Multi-parameter monitoring system for evaluating ulcerative colitis intestinal barrier function

Abstract

The invention relates to the field of ulcerative colitis, and discloses a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis, which is used for realizing the goal of automatically adjusting a treatment scheme according to a monitoring result by constructing a closed-loop feedback treatment system. According to the invention, by designing the nanoprobe of which the surface is modified with a specific binding element and combining with a swallowing capsule endoscopy, real-time imaging and molecular level dynamic tracking of the intestinal barrier are realized, genetically engineered bacteria are constructed to quantify intestinal permeability abnormality and local inflammation level, and an intestinal barrier function damage score and a disease progress prediction result are output; individual monitoring is achieved through transfer learning, ultrasonic elastography, optical coherence tomography, wearable equipment for measuring intestinal electric field impedance spectroscopy and other technologies are adopted, an intestinal wall mechanics-electrophysiology composite parameter set is obtained, and the evaluation accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of ulcerative colitis, and in particular to a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis. Background Art

[0002] Ulcerative colitis (UC) is a chronic inflammatory bowel disease characterized by persistent inflammation in the intestinal mucosa and submucosa, leading to impaired intestinal barrier function. The intestinal barrier is a crucial line of defense for maintaining a stable internal environment. Its dysfunction can lead to bacterial translocation, immune dysregulation, and systemic inflammation, severely impacting patients' quality of life. Therefore, accurate assessment of intestinal barrier function is crucial for the early diagnosis, disease monitoring, and treatment efficacy evaluation of UC.

[0003] Traditional methods for assessing intestinal barrier function rely primarily on clinical symptoms, endoscopic examinations, and histological biopsies, but these methods have numerous limitations. Clinical symptoms often lag behind pathological changes, endoscopic examinations are limited by patient acceptance and operational difficulty, and histological biopsies are invasive and difficult to monitor dynamically. With the rapid advancement of biomedical technology, multi-parameter, non-invasive monitoring methods have become a research hotspot.

[0004] Deficiencies in the existing technology:

[0005] Limitations of single-modality monitoring: Traditional methods often only provide single physiological or pathological information. For example, endoscopy primarily observes intestinal mucosal morphology but cannot directly reflect molecular changes in the intestinal barrier. Single biochemical indicators (such as inflammatory factors in blood or feces) are unlikely to fully reflect the complex state of the intestinal barrier.

[0006] Lack of real-time dynamic monitoring: Most existing methods can only provide static, intermittent monitoring results and cannot track the dynamic changes in intestinal barrier function in real time. For chronic, recurrent diseases such as UC, real-time dynamic monitoring is crucial for timely adjustment of treatment plans.

[0007] Insufficient individualized assessment: The degree of intestinal barrier damage and the rate of disease progression vary significantly among UC patients, making it difficult to accurately assess individual patients using existing methods. There is a lack of personalized monitoring models that integrate patient-specific microbiome, metabolic, and physiological parameters.

[0008] The present invention proposes a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis to improve the above-mentioned deficiencies. Summary of the Invention

[0009] The present invention provides a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis, which is used to automatically adjust the treatment plan according to the monitoring results by constructing a closed-loop feedback treatment system.

[0010] The first aspect of the present invention provides a multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis, and the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis comprises: designing a nanoprobe with a surface-modified specific binding element, the specific binding element targeting the tight junction protein Occludin and / or ZO-1, and simultaneously loading a catalytically active nanozyme, using the nanoprobe to synchronously collect multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals, and combining a swallowable capsule endoscope to achieve real-time imaging and molecular-level dynamic tracking of the intestinal barrier, thereby obtaining a probe-based multimodal data set; constructing a genetically engineered bacterium, which carries a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio, and a sensing module for the inflammatory factor TNF-α or IL-6, and detecting the fluorescence reporter signal or electrical signal output of the genetically engineered bacterium through a fecal sample to obtain an engineered bacterium detection signal. Measure data, quantify abnormal intestinal permeability and local inflammation levels, and obtain quantitative data on intestinal permeability and inflammation; construct a multidimensional database based on probe-based multimodal data sets, engineered bacteria detection data, metagenomic sequencing bacterial community composition data, and clinical biochemical indicators; use graph neural networks or Transformer models to parse the nonlinear associations of the multidimensional database, and output intestinal barrier function damage scores and disease progression prediction results; adapt the pre-trained model to individual patient characteristics through transfer learning to achieve personalized monitoring and obtain a personalized prediction model; use ultrasound elastography to quantify intestinal wall hardness, and use optical coherence tomography to detect the microstructural integrity of the mucosal layer. Use wearable devices to measure the intestinal electric field impedance spectrum to invert the permeability changes between intestinal epithelial cells, and obtain a set of intestinal wall mechanical-electrophysiological composite parameters, which are input into the personalized prediction model, combined with the real-time impedance spectrum to calibrate the model prediction value, and output the corrected disease progression risk level.

[0011] Optionally, in a first implementation of the first aspect of the present invention, the nanoprobe adopts a core-shell structure, the core of which is a SrDy2O4 nanozyme with CT / MRI dual-modal imaging function, the shell is modified with polyethylene glycol and a single-chain antibody targeting Occludin / ZO-1, and the surface of the nanozyme is loaded with the electrochemically active molecule methylene blue; the swallowable capsule endoscope integrates a miniature spectrometer and a wireless transmission module, triggers probe signal acquisition at preset time intervals in the intestine, and transmits multimodal data in real time to an in vitro analysis terminal through near-field communication.

[0012] Optionally, in a second implementation of the first aspect of the present invention, the method for synchronously detecting the fluorescent reporter signal and the electrical signal: using time-resolved fluorescence detection technology, distinguishing the lactulose / mannitol ratio from the inflammatory factor signal through dual fluorescence channels; using screen-printed electrodes to detect the impedance changes of the electroactive biofilm of the engineered bacteria, and combining differential pulse voltammetry to quantify the electrical signal intensity; the stool sample processing method: after enriching the engineered bacteria by centrifugation, using a portable smartphone fluorescence imaging module or a micro electrochemical workstation to collect signals, and using a microfluidic chip to separate the bacterial flora from the host cells; the quantification method includes: calculating the intestinal permeability index based on the standard curve of the lactulose / mannitol ratio and fluorescence intensity; and outputting the inflammatory activity score through a nonlinear regression model of the electrical signal intensity and TNF-α / IL-6 concentration.

[0013] Optionally, in a third implementation of the first aspect of the present invention, a multidimensional database is constructed: the probe multimodal data set is temporally and spatially aligned to extract the fluorescence signal decay rate, electrochemical catalytic efficiency and SERS spectral peak shift characteristic values; K-mer frequency encoding and functional pathway enrichment analysis are performed on the metagenomic sequencing data to construct a microbiota-metabolic function association matrix; and clinical biochemical indicators are time-series normalized, including the fecal calprotectin concentration fluctuation curve and the dynamic changes of the serum CRP / IL-6 ratio.

