on the difference between the latent state of inflammatory activity and the individual dynamic threshold, trend slope, cumulative offset, and measurement variance; determining the weight of the evidence increments based on the counter-evidence result data packets; recursively accumulating evidence based on the evidence increments and evidence increment weights; and outputting graded early warning results based on the cumulative evidence boundary. The graded early warning results include low-risk, medium-risk, and high-risk levels.

[0048] In this embodiment, after obtaining the state data packet and the counter-evidence result data packet, the system performs evidence accumulation to output a graded early warning result. The state data packet includes at least the latent state of inflammatory activity, individual dynamic threshold, trend slope, cumulative offset, and measurement variance; the counter-evidence result data packet includes at least the dilution consistency retest result, immune response kinetics retest result, bleeding retest result, spiked recovery rate, and site expansion validation result. The system first generates an evidence increment based on the state data packet. The evidence increment is used to quantify the new contribution of the current defecation event to the recurrence risk. To suppress the false accumulation caused by low-confidence observations, a penalty term for measurement variance is introduced into the evidence increment, so that the evidence increment decreases when the measurement variance increases. The evidence increment is calculated using the following expression: ; in, To increase the amount of evidence, For the incremental weight of evidence, This represents the latent state of inflammatory activity corresponding to the current defecation event. This represents the individual dynamic threshold corresponding to the current defecation event. The trend slope For cumulative offset, To measure variance, For the threshold exceeding the limit, the weight is... For the trend term weight, Weights for persistent items.

[0049] The weight of the evidence increment is determined by the data packet of rebuttal results and is used to adjust the credibility of the evidence increment by converting the retest and verification results. The system generates the evidence increment weight according to preset mapping rules: when the dilution consistency retest result indicates an improvement in the dilution consistency index, the evidence increment weight is increased; when the immune response kinetics retest result indicates a stable immune response time series signal, the evidence increment weight is increased; when the bleeding retest result is positive, the evidence increment weight is decreased; when the spike recovery rate falls into the preset acceptable range, the evidence increment weight is increased; when the site expansion verification result indicates a decrease in spatial heterogeneity uncertainty, the evidence increment weight is increased. The mapping rules are fixed in the system in the form of a rule table, which gives the weight adjustment direction and adjustment range corresponding to each type of rebuttal result, making the generation of evidence increment weights deterministic and traceable.

[0050] The system then recursively calculates the accumulated evidence and outputs tiered early warning results. The accumulated evidence is calculated using the following expression: ; in, This is the cumulative evidence corresponding to the current defecation event. This represents the cumulative evidence corresponding to the previous defecation event. The system compares the cumulative evidence with the cumulative evidence boundary to output a graded warning result: when the cumulative evidence is below the first boundary, a low-risk level is output; when the cumulative evidence is between the first and second boundaries, a medium-risk level is output; and when the cumulative evidence is above the second boundary, a high-risk level is output. The cumulative evidence boundary is stored by the system as part of the individual parameter set and is recorded in association with the rebuttal result data packet so that the cumulative evidence boundary and the evidence increment weight generation rule can be calibrated and updated after the relapse outcome label is obtained.

[0051] After obtaining the relapse outcome label, updating the individual parameter set includes: calibrating and updating the individual dynamic threshold generation rule, cumulative evidence boundary and observation gain calculation rule based on the relapse outcome label, and updating the basic noise of the detection platform based on the relapse outcome label for synthesizing measurement variance.

[0052] In this embodiment, after outputting the graded early warning results, the system enters the individual parameter set update stage. The recurrence outcome label indicates whether inflammatory bowel disease recurred within a preset follow-up window. The recurrence outcome label is generated and written into the system from clinical follow-up records, endoscopic evaluation conclusions, or physician diagnostic conclusions. The system uses the recurrence outcome label as a supervisory signal to calibrate and update the individual dynamic threshold generation rules, cumulative evidence boundaries, and observation gain calculation rules, ensuring consistency between the early warning results of subsequent defecation events and the actual outcomes, and reducing bias caused by baseline differences between individuals.

[0053] The calibration and update of the individual dynamic threshold generation rule are performed based on the correspondence between the latent state of inflammatory activity and the relapse outcome label corresponding to the most recent defecation events. The system maintains central tendency and dispersion as individual baseline statistics, which are updated using exponential smoothing recursion. The calculation expression for the exponential smoothing recursion update has been given previously and will not be repeated here. When the relapse outcome label is "relapse," the system increases the update intensity of dispersion, making the individual dynamic threshold more sensitive to increased risk; when the relapse outcome label is "no relapse," the system decreases the update intensity of dispersion, making the individual dynamic threshold more robust to short-term fluctuations. The system generates the individual dynamic threshold based on the updated central tendency and dispersion, and writes the smoothing coefficient and dispersion amplification coefficient from the individual dynamic threshold generation rule into the individual parameter set for use in calculating the individual dynamic threshold for subsequent defecation events.

[0054] The calibration and update of the cumulative evidence boundary is used to control the sensitivity and specificity of the graded early warning results. The system aligns the graded early warning results in the follow-up window with the relapse outcome label, statistically analyzes the hit and false alarm rates corresponding to low-risk, medium-risk, and high-risk levels, and adjusts the first and second boundaries according to a preset update table: when the relapse outcome label is "relapse" and the historical early warning level has repeatedly fallen into the low-risk level, the first and second boundaries are lowered; when the relapse outcome label is "no relapse" and the historical early warning level has repeatedly fallen into the high-risk level, the first and second boundaries are raised. The preset update table is fixed in the system in the form of a rule table, recording the boundary adjustment direction and adjustment magnitude corresponding to different error types, thereby ensuring that the update of the cumulative evidence boundary has determinism and traceability.

[0055] The calibration update of the observation gain calculation rules is used to ensure that the mapping between observation gain and measurement variance conforms to the actual fluctuation level of the individual and the detection platform. After the relapse outcome label arrives, the system reads the measurement variance, observation gain, and retest and verification results in the data packet of the reconfirmation results within the follow-up window to determine whether the observation gain is sufficient to suppress low-confidence observations. When the relapse outcome label is non-relapsed and there are high-risk false alarms due to increased measurement variance, the system strengthens the mapping strength where the observation gain decreases as the measurement variance increases; when the relapse outcome label is relapsed and there are low-risk missed alarms due to small measurement variance, the system weakens the observation gain suppression strength. The mapping strength is stored in the individual parameter set as rule parameters for subsequent observation gain calculations for defecation events.

[0056] The baseline noise of the detection platform is used to synthesize the measurement variance, reflecting the baseline fluctuations caused by instrument, reagent batch, and operational errors under the same sample and testing procedure. When updating the baseline noise of the detection platform based on the relapse outcome label, the system selects defecation events with the relapse outcome label of "no relapse" and whose evidence data package indicates stable dilution consistency index, stable immunoreaction kinetic uncertainty, and spike recovery rate within a preset acceptable range as calibration samples. The lower limit of the measurement variance of the calibration samples is statistically analyzed, and the value range of the baseline noise of the detection platform is updated accordingly. The updated baseline noise of the detection platform is written into the individual parameter set and participates in the synthesis as a fixed noise component in subsequent synthesis of the measurement variance to reduce false alarms caused by platform state drift. Through the above-mentioned individual parameter set update mechanism, the early warning strategy can be aligned with the individual baseline and the detection platform state during continuous use, thereby improving the stability and traceability of the graded early warning results.

[0057] In one specific embodiment of the present invention, the following example uses a subject in remission of inflammatory bowel disease as the subject, continuously collecting 12 defecation events (each approximately 4 days apart) and running the process of the present invention online. For each defecation event, samples were taken from three locations within the same defecation event and lysed for extraction; the immunoassay platform performed calprotectin immunoassay on extracts at different dilution ratios and output concentration and immune response time series signals; subsequently, observation data packets, status data packets, event codes, and rebuttal result data packets were generated sequentially, and evidence was recursively accumulated; finally, a low-risk, medium-risk, or high-risk level was output, and detection design instructions were simultaneously given.