[0014] Optionally, in a fourth implementation of the first aspect of the present invention, multimodal biophysical parameters are collected synchronously: high-frequency ultrasound elastography is used to obtain the intestinal wall shear wave velocity and Young's modulus distribution map, and the intestinal wall hardness gradient is quantified; frequency domain optical coherence tomography is used to scan the mucosal layer to extract goblet cell density, mucus layer thickness and crypt structure integrity index; an alternating electric field is applied through a wearable flexible electrode array, and the dynamic changes of intestinal epithelial cell resistance and transmembrane capacitance are inverted based on the Cole-Cole impedance model; composite parameter set construction and feature fusion: the intestinal wall shear wave velocity, goblet cell density and intestinal epithelial cell resistance value are spatiotemporally aligned to generate a three-dimensional intestinal segment mechanical-electrophysiological fusion thermal map;

[0015] Define the composite parameter barrier stress index:

[0016]

[0017] Among them, SWV max Indicates the maximum value of the intestinal wall shear wave velocity, which reflects the hardness or elasticity of the intestinal wall tissue and can be measured in real time by ultrasound elastography. min The minimum value of the electrical resistance between intestinal epithelial cells is an important indicator for evaluating intestinal barrier function and reflects the integrity of the tight junctions between epithelial cells. α and β are adaptive weight coefficients based on the patient's historical data.

[0018] Dynamic calibration of personalized prediction model: Establish an impedance spectrum-histopathology correlation matrix: Through the solution of the finite element inverse problem, map the real-time impedance phase angle to the opening and closing probability of tight junctions between epithelial cells; Use the extended Kalman filter algorithm to dynamically correct the prediction results of the Transformer model with real-time impedance data as the observed values; The risk correction factor is output by the calibrated model:

[0019]

[0020] Among them, the predicted TEER refers to the resistance value between intestinal epithelial cells predicted by the personalized prediction model; the measured TEER refers to the resistance value between intestinal epithelial cells obtained by actual measurement; d(BSI) / dt represents the change rate of BSI with time, reflecting the dynamic change of intestinal barrier function;

[0021] Risk level stratification output: According to the RCF value and the BSI threshold interval, the disease progression risk is divided into three levels: Level I: RCF≤0.2 and BSI<30; Level II: 0.2<RCF≤0.5 or 30≤BSI<60; Level III: RCF>0.5 or BSI≥60; The risk level is logically OR-operated with the degree of OCT mucosal structure damage, and if more than 50% of the crypts are deformed, it will be automatically upgraded by one level to generate the final corrected risk level.

[0022] Optionally, in the fifth implementation manner of the first aspect of the present invention, it further includes the verification of the combined dynamic regulation of gut microbiota-metabolome: Using the causal inference algorithm based on Bayesian network, with phage intervention as the exogenous variable, analyze the causal chain of microbiota species, metabolites and barrier function parameters;

[0023] Define the contribution degree of the key regulatory pathway as CRC:

[0024]

[0025] Among them, represents the influence degree of metabolite concentration change on BSI; represents the influence degree of microbiota abundance change on metabolite concentration;

[0026] Screen the gut microbiota-metabolite pairs with CRC>0.5 as treatment targets; Prediction of the microbiota-host metabolism model: Construct a genome-scale metabolic network model, integrate patient-specific microbiota and host intestinal cell metabolic fluxes; Predict the short-chain fatty acid synthesis rate and mucin secretion amount after targeted regulation through flux balance analysis; If the predicted SCFA synthesis rate increases by ≥30% and the error from the measured value is <15%, it is determined that the microbiota intervention plan is effective.

[0027] Optionally, in the sixth implementation of the first aspect of the present invention, it also includes a multimodal monitoring device collaboration and real-time decision-making system: wearable-implantable device collaborative architecture: stretchable electronic skin: integrated flexible ultrasonic transducer for detecting intestinal wall hardness, impedance sensor for TEER monitoring and optical microlens array for OCT signal acquisition, adhered to the abdominal skin to achieve continuous monitoring; smart capsule endoscope: loaded with the nanoprobe and micro metabolite sensor described in claim 2, detects the butyrate concentration in the cavity, stays in the intestinal lesion site for ≥72 hours, and is powered by a biofuel cell; subcutaneously implanted microprocessor: receives data from multiple devices, executes a lightweight version of the personalized prediction model described in claim 4, predicts the risk level and triggers the alarm threshold; edge-cloud hybrid computing framework: edge end: on the implanted microprocessor A lightweight GNN model after knowledge distillation is deployed in the cloud to calculate the barrier function trend index in real time; on the cloud: after receiving the full amount of data, a high-precision Transformer model is reconstructed to generate weekly optimization suggestions for the barrier repair path, including the timing of microbial transplantation and drug dosage adjustment; multi-center data security sharing and model incremental learning are achieved through blockchain smart contracts; closed-loop feedback treatment system: when BFTI exceeds the threshold for 3 consecutive hours, at least one of the following interventions is automatically activated: releasing sustained-release nanoprobes loaded with anti-TNF-α antibodies through smart capsules; triggering transcutaneous electrical stimulation of wearable devices at a frequency of 5Hz to promote goblet cell regeneration; sending signals to the associated drug pump to adjust the drug release rate of the mesalazine microneedle patch; automatically starting the metabolic flux prediction model described in claim 6 after the intervention to evaluate the efficacy and perform cyclic optimization.

[0028] The second aspect of the present invention provides a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis, and the multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis includes: an acquisition module for designing a nanoprobe with a surface-modified specific binding element, the specific binding element targeting the tight junction protein Occludin and / or ZO-1, and simultaneously loading a catalytically active nanozyme, and using the nanoprobe to synchronously acquire multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals, and combining with a swallowable capsule endoscope to achieve real-time imaging and molecular-level dynamic tracking of the intestinal barrier, to obtain a probe-based multimodal data set; a processing module for constructing a genetically engineered bacterium, which carries a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio, and a sensing module for the inflammatory factor TNF-α or IL-6, and detects the fluorescence reporter signal or electrical signal output of the genetically engineered bacteria through a fecal sample to obtain the engineered bacteria detection signal. The system measures data, quantifies abnormal intestinal permeability and local inflammation levels, and obtains quantitative data on intestinal permeability and inflammation; the analysis module is used to construct a multidimensional database based on probe-based multimodal data sets, engineered bacteria detection data, metagenomic sequencing bacterial community composition data, and clinical biochemical indicators; the graph neural network or Transformer model is used to analyze the nonlinear correlation of the multidimensional database, and output the intestinal barrier function damage score and disease progression prediction results; the pre-trained model is adapted to individual patient characteristics through transfer learning to achieve personalized monitoring and obtain a personalized prediction model; the calibration module is used to quantify intestinal wall hardness using ultrasound elastography, and to detect the microstructural integrity of the mucosal layer using optical coherence tomography. The intestinal electric field impedance spectrum is measured by wearable devices to invert the permeability changes between intestinal epithelial cells, and the intestinal wall mechanical-electrophysiological composite parameter set is obtained, which is input into the personalized prediction model, combined with the real-time impedance spectrum calibration model prediction value to output the corrected disease progression risk level.