[0058] Three sampling sites from the same defecation event were tested at dilution ratios of 1:50, 1:100, and 1:200, respectively. The testing platform simultaneously performed internal standard spiking to obtain the spiked recovery rate. Recovery correction was first applied to each test result to obtain the site-corrected fecal calprotectin concentration: ; in, For point At dilution ratio The concentration of fecal calprotectin was corrected at the following locations; Raw readings from the immunoassay test; For point The corresponding spiked recovery rate conversion ratio. Subsequently, within each sampling point, the dilution consistency index was calculated based on the sampling point-corrected fecal calprotectin concentration at the three dilution ratios; the kinetic uncertainty of the immune response was obtained based on the fitting residuals and rising segment jitter of the immune response time series signal; and the spatial heterogeneity uncertainty was obtained based on the dispersion of the fecal calprotectin concentration at the three sampling points. For example... Figure 2 As shown, the horizontal axis represents days, and the vertical axis represents index values. The blue broken line (marked with dots) represents the dilution consistency index, the orange broken line (marked with squares) represents the immunoreaction kinetic uncertainty, and the green broken line (marked with triangles) represents the spatial heterogeneity uncertainty. The figure shows that the dilution consistency index significantly increased on day 16, indicating a more pronounced dilution consistency anomaly; the immunoreaction kinetic uncertainty fluctuated within a narrow range overall, but showed an increase at certain time points; the spatial heterogeneity uncertainty fluctuated significantly and increased in the later stages (such as around days 28, 36, and 44), indicating an increased contribution of location differences to measurement uncertainty. This figure is used to illustrate the temporal evolution of quality indicators in the observation data package and provides a basis for subsequent measurement variance synthesis and event code triggering. The representative values ​​of the three locations are weighted and fused to obtain the event-level fecal calprotectin observation value. The weights are jointly constrained by the dilution consistency index, immunoreaction kinetic uncertainty, and spatial heterogeneity uncertainty, reducing the contribution of abnormal locations to the event-level fecal calprotectin observation value, thereby avoiding single-point anomalies directly dominating subsequent early warnings.

[0059] In handling measurement uncertainty, the measurement variance is obtained by synthesizing the uncertainty components corresponding to the dilution consistency index, the uncertainty of immune reaction kinetics, the spatial heterogeneity uncertainty, and the fundamental noise of the detection platform. The fundamental noise of the detection platform is obtained by statistical analysis of samples from the platform blank and stable periods. The observation gain is obtained by monotonically mapping the measurement variance, and satisfies the condition that the observation gain decreases as the measurement variance increases. Figure 3 As shown in the figure, the horizontal axis represents days; the left vertical axis represents measurement variance (blue broken line, marked with dots), and the right vertical axis represents observation gain (red broken line, marked with squares). It can be seen from the figure that around day 16, measurement variance increases significantly while observation gain decreases synchronously; around day 20, measurement variance is at a low level while observation gain is at a high level; thereafter, each time point shows an inverse relationship of "increased measurement variance leads to decreased observation gain, and decreased measurement variance leads to increased observation gain." This emphasizes historical continuity when observation quality is poor, and emphasizes the contribution of the current observation to the latent state update when observation quality is good.

[0060] The latent state of inflammatory activity is updated using a weighted fusion of observational gain, combining the latent state of inflammatory activity from the previous defecation event with the observed fecal calprotectin value at the event level of the current defecation event. A lower observational gain emphasizes historical continuity, while a higher gain emphasizes the current observation. The trend slope of the latent state of inflammatory activity is calculated chronologically based on multiple consecutive defecation events. The difference between the updated latent state of inflammatory activity and the individual dynamic threshold is recursively accumulated as a cumulative offset, limited to zero or a positive value. The individual dynamic threshold is generated from the central trend and dispersion of the latent state of inflammatory activity corresponding to historical defecation events, used to adapt to long-term baseline differences and short-term fluctuations. Figure 4 As shown, the horizontal axis represents days (corresponding to the occurrence time of 12 defecation events), and the vertical axis represents the concentration and latent state in the same dimension. The blue broken line (marked with dots) represents the observed value of fecal calprotectin at the event level, the orange broken line (marked with squares) represents the updated latent state of inflammatory activity, and the green dashed line represents the individual dynamic threshold. As can be seen from the figure, in the early stage (day 0 to day 12), the observed value and latent state were in a low fluctuation range; on day 16, the observed value showed a sudden increase, while the upward movement of the latent state was relatively limited; from day 32 onwards, the observed value and latent state rose continuously in sync, and exceeded the individual dynamic threshold multiple times; the individual dynamic threshold slowly declined in the early stage, and gradually moved upward as the latent state rose in the later stage, reflecting the adaptiveness of the threshold to the individual baseline and stage level.

[0061] In terms of anomaly identification and rebuttal linkage, the anomaly type is determined based on dilution consistency index, immune response kinetic uncertainty, spatial heterogeneity uncertainty, measurement variance and observation gain. The anomaly type is then mapped to event codes that indicate the rebuttal process. Each event code corresponds one-to-one with a rebuttal action, including dilution consistency retest, immune response kinetic retest, bleeding retest, spike recovery and site expansion. Upon receiving the event code, the disproving process is executed, generating a disproving result data packet. Simultaneously, detection design instructions are generated: First, a dilution consistency retest is performed, adding a new dilution ratio to the existing dilution ratio and recalculating the dilution consistency index. Second, an immunokinetics retest is performed, re-immunizing the extract and re-acquiring the immunoreaction time series signal to re-obtain the immunokinetic uncertainty. Third, a bleeding retest is performed, detecting fecal occult blood in fecal samples from the same defecation event to generate a bleeding retest result. Fourth, a spiked recovery is performed, dividing the extract and adding calprotectin standard to the spiked samples, obtaining the spiked recovery rate based on the immunodetection results before and after spiked analysis. Fifth, a site expansion is performed, expanding the fecal samples from three sites to five sites from the same defecation event, lysing and extracting the expanded extracts separately, mixing the five expanded extracts to obtain a combined extract, and selecting one of the expanded extracts as the fixed site extract. The spatial heterogeneity uncertainty is recalculated based on the immunodetection results of the combined extract and the fixed site extract. The data package containing the evidence of rebuttal is used to adjust the credibility of the evidence increment in the future. The test design instructions are used to constrain the organization of the next test, such as increasing the dilution ratio, shortening the next test interval, enabling the five-point mode, or adding occult blood retesting.

[0062] In terms of early warning output, evidence increments are generated based on the difference between the latent state of inflammatory activity and the individual dynamic threshold, the trend slope, the cumulative offset, and the measurement variance; the weight of the evidence increment is determined based on the data package of refutation results; the accumulated evidence is recursively calculated based on the evidence increment and its weight, and compared with the cumulative evidence boundary to output a graded early warning result, which includes low-risk, medium-risk, and high-risk levels. Figure 5 As shown, the horizontal axis represents days, and the vertical axis represents cumulative evidence. The blue broken line (marked with dots) represents the recursive cumulative evidence, the orange dashed line represents the first boundary, and the green dotted line represents the second boundary. The graph shows that in the early stages, the cumulative evidence is close to zero and rises slowly; from day 20 to day 28, the cumulative evidence is at a low level and accumulates slowly; on day 32, the cumulative evidence jumps and exceeds both boundary lines, and then continues to rise and remain above the boundary lines on days 36, 40, and 44, visually reflecting the process of the graded warning entering and remaining in the high range from a low level. The cumulative evidence boundaries are maintained in the individual parameter set, facilitating adaptive updates based on the relapse outcome label.