[0029] A third aspect of the present invention provides a multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis performs the above-mentioned multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis.

[0030] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein, which, when executed on a computer, enables the computer to execute the above-mentioned multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis.

[0031] The technical solution provided by the present invention has the following beneficial effects:

[0032] By modifying the specific binding elements targeting tight junction proteins (Occludin and / or ZO-1) and loading catalytically active nanozymes, real-time imaging and dynamic tracking of the intestinal barrier at the molecular level can be achieved, and multimodal signals such as fluorescence, electrochemistry, and oxidative stress catalytic reactions can be simultaneously collected.

[0033] The biosensor module carries a biosensor that responds to intestinal permeability markers (lactulose / mannitol ratio) and inflammatory factors (TNF-α or IL-6) and quantifies abnormal intestinal permeability and local inflammation levels through stool sample testing.

[0034] Integrate probe multimodal datasets, engineered bacteria detection data, metagenomic sequencing bacterial composition data and clinical biochemical indicators, use graph neural networks or Transformer models to analyze nonlinear associations, and output intestinal barrier function damage scores and disease progression prediction results.

[0035] Transfer learning is used to adapt pre-trained models to individual patient characteristics, enabling personalized monitoring. Combining ultrasound elastography, optical coherence tomography, and intestinal electric field impedance spectroscopy (IED) measurements using wearable devices, a composite set of intestinal wall mechanical and electrophysiological parameters is generated and fed into a personalized prediction model for dynamic calibration.

[0036] Through phage-directed regulation and metabolomics feedback, we construct a cross-omics causal network, screen key regulatory pathways, and predict metabolic changes after targeted regulation. We integrate wearable and implantable devices and build an edge-cloud hybrid computing framework to achieve a closed-loop feedback therapy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Schematic diagram of an embodiment of a multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to an embodiment of the present invention;

[0038] Figure 2 Schematic diagram of another embodiment of a multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to an embodiment of the present invention;

[0039] Figure 3 Schematic diagram of a multi-parameter monitoring system for evaluating intestinal barrier function in ulcerative colitis according to an embodiment of the present invention;

[0040] Figure 4 Schematic diagram of an embodiment of a multi-parameter monitoring device for evaluating intestinal barrier function in ulcerative colitis according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] Embodiments of the present invention provide a multi-parameter monitoring system for evaluating intestinal barrier function in ulcerative colitis, which automatically adjusts treatment plans based on monitoring results by establishing a closed-loop feedback treatment system. In the specification and claims of the present invention, and in the accompanying drawings, the terms "first," "second," "third," "fourth," and so on (if any) are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence.

[0042] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis in the present invention includes:

[0043] 101. Design a nanoprobe with a surface-modified specific binding element that targets tight junction proteins occludin and / or ZO-1 and simultaneously loads catalytically active nanozymes. Use the nanoprobe to simultaneously collect multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals. Combined with a swallowable capsule endoscopy, real-time imaging and molecular-level dynamic tracking of the intestinal barrier are achieved, resulting in a probe-based multimodal dataset.

[0044] It is understood that the execution subject of the present invention can be a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0045] It should be noted that during the design and preparation of the nanoprobe, the core structure used gold nanoparticles (AuNPs, 50 nm in diameter) as a carrier, which exhibits high biocompatibility and surface plasmon resonance, facilitating fluorescence signal enhancement. Surface modification involved the specific binding element, which was conjugated via a thiol bond to an anti-occludin monoclonal antibody (10 μg / mL) and an anti-ZO-1 aptamer (sequence: 5'-ATC CTA TCA CTA ATT AGG GTA GGG CGG GTT GGG-3'). The binding efficiency was verified to be 97% (ELISA assay).

[0046] Nanozyme loading: Platinum nanoparticles (PtNPs, 5 nm in diameter) were covalently modified on the surface of AuNPs to mimic peroxidase activity, catalyzing the conversion of H2O2 to hydroxyl radicals (·OH) with a catalytic efficiency three times that of natural horseradish peroxidase (kinetic parameters: Km = 0.2 mM, Vmax = 2.5 μM / min). Fluorescent labeling: The near-infrared fluorescent dye Cy5.5 (emission wavelength 710 nm) was attached to the surface of AuNPs to avoid intestinal background interference.

[0047] In vitro validation experiment, targeted binding experiment: The nanoprobe was co-incubated with human colon epithelial cells (Caco-2). Immunofluorescence showed that the occludin / ZO-1 binding signal in the inflammation model group (TNF-α induced) was 5 times stronger than that in the normal group (fluorescence intensity: 200AU in the normal group vs. 1000AU in the inflammation group).

[0048] Catalytic activity testing: In a simulated intestinal environment (pH 6.8), the nanoprobe catalyzed the production of ·OH from 1 mM H₂O₂. The absorbance (650 nm) increased by 0.8 over 5 minutes as measured by the TMB colorimetric assay, corresponding to a H₂O₂ decomposition rate of 0.16 mM / min. Animal model validation: DSS-induced ulcerative colitis mouse model (n=10) and healthy mice (n=5) were used as controls. Probe administration: The nanoprobe (dose: 1 mg / kg) was administered via gavage and assayed 2 hours later.

[0049] The swallowable capsule endoscopy system integrates hardware design: a miniature fluorescence imaging module equipped with a 710nm excitation light source and a CMOS sensor (resolution 1280×720). The electrochemical sensor uses a three-electrode system (working electrode modified with carbon nanotubes, detection range 1-100μM H₂O₂). Data transmission uses Bluetooth Low Energy (BLE) to transmit data to an external receiver in real time. The capsule endoscope collects fluorescence images, electrochemical signals (current values), and catalytic reaction products (H₂O₂ concentration) every 10 seconds for continuous monitoring for 6 hours.

[0050] Multimodal signal acquisition and results, fluorescence imaging: Inflammation areas showed high-intensity Cy5.5 signals (average grayscale value 950AU vs. 200AU in normal areas), which were negatively correlated with pathological sections (occludin expression decreased by 60%) (r = -0.89). Electrochemical signals: The H2O2 concentration in the inflamed intestine increased to 25μM (8μM in the normal group), and the current response value increased from 50nA to 180nA. Catalytic reaction kinetics: The catalytic efficiency of the nanozyme increased by 2 times in the inflamed area (catalytic rate 0.32mM / min vs. 0.16mM / min in the normal group), reflecting the level of local oxidative stress.

[0051] Multimodal dataset integration: A signal-Disease Activity Index (DAI) model was established using machine learning (random forest algorithm), with a prediction accuracy of 92% (AUC = 0.94). Histological verification: Immunohistochemistry showed a significant correlation between ZO-1 expression and the decrease in electrochemical signal (r = 0.85, p < 0.01). Biocompatibility: The probe was metabolized and excreted by the intestine within 72 hours, and there was no significant increase in serum inflammatory factors (IL-6, TNF-α) (p > 0.05). Capsule excretion rate: All animals naturally excreted the capsule within 24 hours, with no retention.