[0063] To facilitate reproducibility, the core calculation results for each defecation event are given (observed values ​​are dimensionless concentrations, while indices and variances are dimensionless or relative quantities): Defecation Event 1 (Day 0): Observed value 114.4, dilution concordance index 0.370, immune response kinetic uncertainty 0.074, spatial heterogeneity uncertainty 0.091, measurement variance 0.0418, observation gain 0.323, latent inflammatory activity 114.4, individual dynamic threshold 159.2, cumulative evidence 0.000, low risk level, event code: dilution concordance retest. Defecation Event 2 (Day 4): Observed value 123.0, dilution concordance index 0.216, immune response kinetic uncertainty 0.052, spatial heterogeneity uncertainty 0.056, measurement variance 0.0153, observation gain 0.566, latent inflammatory activity 119.3, individual dynamic threshold 152.7, cumulative evidence 0.000, low risk level, event code: conventional. Event 3 (Day 8): Observed value 101.8, dilution concordance index 0.293, immune response kinetic uncertainty 0.042, spatial heterogeneity uncertainty 0.092, measurement variance 0.0265, observation gain 0.430, latent inflammatory activity 111.7, individual dynamic threshold 148.6, cumulative evidence 0.000, low risk level, event code: dilution concordance retest. Event 4 (Day 12): Observed value 125.5, dilution concordance index 0.194, immune response kinetic uncertainty 0.060, spatial heterogeneity uncertainty 0.074, measurement variance 0.0139, observation gain 0.590, latent inflammatory activity 119.9, individual dynamic threshold 145.1, cumulative evidence 0.000, low risk level, event code: conventional. Event 5 (Day 16): Observed value 359.0, dilution concordance index 0.981, immune response kinetic uncertainty 0.138, spatial heterogeneity uncertainty 0.200, measurement variance 0.0596, observation gain 0.251, latent inflammatory activity 183.0, individual dynamic threshold 144.7, cumulative evidence 0.028, low risk level, event code: dilution concordance retest, immune response kinetic retest, bleeding retest, spiked recovery. Event 6 (Day 20): Observed value 97.0, dilution concordance index 0.186, immune response kinetic uncertainty 0.057, spatial heterogeneity uncertainty 0.117, measurement variance 0.0143, observation gain 0.584, latent inflammatory activity 132.9, individual dynamic threshold 159.7, cumulative evidence 0.028, low risk level, event code: routine.Event 7 (Day 24): Observation value 134.1, dilution concordance index 0.370, immune response kinetic uncertainty 0.063, spatial heterogeneity uncertainty 0.080, measurement variance 0.0330, observation gain 0.377, latent inflammatory activity 133.4, individual dynamic threshold 149.2, cumulative evidence 0.028, low risk level, event code: dilution concordance test-through. Event 8 (Day 28): Observation value 131.7, dilution concordance index 0.205, immune response kinetic uncertainty 0.060, spatial heterogeneity uncertainty 0.062, measurement variance 0.0141, observation gain 0.587, latent inflammatory activity 132.4, individual dynamic threshold 145.2, cumulative evidence 0.028, low risk level, event code: standard. Event 9 (Day 32): Observation value 300.1, dilution concordance index 0.215, immune response kinetic uncertainty 0.068, spatial heterogeneity uncertainty 0.185, measurement variance 0.0092, observation gain 0.685, latent inflammatory activity 247.6, individual dynamic threshold 152.3, cumulative evidence 1.010, high risk level, event code: standard. Event 10 (Day 36): Observation value 316.4, dilution concordance index 0.390, immune response kinetic uncertainty 0.074, spatial heterogeneity uncertainty 0.124, measurement variance 0.0192, observation gain 0.510, latent inflammatory activity 282.7, individual dynamic threshold 208.0, cumulative evidence 1.038, high risk level, event code: dilution concordance test-through. Event 11 (Day 40): Observed value 448.1, dilution concordance index 0.422, immune response kinetic uncertainty 0.082, spatial heterogeneity uncertainty 0.094, measurement variance 0.0184, observation gain 0.521, latent inflammatory activity 368.9, individual dynamic threshold 259.7, cumulative evidence 1.160, high risk level, event code: dilution concordance test-through. Event 12 (Day 44): Observed value 563.3, dilution concordance index 0.456, immune response kinetic uncertainty 0.090, spatial heterogeneity uncertainty 0.244, measurement variance 0.0267, observation gain 0.428, latent inflammatory activity 450.2, individual dynamic threshold 333.5, cumulative evidence 1.292, high risk level, event code: dilution concordance test-through, site expansion.