[0052] 102. Construct genetically engineered bacteria that carry a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio and a sensing module for the inflammatory factors TNF-α or IL-6. Fluorescent reporter signals or electrical signal outputs from the genetically engineered bacteria are detected in fecal samples to obtain engineered bacteria detection data, quantify intestinal permeability abnormalities and local inflammation levels, and obtain quantitative intestinal permeability and inflammation data.

[0053] It should be noted that the chassis strain for the design and construction of genetically engineered bacteria is: non-pathogenic Escherichia coli Nissle1917 (EcN) is selected as the host, which has the ability to colonize the intestine and is safe.

[0054] Dual-function biosensor module: Intestinal permeability sensing module: Lactulose response: A lactulose-specific promoter (PlacF, derived from a lactose operon mutant with a lactulose response threshold of 0.1 mM) was designed to drive the expression of red fluorescent protein (RFP, emission wavelength 610 nm). Mannitol response: A mannitol-inducible promoter (PmtlA, response threshold 0.05 mM) was used to drive the expression of green fluorescent protein (GFP, emission wavelength 510 nm). Signal ratio calculation: The lactulose / mannitol ratio (L / M) was calculated by the RFP / GFP fluorescence intensity ratio.

[0055] Inflammatory factor sensing module: TNF-α sensing: A chimeric receptor was constructed by fusing a TNF-α single-chain antibody (scFv) to a transmembrane histidine kinase (EnvZ). TNF-α binding activates the downstream OmpC promoter (PompC), driving the expression of blue fluorescent protein (BFP, emission wavelength 450 nm) (detection limit: 1 ng / mL). IL-6 sensing: An IL-6 aptamer (sequence: 5'-GGC AGC TCGTGC CGA TTC GAA TTC GTC GAC-3') was designed and combined with a riboswitch to regulate the expression of violet fluorescent protein (VFP, emission wavelength 430 nm) (detection limit: 0.5 ng / mL).

[0056] In vitro validation experiment, permeability module test: The engineered bacteria were cultured in simulated intestinal culture medium (containing different concentrations of lactulose / mannitol): at 0.5mM lactulose, the RFP intensity reached 5000AU; at 0.3mM mannitol, the GFP intensity reached 4800AU. When the L / M ratio was 2 (simulating increased permeability), the RFP / GFP fluorescence ratio was 3.2 (linear range 0.5-5, R 2 =0.98).

[0057] Inflammation module testing: When TNF-α (10 ng / mL) is added, the BFP intensity rises to 8000 AU within 4 hours (background <500 AU); when IL-6 (5 ng / mL) is added, the VFP intensity reaches 6500 AU (background <300 AU). Cross-reactivity testing: The TNF-α module is unresponsive to IL-6 (signal change <5%), and vice versa. Animal model validation: Experimental animals: DSS-induced ulcerative colitis mice (n=12) and healthy controls (n=6).

[0058] Engineering bacteria administration: Oral administration of an engineered bacterial suspension (1×10^9 CFU / mouse) once daily for 3 days. Fecal sample collection and analysis: Feces were collected on day 3, lysed, and analyzed for fluorescence signals by flow cytometry. Results: The L / M fluorescence ratio in the inflammation group was 4.8±0.5 (1.2±0.3 in the control group), highly correlated with the conventional urine L / M ratio (4.5±0.6) (r=0.92, p<0.001). TNF-α module signal: BFP intensity was 7200±800 AU (800±200 AU in the control group), which correlated with the serum ELISA result (TNF-α=9.2±1.1 ng / mL) (r=0.88). IL-6 module signal: VFP intensity was 5800±600 AU (400±150 AU in the control group), which correlated with tissue IL-6 levels (6.8±0.9 ng / g) (r=0.85).

[0059] Detection equipment and signal reading, portable stool detector: integrated multi-channel fluorescence detection module (excitation / emission wavelength covers 450-610nm), detection limit: 10^3 CFU / g feces. Electrochemical backup module: for non-fluorescent scenarios, engineered bacteria express horseradish peroxidase (HRP), catalyze TMB color development, and quantify inflammatory factors through current signals (0-100nA) (sensitivity: 0.1ng / mL). Data analysis and verification, signal-pathology association model: through multivariate regression analysis, the L / M fluorescence ratio combined with TNF-α / BFP signal predicts the disease activity index (DAI) with an accuracy of 89% (AUC = 0.91).

[0060] Histological verification: Decreased occludin expression in colonic tissue was negatively correlated with increased L / M fluorescence ratio (r=-0.84, p<0.01). TNF-α / BFP signal was positively correlated with intestinal pathological score (Histological Activity Index, HAI) (r=0.79, p<0.01).

[0061] Safety assessment, colonization and clearance of engineered bacteria: After three days of colonization in the intestine, the bacteria were cleared by antibiotics (ciprofloxacin) with an efficiency of >99.9%. Biosafety: The engineered bacteria did not induce a systemic inflammatory response (serum IL-6 and TNF-α levels were no different from those in the control group, p>0.05).

[0062] 103. Construct a multidimensional database based on probe-based multimodal datasets, engineered bacteria detection data, microbial composition data from metagenomic sequencing (original input source: fecal sample metagenomic sequencing), and clinical biochemical indicators (original input source: routine blood / fecal testing). Use graph neural networks or Transformer models to analyze nonlinear associations in the multidimensional database and output intestinal barrier function damage scores and disease progression prediction results. Use transfer learning to adapt pre-trained models to individual patient characteristics, enabling personalized monitoring and obtaining personalized prediction models.

[0063] It should be noted that the multi-dimensional database construction and data sources and preprocessing are: nanoprobe multimodal data (from capsule endoscopy): fluorescence signal (Cy5.5 grayscale value: 0-1000 AU), electrochemical signal (H2O2 concentration: 1-100 μM), catalytic reaction rate (0.1-0.5 mM / min). Data volume: 24 hours of monitoring per patient generates 8640 data points (10 seconds / time).

[0064] Genetically engineered bacteria detection data (from stool samples): L / M fluorescence ratio (range 1-10), TNF-α / BFP intensity (0-10,000 AU), IL-6 / VFP intensity (0-8,000 AU). Metagenomic sequencing data (from stool samples): 16S rRNA sequencing (Illumina NovaSeq) was used to analyze bacterial composition (e.g., Bacteroidetes abundance decreased to 15 ± 5% vs. 35 ± 8% in the healthy group).

[0065] Key functional genes: The expression of tight junction protein degradation gene (clbA) was upregulated 3 times.

[0066] Clinical biochemical indicators (blood / feces): serum CRP (0-50 mg / L), fecal calprotectin (50-2,000 μg / g), erythrocyte sedimentation rate (ESR: 1-100 mm / h).

[0067] Database Architecture: A graph database (Neo4j) is used to store data. Nodes include patient ID, probe signal, microbial species, inflammatory factors, and biochemical markers; edges represent association weights (e.g., microbial-inflammation correlation coefficient). Sample dataset: 500 patients (300 with ulcerative colitis and 200 healthy controls), totaling 1.2 TB of data.