[0064] The verifiable calculations are as follows. First, in fecal event 5, an abnormal increase occurred, but the dilution consistency index and the uncertainty of immune response kinetics simultaneously worsened, significantly increasing the measurement variance and reducing the observation gain. The latent state of inflammatory activity only shifted from approximately 120 to 183 without being "overwhelmed" by a single abnormality. At the same time, the reversal process triggered bleeding retesting and spiked recovery, and reduced the weight of incremental evidence (0.25 in this embodiment). The cumulative evidence remained in the low-risk range, demonstrating the effect of identifying and suppressing abnormal observations. The original detection point data for this event can be verified in the original detection point data file. For example, the site-corrected fecal calprotectin concentrations at 1:50, 1:100, and 1:200 were approximately 499.4, 419.8, and 317.5, respectively, showing significant dilution inconsistency. Secondly, starting with defecation event 9, a continuous increase in fecal calprotectin levels was observed, with normal quality indicators, small measurement variance, and high observation gain. The latent state of inflammatory activity rapidly followed the upward shift of the event-level fecal calprotectin observation value and crossed the individual dynamic threshold. The accumulated evidence crossed the second boundary at once, resulting in a high-risk level. Subsequently, defecation events 10 to 12 maintained a stable high-risk output with the support of retesting and site expansion. According to the relapse outcome labeling window set in this embodiment (marked as relapse from day 42), the high-risk warning appeared as early as day 32, achieving a graded warning approximately 10 days in advance.

[0065] After the relapse outcome label was obtained on day 44, the individual dynamic threshold generation rule, cumulative evidence boundary, and observation gain calculation rule were calibrated and updated: the inflammatory activity latent state sequence of several defecation events before relapse was used as a positive window, and the dispersion coefficient in the individual dynamic threshold generation rule was appropriately lowered to improve sensitivity; the second boundary was slightly lowered to reduce high-risk trigger hysteresis; the basic noise of the detection platform was re-estimated using the residuals of the stable period and the pre-relapse window and written into the individual parameter set for subsequent synthetic measurement variance, thereby improving the response speed to the relapse trend in the next cycle without sacrificing the ability to suppress evidence against contradiction.

[0066] like Figure 6 As shown, an inflammatory bowel disease recurrence early warning system based on fecal calprotectin is used to implement an inflammatory bowel disease recurrence early warning method based on fecal calprotectin. The system includes: The observation generation module collects fecal samples from three locations during the same defecation event, lyses and extracts the resulting solutions, performs immunoassay on the extracts at different dilution ratios to obtain the fecal calprotectin concentration at each site, and generates an observation data package containing dilution consistency indices, immunoreaction kinetic uncertainty, and spatial heterogeneity uncertainty. The observation generation module can be hardware-wise composed of a disposable sampling device, a sample preservation container, a lysis extraction unit, and an immunoassay unit. The lysis extraction unit may include a quantitative dispensing mechanism, a sealed oscillating mixing mechanism, a filtration or centrifugation separation mechanism, and a waste collection structure, used to prepare extracts from the three fecal samples separately. The immunoassay unit can employ a lateral chromatography reader or a microplate reader, equipped with a multi-channel pipetting mechanism to perform sample dispensing at different dilution ratios, and uses optical detection devices to acquire reaction signals to obtain the fecal calprotectin concentration at each site. Simultaneously, a local processor calculates the dilution consistency indices, immunoreaction kinetic uncertainty, and spatial heterogeneity uncertainty, and encapsulates these into an observation data package.

[0067] The state update module is used to synthesize measurement variance and calculate observation gain based on observation data packets. It then updates the latent state of inflammatory activity based on the observation gain and generates trend slope, cumulative offset, and individual dynamic thresholds to form a state data packet. The state update module can be hardware-wise composed of an embedded processor, memory, and a secure storage unit. The processor performs observation data packet parsing, measurement variance synthesis, and observation gain calculation, and iteratively updates the latent state of inflammatory activity. The memory stores the sequence of latent inflammatory activity states and individual parameter sets from historical defecation events to support the generation of trend slope, cumulative offset, and individual dynamic thresholds. The secure storage unit stores individualized threshold rules and boundary rules, ensuring the update process is traceable and not easily modified without authorization, thus outputting the state data packet for subsequent modules to call.