[0068] Model design and training, Graph Neural Network (GNN) architecture: Node feature encoding: Probe signal (3D), microbial abundance (50D), and biochemical markers (5D) are embedded into 128-dimensional vectors. Graph structure construction: A heterogeneous graph of patient-probe-microbial-biochemical markers is constructed, and edge weights are selected using mutual information (MI>0.3). Model parameters: 3-layer GraphSAGE, with an attention mechanism aggregating neighbor information, outputting a gut barrier damage score (0-10) and a 3-year probability of disease progression (0-1).

[0069] Transformer model architecture: Input sequence: Time series probe data (8640 points) + static data (microbial flora, biochemical indicators), concatenated into a 768-dimensional embedding. Self-attention mechanism: 12 attention heads, calculating the contribution of different signals to the score (e.g., H2O2 concentration accounts for 25%). Output layer: Jointly trained for regression (scoring) and classification (progression to severe colitis).

[0070] Training and Validation: The dataset was divided into a training set (400 cases), a validation set (50 cases), and a test set (50 cases). Performance: GNN model: Intestinal injury score prediction MAE = 0.8 (error 0.5 in the normal group, 1.1 in the inflammation group), disease progression AUC = 0.93. Transformer model: MAE = 0.7, AUC = 0.95.

[0071] Transfer learning enables personalized predictions. Pre-trained model: A basic GNN / Transformer model trained on 500 patient data. Personalized adaptation strategy: Dynamic fine-tuning layer: A new patient-specific adaptation layer (2 fully connected layers) is added, and only the adaptation layer parameters are updated (learning rate 1e-4). Data augmentation: Synthetic data (GAN-generated similar patient features) is used to expand the individual data to 100 samples. Effectiveness validation: Across 10 new patients, the transferred model's prediction score error decreased by 40% (MAE decreased from 1.2 to 0.7), and the personalized AUC increased to 0.89.

[0072] Multi-dimensional correlation analysis revealed key associations (SHAP value analysis): Microbial-probe signal interaction: For every 10% decrease in Bacteroidetes abundance, H2O2 concentration increased by 15% (SHAP value = 0.32). Inflammatory factor-L / M ratio: When TNF-α > 5 ng / mL, L / M ratios > 4 contributed 45% of the total weight. Visualization output: A heat map showed a strong positive correlation between probe signal and microbial clbA gene expression (r = 0.82, p < 0.001).

[0073] Clinical validation and results, compared to the gold standard: The intestinal barrier damage score was significantly correlated with the endoscopic UCEIS score (r = 0.86, p < 0.001). Accuracy of disease progression prediction: Among patients with a model-predicted probability of progression > 0.8 within 3 years, 85% actually developed severe colitis (compared to < 10% in the control group). Clinical decision support: Among patients recommended "anti-TNF-α therapy" by the model, treatment efficacy increased by 30% (compared to the group receiving traditional guidelines).

[0074] System security, explainability, and privacy protection: Data anonymization (differential privacy, ε = 0.1) eliminates the risk of raw data leakage during model inference. Interpretability tools provide patient-specific reports, annotating key drivers (e.g., "high H2O2 concentration + low Bacteroidetes abundance" contributes 60%).

[0075] 104. Ultrasound elastography is used to quantify the hardness of the intestinal wall, combined with optical coherence tomography to detect the microstructural integrity of the mucosal layer. Wearable devices are used to measure the intestinal electric field impedance spectrum to invert the changes in permeability between intestinal epithelial cells, and obtain a set of intestinal wall mechanical-electrophysiological composite parameters. These parameters are input into a personalized prediction model, combined with the real-time impedance spectrum calibration model prediction value, and the corrected disease progression risk level is output.

[0076] It should be noted that the equipment and parameters for multimodal data acquisition and parameter set construction, as well as ultrasound elastography quantification of intestinal wall stiffness, were measured using a high-frequency ultrasound elastography system (15 MHz probe frequency, 0.1 mm resolution) combined with shear wave elastography (SWE) technology to measure the elastic modulus of the colonic wall (in kPa). The healthy reference value is 5.0 ± 1.2 kPa for normal intestinal wall elasticity, while it is significantly elevated in inflamed intestinal wall (15.3 ± 4.5 kPa).

[0077] Data collection: Intestinal ultrasound examinations were performed on patients with ulcerative colitis (n=50) and healthy controls (n=30), and elastic modulus distribution maps of the diseased area (such as the sigmoid colon) were collected.

[0078] Result example: The median elastic modulus of patients in the active stage was 18.6 kPa (range 10-32 kPa), which was positively correlated with the endoscopic severity score (UCEIS) (r=0.78, p<0.001).

[0079] Optical coherence tomography (OCT) to examine mucosal microstructure

[0080] Equipment and parameters: Frequency domain OCT system (central wavelength 1.3 μm, axial resolution 8 μm, lateral resolution 15 μm) was used, with a scanning depth of 2 mm. Quantitative indicators: mucosal crypt density (unit: per mm 2 ), crypt depth (μm), and mucosal layer thickness (μm).

[0081] Data verification: Analysis of inflamed mucosa (biopsy confirmed) showed that the crypt density decreased to 40 ± 12 / mm 2 (Healthy group 85±20 / mm 2 ), the crypt depth decreased to 80±25μm (150±30μm in the healthy group), and was significantly correlated with the pathological score (Geboes score) (r=-0.82, p<0.001).

[0082] Wearable intestinal electric field impedance spectroscopy measurement

[0083] Device design: Flexible wearable electrode array (5×5cm 2 , 16 channels), covers the abdominal colon projection area, and supports multi-band impedance measurement (10Hz-1 MHz). Inversion model: Based on the Cole-Cole equivalent circuit model, calculates the intestinal epithelial cell resistance (unit: Ω·cm 2 ) and transmembrane capacitance (μF / cm 2 ).

[0084] Clinical data: The low-frequency (100Hz) impedance of the healthy group was 120±20Ω·cm 2 , and the inflammation group decreased to 45±15Ω·cm 2 (Reflecting increased permeability.) The impedance spectrum characteristics were significantly correlated with the lactulose / mannitol (L / M) ratio (r=0.75, p<0.001).

[0085] Composite parameter set integration and model input

[0086] Parameter set definition: Mechanical parameters: elastic modulus (kPa), mucosal thickness (μm). Microstructural parameters: crypt density (number / mm 2 ), crypt depth (μm). Electrophysiological parameters: low-frequency impedance (Ω·cm 2 ), transmembrane capacitance (μF / cm 2 ).

[0087] Data normalization: Z-score standardization was used to eliminate dimensional differences.

[0088] Example dataset:

[0089]

[0090] Personalized prediction model construction, model architecture and training, input layer: composite parameter set (6-dimensional vector) + real-time impedance spectrum (100 frequency point data). Core algorithm: Long short-term memory network (LSTM): processes time series impedance data (sampling interval is 10 minutes). Random forest regressor: analyzes the nonlinear relationship between mechanical-microstructural parameters and disease progression. Output layer: short-term (1 week) inflammation risk level (low / medium / high). Long-term (6 months) disease progression probability (0-1). Real-time impedance spectrum calibration, dynamic update strategy: impedance spectrum is collected every 2 hours, and the model input is updated through a sliding window (window size is 24 hours).

[0091] Calibration formula:

[0092] α=0.3 (empirical coefficient), ΔZ is the deviation of the current impedance from the baseline value.

[0093] Model validation, training set: 300 patient data (200 for training and 100 for validation).

[0094] Performance indicators: short-term risk prediction AUC = 0.91 (sensitivity 85%, specificity 89%). Long-term progression prediction MAE = 0.12 (error range ± 10%).

[0095] System integration and clinical validation, with a hardware-software collaborative workflow. Hardware: Ultrasound-OCT integrated endoscopic probe (12mm diameter, suitable for colonoscopy). Wearable impedance monitoring belt (72-hour battery life, Bluetooth data transmission). Software platform: Real-time display of intestinal wall elastic modulus thermal maps, OCT mucosal structure 3D reconstruction, and impedance dynamic curves. Risk level warning interface (red / yellow / green color code).

[0096] Clinical trial results, cohort study (n=80): Among patients (n=25) predicted to be at high risk by the model, 92% experienced endoscopic exacerbation within 3 months (vs. 76% as assessed by clinicians). Impedance calibration reduced prediction error by 35% (MAE from 0.18 to 0.12).

[0097] Typical case analysis, case UC-09: Initial parameters: elastic modulus 16.8 kPa, crypt density 42 / mm 2 , impedance 52Ω·cm 2 → Predicted risk level: "Medium". Real-time impedance monitoring shows a sudden drop to 28Ω·cm 2 (Within 24 hours), the model revised the risk to “high.” Endoscopic review confirmed the expansion of the mucosal ulcer (UCEIS increased from 4 to 6), and intensive treatment was initiated promptly.

[0098] Biosafety: The wearable electrode passed the biocompatibility test (ISO 10993) and showed no skin irritation (mouse model, n=20). Model interpretability: SHAP value analysis showed that low-frequency impedance contributed the most (weighted 42%), followed by elastic modulus (35%).

[0099] Provide risk driver report (such as "impedance decreased by 30% + crypt density <40 / mm 2 " triggering a high-risk warning).

[0100] In this embodiment, a dynamic risk assessment (AUC > 0.9) for ulcerative colitis was achieved using a composite parameter set of ultrasound elastography (to quantify stiffness), optical coherence tomography (OCT) (to assess mucosal integrity), and wearable impedance spectroscopy (to invert permeability) combined with an LSTM-random forest hybrid model. Real-time impedance calibration reduced prediction error by 35%, and the system can be integrated into clinical workflows, providing precise support for early intervention.

[0101] See also Figure 2 Another embodiment of the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis in the embodiments of the present invention includes:

[0102] 201. Design a nanoprobe with a surface-modified specific binding element that targets tight junction proteins occludin and / or ZO-1 and simultaneously loads catalytically active nanozymes. Use the nanoprobe to simultaneously collect multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals. Combined with a swallowable capsule endoscope, real-time imaging and molecular-level dynamic tracking of the intestinal barrier are achieved, resulting in a probe-based multimodal dataset.

[0103] Specifically, the nanoprobe adopts a core-shell structure, with the core being a SrDy2O4 nanozyme with CT / MRI dual-modal imaging capabilities, the shell modified with polyethylene glycol and a single-chain antibody targeting Occludin / ZO-1, and the surface of the nanozyme loaded with the electrochemically active molecule methylene blue; the synchronous acquisition of multimodal signals is specifically as follows: the fluorescence resonance energy transfer signal after the probe binds to the target protein is detected by fluorescence lifetime imaging to quantify the expression level of Occludin / ZO-1; the electrochemical current changes generated by nanozyme-catalyzed hydrogen peroxide are used to monitor the intensity of local oxidative stress in the intestine in real time; surface-enhanced Raman scattering labels are combined to detect conformational changes in tight junction proteins; the swallowable capsule endoscope integrates a miniature spectrometer and a wireless transmission module, which triggers probe signal acquisition at preset time intervals in the intestine, and transmits multimodal data in real time to an in vitro analysis terminal through near-field communication; time-resolved signal processing is performed on the multimodal data set, and the surface and deep barrier damage of the intestinal mucosa are distinguished by the difference between the fluorescence decay curve and the electrochemical response time, thereby improving the spatial resolution to below 50μm.

[0104] It should be noted that the design of the nanoprobe has a core-shell structure: the inner core: uses SrDy2O4 nanozyme with CT / MRI dual-modal imaging function. The nanozyme has high catalytic activity, stability and versatility, can maintain activity under a wide range of temperature and pH conditions, and is suitable for in vivo monitoring. The outer shell: modified with polyethylene glycol (PEG) and a single-chain antibody targeting Occludin / ZO-1. The modification with polyethylene glycol can improve the nanoprobe's anti-protein nonspecific adsorption ability and prolong its circulation time in the body. The single-chain antibody acts as a specific binding element, enabling the nanoprobe to target the tight junction protein Occludin and / or ZO-1. Loading electrochemically active molecules: The electrochemically active molecule methylene blue (MB) is loaded on the surface of the nanozyme to generate electrochemical signals.

[0105] Specific Data: Nanoprobe Particle Size: The average particle size of the nanoprobes was set to 50 nm, as measured by transmission electron microscopy (TEM). Polyethylene glycol molecular weight: A polyethylene glycol with a molecular weight of 2000 was selected for modification. Single-chain antibody (scFv) concentration: During nanoprobe preparation, the single-chain antibody (scFv) concentration was 1 mg / mL.

[0106] Synchronous acquisition of multimodal signals, including fluorescence lifetime imaging: Fluorescence resonance energy transfer (FRET) is used to detect the fluorescence resonance energy transfer signal after the probe binds to the target protein. Using CFP as the donor fluorescent protein and YFP as the acceptor fluorescent protein, changes in CFP fluorescence lifetime are used to quantify occludin / ZO-1 expression levels.

[0107] Specific data: In in vitro experiments, when the probe bound to the target protein, the fluorescence lifetime of CFP decreased from 2.5ns to 2.0ns, indicating that effective fluorescence resonance energy transfer occurred.

[0108] Electrochemical Signaling: Nanozyme-catalyzed Hydrogen Peroxide: The electrochemical current generated by nanozymes catalyzing hydrogen peroxide is used to monitor the intensity of local oxidative stress in the intestine in real time. Measurements are performed using an electrochemical workstation, with a glassy carbon electrode modified with nanoprobes as the working electrode.

[0109] Specific data: Under the condition of hydrogen peroxide concentration of 1mM, the electrochemical current generated by nanozyme catalysis is 10μA.

[0110] Surface-enhanced Raman scattering (SERS): Detecting conformational changes in tight junction proteins: Surface-enhanced Raman scattering (SERS) tags are used to detect conformational changes in tight junction proteins occludin and / or ZO-1. Gold nanoparticles are used as SERS substrates to monitor protein conformational changes by measuring changes in the position and intensity of specific Raman peaks.