[0068] The disproving trigger module is used to generate event codes based on the status data packet and execute the disproving process based on the event codes to generate a disproving result data packet and detection design instructions. The disproving trigger module can be hardware-wise composed of an event judgment and processing unit, peripheral interfaces, and linkage interfaces with executable detection components. The event judgment and processing unit generates event codes based on the status data packet and drives a preset disproving process. The peripheral interfaces are used to connect to a occult blood detection reader, a pipetting mechanism for spiked recovery, and an extended sampling device identification device, etc., to convert the "requiring retesting or verification" action into executable detection design instructions. The linkage interface is used to issue new dilution ratios, retesting times, and sampling point configurations to the immunoassay unit and to collect retest data to form a disproving result data packet.

[0069] The evidence warning module is used to accumulate evidence based on status data packets and rebuttal result data packets to output graded warning results, and to update the individual parameter set after obtaining the relapse outcome label. The evidence warning module can be hardware-wise composed of a data fusion processor, a communication unit, and a human-computer interaction unit. The data fusion processor receives status data packets and rebuttal result data packets, performs evidence accumulation and grading judgment, and outputs graded warning results. The communication unit is used to send the warning results and test design instructions to a mobile terminal or clinical information system, and to receive subsequent relapse outcome labels. The human-computer interaction unit can be a display screen or a mobile application, used to present the risk level and recommended retesting actions. Simultaneously, after obtaining the relapse outcome label, the data fusion processor updates the individual parameter set and writes the updated parameters into memory for reuse in subsequent defecation events.

[0070] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for early warning of relapse of inflammatory bowel disease based on fecal calprotectin, characterized in that, Includes the following steps: Three fecal samples were collected from the same defecation event and lysed to obtain extracts. The extracts were subjected to immunoassay at different dilution ratios to obtain the concentration of fecal calprotectin at the site, and an observation data package containing dilution consistency index, immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty was generated. Based on the observation data package, the measurement variance is synthesized and the observation gain is calculated. Based on the observation gain, the latent state of inflammatory activity is updated and the trend slope, cumulative offset and individual dynamic threshold are generated to form a state data package. Based on the state data packet, an event code is generated, and based on the event code, a disproving process is executed to generate a disproving result data packet and a detection design instruction; Based on the status data packet and the counter-evidence result data packet, evidence is accumulated to output a graded early warning result, and the individual parameter set is updated after obtaining the relapse outcome label.

2. The method according to claim 1, characterized in that, Immunoassay was performed on the extract at different dilution ratios to obtain the concentration of fecal calprotectin at the specified sites, including: Collect immune response time-series signals and obtain the kinetic uncertainty of the immune response based on the immune response time-series signals; The extract was subjected to internal standard channel immunoassay to obtain the internal standard recovery rate. Based on the internal standard recovery rate, the fecal calprotectin concentration at the site was corrected to obtain the site-corrected fecal calprotectin concentration. The spatial heterogeneity uncertainty is obtained by correcting the dispersion of fecal calprotectin concentration at three locations. The dilution consistency index is obtained by correcting the consistency of fecal calprotectin concentration at different dilution ratios.

3. The method according to claim 1, characterized in that, The process of synthesizing measurement variance and calculating observation gain based on the observation data package includes: Based on the fecal calprotectin concentrations at the three locations, event-level fecal calprotectin observations are generated. These event-level fecal calprotectin observations are obtained by weighted fusion of the fecal calprotectin concentrations at the three locations according to the weights determined by the dilution consistency index, the uncertainty of immune response kinetics, and the uncertainty of spatial heterogeneity, and are used to update the latent state of inflammatory activity.

4. The method according to claim 1, characterized in that, The measurement variance synthesized based on the observation data packets includes: The measurement variance is obtained by combining the uncertainty component corresponding to the dilution consistency index, the uncertainty of the immune response kinetics, the uncertainty of spatial heterogeneity, and the basic noise of the detection platform.

5. The method according to claim 1, characterized in that, Calculating the observation gain includes determining the observation gain based on the measurement variance, and decreasing the observation gain as the measurement variance increases; Updating the latent state of inflammatory activity includes weighted fusion updating of the latent state of inflammatory activity corresponding to the previous defecation event and the event-level fecal calprotectin observation value corresponding to the current defecation event based on the observation gain. The generation of the trend slope involves calculating the trend based on the latent state of inflammatory activity corresponding to multiple defecation events in chronological order. Generating the cumulative offset involves recursively accumulating the difference between the updated latent state of inflammatory activity and the individual dynamic threshold, and limiting the cumulative offset to zero or a positive value. The generation of individual dynamic thresholds involves determining individual dynamic thresholds based on the central tendency and dispersion of the latent state of inflammatory activity corresponding to historical defecation events.

6. The method according to claim 1, characterized in that, Generating event codes based on the status data packet includes: Based on the dilution consistency index, the immune response kinetic uncertainty, the spatial heterogeneity uncertainty, the measurement variance, and the observation gain, anomaly types are determined, and the anomaly types are mapped to event codes used to indicate dilution consistency retesting, immune response kinetic retesting, hemorrhage retesting, spike recovery, and site expansion in the proof-of-conformity process.

7. The method according to claim 6, characterized in that, Executing the disproving process based on the event code to generate a disproving result data packet and detection design instructions includes: Based on the immunoassays at different dilution ratios, an additional immunoassay at a different dilution ratio was added, and the dilution consistency index was recalculated. The extract was subjected to immunoassay again and the immunoreaction time series signal was collected to obtain the immunoreaction kinetic uncertainty again. Fecal occult blood testing was performed on fecal samples from the same defecation event to generate bleeding retest results; the extract was aliquoted and calprotectin standard was added to the spiked sample, and the spiked recovery rate was obtained based on the immunoassay results before and after spiking; The fecal samples from the same defecation event were expanded from three sites to five sites, and each site was lysed and extracted to obtain expanded extracts. The five expanded extracts were mixed to obtain a combined extract, and one of the expanded extracts was selected as the fixed site extract. The spatial heterogeneity uncertainty was recalculated based on the immunoassay results of the combined extract and the fixed site extract. Based on the retest results and verification results, the counter-evidence result data package is generated, and the detection design instructions are generated.

8. The method according to claim 1, characterized in that, Based on the accumulated evidence from the status data packet and the counter-evidence result data packet, the graded early warning result is output, including: Evidence increments are generated based on the difference between the latent state of inflammatory activity and the individual dynamic threshold, the trend slope, the cumulative offset, and the measurement variance. Evidence increment weights are determined based on the data package of the evidence of contradiction. Cumulative evidence is recursively calculated based on the evidence increments and evidence increment weights, and graded early warning results are output based on the cumulative evidence boundaries. The graded early warning results include low-risk, medium-risk, and high-risk levels.

9. The method according to claim 8, characterized in that, Updating the individual parameter set after obtaining the relapse outcome label includes: Based on the relapse outcome label, the individual dynamic threshold generation rule, cumulative evidence boundary and observation gain calculation rule are calibrated and updated, and the basic noise of the detection platform is updated based on the relapse outcome label for synthesizing measurement variance.

10. A fecal calprotectin-based early warning system for recurrence of inflammatory bowel disease, used to implement the fecal calprotectin-based early warning method for recurrence of inflammatory bowel disease as described in any one of claims 1-9, characterized in that, The system includes: The observation generation module is used to collect fecal samples from three locations during the same defecation event, lyse and extract the extract, perform immunoassay on the extract at different dilution ratios to obtain the concentration of fecal calprotectin at the site, and generate an observation data package containing dilution consistency index, immunoreaction kinetic uncertainty and spatial heterogeneity uncertainty. The state update module is used to synthesize measurement variance and calculate observation gain based on observation data packets, update the latent state of inflammatory activity based on observation gain, and generate trend slope, cumulative offset and individual dynamic threshold to form state data packets. The disproving trigger module is used to generate event codes based on the status data packet and execute the disproving process based on the event codes to generate a disproving result data packet and detection design instructions. The evidence warning module is used to accumulate evidence based on status data packets and counter-evidence result data packets to output graded warning results, and to update the individual parameter set after obtaining the relapse outcome label.

Images