[0111] Specific data: In in vitro experiments, when the conformation of tight junction proteins changed, the intensity of the Raman peak at 1000 cm^-1 in the SERS spectrum increased by 2 times.

[0112] Integration and use of swallowable capsule endoscope, capsule endoscope structure: integrated micro-spectrometer and wireless transmission module: The capsule endoscope is integrated with a micro-spectrometer and a wireless transmission module for collecting and transmitting multimodal signals.

[0113] Specific data: The diameter of the capsule endoscope is 11mm, the length is 26mm, and the wireless transmission rate is 1Mbps.

[0114] Signal acquisition and transmission: Preset time interval triggering: Probe signal acquisition is triggered at preset intervals (e.g., every 5 minutes) within the intestine. Near-field communication transmission: Multimodal data is transmitted to an in vitro analysis terminal in real time via near-field communication. Time-resolved signal processing of multimodal datasets: Fluorescence decay curve and electrochemical response time difference: Time-resolved signal processing is performed on multimodal datasets, and the difference between fluorescence decay curve and electrochemical response time is used to distinguish between surface and deep intestinal mucosal barrier damage.

[0115] Specific data: In the surface barrier damage model, the half-life of the fluorescence decay curve is 1.0ns, and the electrochemical response time is 0.5s; in the deep barrier damage model, the half-life of the fluorescence decay curve is 1.5ns, and the electrochemical response time is 1.0s.

[0116] Spatial resolution is improved to below 50 μm: Through the above-mentioned signal processing method, the spatial resolution is improved to below 50 μm, enabling fine imaging of the intestinal barrier and dynamic tracking at the molecular level.

[0117] 202. Construct genetically engineered bacteria that carry a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio and a sensing module for the inflammatory factors TNF-α or IL-6. Fluorescent reporter signals or electrical signal outputs from the genetically engineered bacteria are detected in fecal samples to obtain engineered bacteria detection data, quantify intestinal permeability abnormalities and local inflammation levels, and obtain quantitative intestinal permeability and inflammation data.

[0118] Specifically, the dual-channel biosensor design of genetically engineered bacteria: permeability detection module: based on the luciferase gene regulated by the osmotic pressure response promoter, its expression intensity is positively correlated with the lactulose / mannitol ratio; inflammatory factor detection module: through the engineering transcription factor to control the expression of conductive protein, generate an electrochemical signal related to the TNF-α / IL-6 concentration; fluorescence reporter signal and electrical signal synchronous detection method: using time-resolved fluorescence detection technology, the lactulose / mannitol ratio and inflammatory factor signals are distinguished through dual fluorescence channels; screen-printed electrodes are used to detect the impedance changes of the electroactive biofilm of engineered bacteria, combined with differential pulse voltammetry to quantify the electrical signal intensity; fecal sample processing method: after enriching the engineered bacteria by centrifugation, a portable smartphone fluorescence imaging module or a micro electrochemical workstation is used to collect signals, and the separation of the bacterial community and host cells is achieved through a microfluidic chip; quantification methods include: calculating the intestinal permeability index based on the standard curve of the lactulose / mannitol ratio and fluorescence intensity; outputting the inflammatory activity score through a nonlinear regression model of electrical signal intensity and TNF-α / IL-6 concentration.

[0119] It should be noted that the construction of the genetically engineered bacteria includes a dual-channel biosensor design and a permeability detection module. Principle: An osmotic-responsive promoter regulates luciferase gene expression. When intestinal permeability increases, the lactulose / mannitol ratio rises, osmotic pressure changes, the promoter is activated, and luciferase gene expression is enhanced. Specific design: An osmotic-sensitive promoter is selected and linked to the luciferase gene to create a permeability detection module.

[0120] Inflammatory Factor Detection Module: Principle: Engineered transcription factors control the expression of conductive proteins. When the concentration of inflammatory factors TNF-α or IL-6 increases, the transcription factors are activated, increasing the expression of conductive proteins and generating an electrochemical signal. Detailed Design: Select a transcription factor sensitive to TNF-α or IL-6 and link it to the conductive protein gene to construct an inflammatory factor detection module.

[0121] Preparation of genetically engineered bacteria, integrating the above two modules into a genetically engineered bacterium, and realizing the function of a dual-channel biosensor through genetic engineering technology.

[0122] Synchronous detection method of fluorescence reporter signal and electrical signal, time-resolved fluorescence detection technology: Principle: Luciferase is used to catalyze the fluorescence signal produced by luciferin, and time-resolved fluorescence detection technology is used to distinguish the lactulose / mannitol ratio from the inflammatory factor signal.

[0123] Specific steps: Use a portable smartphone fluorescence imaging module to collect fluorescence signals. Use dual fluorescence channels to detect the fluorescence signals of the permeability detection module and the inflammatory factor detection module respectively.

[0124] Screen-printed electrodes and differential pulse voltammetry: Principle: Use screen-printed electrodes to detect impedance changes in the electroactive biofilm of engineered bacteria, combined with differential pulse voltammetry to quantify the electrical signal intensity.

[0125] The specific steps are as follows: After centrifugation, the engineered bacteria in the fecal sample are concentrated and coated onto a screen-printed electrode. A microelectrochemical workstation is used to apply voltage pulses, measure the current difference, and plot a voltammetric curve. Differential pulse voltammetry is used to quantify the electrical signal intensity, reflecting the concentration of inflammatory factors.

[0126] Fecal Sample Processing Method, Centrifugation Enrichment: Fecal samples are centrifuged to enrich the engineered bacteria. Signal Acquisition: Fluorescence or electrical signals are collected using a portable smartphone fluorescence imaging module or a microelectrochemical workstation. Microbial and Host Cell Separation: Microfluidic chips are used to separate the bacterial and host cells, improving detection accuracy. Quantification Method, Intestinal Permeability Index Calculation: Principle: The intestinal permeability index is calculated based on a standard curve of lactulose / mannitol ratio and fluorescence intensity.

[0127] Specific steps: Prepare standard solutions with different lactulose / mannitol ratios. Measure the fluorescence intensity of the standard solutions and plot a standard curve. Based on the fluorescence intensity of the engineered bacteria in the stool sample, calculate the lactulose / mannitol ratio from a table and calculate the intestinal permeability index.

[0128] Inflammatory activity score output: Principle: The inflammatory activity score is output through a nonlinear regression model of electrical signal intensity and TNF-α / IL-6 concentration.

[0129] Specific steps: Prepare standard solutions with varying TNF-α / IL-6 concentrations. Measure the electrical signal intensity of the standard solutions and establish a nonlinear regression model. Substitute the electrical signal intensity of the engineered bacteria in the stool sample into the model to calculate the TNF-α / IL-6 concentration and output an inflammatory activity score.

[0130] Specific data of the embodiment, construction of genetically engineered bacteria: permeability detection module Correlation coefficient between luciferase gene expression intensity and lactulose / mannitol ratio: R 2 =0.98. Correlation coefficient between the electrical signal intensity of the inflammatory factor detection module and the concentration of TNF-α / IL-6: R 2 =0.95. Fluorescence reporter signal and electrical signal detection: Sensitivity of fluorescent signal detection by the portable smartphone fluorescence imaging module: 10^-9 M. Sensitivity of electrical signal detection by the microelectrochemical workstation: 10^-12 A. Fecal sample processing: Recovery rate of engineered bacteria enriched by centrifugation: 90%. Purity of bacterial flora and host cells separated by microfluidic chip: 95%. Quantitative analysis: Slope of the intestinal permeability index standard curve: 0.85. Coefficient of determination of the nonlinear regression model for inflammatory activity score: R 2=0.97. Actual stool sample test results: Intestinal permeability index was 1.2 (normal range: 0.5-1.0), and inflammatory activity score was 80 (normal range: 0-50).

[0131] 203. Construct a multidimensional database based on probe-based multimodal datasets, engineered bacteria detection data, microbial composition data from metagenomic sequencing (original input source: fecal sample metagenomic sequencing), and clinical biochemical indicators (original input source: routine blood / fecal testing). Use graph neural networks or Transformer models to analyze nonlinear associations in the multidimensional database and output intestinal barrier function damage scores and disease progression prediction results. Use transfer learning to adapt the pre-trained model to individual patient characteristics, enabling personalized monitoring and obtaining a personalized prediction model.

[0132] Specifically, multi-dimensional database construction: spatiotemporal alignment of probe multimodal datasets, extraction of fluorescence signal decay rate, electrochemical catalytic efficiency and SERS spectral peak shift characteristic values; K-mer frequency encoding and functional pathway enrichment analysis of metagenomic sequencing data, construction of microbiome-metabolic function association matrix; time series normalization of clinical biochemical indicators, including fecal calprotectin concentration fluctuation curve and dynamic changes of serum CRP / IL-6 ratio; graph neural network modeling: definition of heterogeneous graph node types: including probe molecular targets, microbial species, metabolites, and clinical indicators; definition of edge relationship weights: initialization of connection strength between nodes based on prior knowledge of intestinal barrier-related biological networks; use of dynamic graph attention mechanism to dynamically update edge weights according to individual patient data to generate barrier function damage scores; Transformer time series prediction model: input layer design: encoding multimodal data into embedding vectors of time segments, embedding dimensions and biomarkers Strong correlation between object categories; introducing biological constraints in the multi-head attention mechanism: applying a temporal continuity loss function to inflammatory factor-related features, and setting a threshold penalty term for microbial abundance mutations; joint prediction of the output layer: generating the probability of disease progression within 3 months and barrier repair priority recommendations based on Monte Carlo Dropout uncertainty quantification; transfer learning and personalized adaptation: pre-training stage: training a general barrier assessment model on an IBD multi-center dataset; fine-tuning stage: freezing the underlying feature extraction layer of the model, and only updating the patient-specific fully connected layer parameters to adapt to the individual microbial baseline and drug response characteristics; small sample optimization: using synthetic data augmentation technology (generating virtual patient trajectories based on generative adversarial networks GAN) to solve the problem of sparse data in the initial stage of individual monitoring; verification and output: comparing the ROC curves of the GNN / Transformer model and the traditional logistic regression model through cross-validation; outputting a visual decision map: labeling key driving factors.

[0133] It should be noted that the construction of a multi-dimensional database includes the processing of probe multimodal data sets, data alignment: the probe multimodal data sets are temporally and spatially aligned to ensure the synchronization of different modal data in time and space. Feature extraction: Fluorescence signal decay rate: Extract the signal attenuation characteristics of the fluorescent probe in the intestine to reflect changes in intestinal permeability. Electrochemical catalytic efficiency: Measure the catalytic efficiency of the electrochemical probe to reflect the level of local inflammation. SERS spectral peak shift characteristic value: Use surface enhanced Raman scattering spectroscopy (SERS) technology to extract the spectral peak shift characteristic value to reflect changes in molecular structure in the intestine.

[0134] Specific data: The fluorescence signal decay rate ranges from 0.8 to 1.2 (dimensionless), and the electrochemical catalytic efficiency is 50 to 150 μA / cm 2 , the characteristic value of SERS spectrum peak shift is 10-20cm -1 .

[0135] Metagenomic sequencing data processing, K-mer frequency encoding: K-mer frequency encoding is performed on metagenomic sequencing data to convert sequence information into numerical features. Functional pathway enrichment analysis: Functional pathway enrichment analysis is performed to construct a microbial community-metabolism function association matrix to reflect the interaction between microbial community and metabolic function.

[0136] Specific data: The K-mer length is 7, the generated K-mer frequency range is 1000-5000 (dimensionless), and the number of key pathways obtained by functional pathway enrichment analysis is 10-30.

[0137] Clinical biochemical index processing and time series normalization: Time series normalization processing is performed on clinical biochemical indicators such as the fluctuation curve of fecal calprotectin concentration and the dynamic changes of serum CRP / IL-6 ratio to eliminate the dimensional differences between data at different time points.

[0138] Specific data: The fluctuation curve of fecal calprotectin concentration ranges from 50 to 200 μg / g, and the dynamic change range of serum CRP / IL-6 ratio is 0.5 to 2.0 (dimensionless).

[0139] Graph neural network modeling, heterogeneous graph node type definition: Node types include probe molecular targets, bacterial species, metabolites, clinical indicators, etc. Edge relationship weights are defined and initialized: Based on prior knowledge of the intestinal barrier biological network, the connection strength between nodes is initialized. Dynamic update: Using a dynamic graph attention mechanism, edge weights are dynamically updated based on individual patient data to generate a barrier function impairment score.

[0140] Specific data: The initial edge weight range is 0.1-1.0 (dimensionless), and the edge weight range after dynamic update is 0.05-1.5 (dimensionless).

[0141] Barrier function impairment score generation, ranging from 0 to 100, with higher scores indicating more severe barrier function impairment. Specific data: A patient's barrier function impairment score was 65. Transformer time series prediction model, input layer design, embedding vector: Encodes multimodal data into time-segment embedding vectors, with the embedding dimension strongly correlated with the biomarker category. Specific data: The embedding dimension is 128. Biological constraints are introduced into the multi-head attention mechanism. Temporal continuity loss function: A temporal continuity loss function is applied to features related to inflammatory factors to ensure temporal consistency of the prediction results. Microbial abundance mutation threshold penalty: A threshold penalty is set for microbial abundance mutation to prevent unreasonable mutations in the prediction results. Joint prediction at the output layer: Disease progression probability: Generates the probability of disease progression within 3 months based on Monte Carlo Dropout uncertainty quantification. Barrier repair priority recommendation: Provides barrier repair priority recommendations based on the prediction results. Specific data: A patient's probability of disease progression within 3 months is 40%. The barrier repair priority recommendation is to prioritize inflammatory factor regulation and microbial balance.

[0142] Transfer learning and personalized adaptation, pre-training phase, dataset: A universal barrier assessment model was trained on a multi-center IBD dataset. Fine-tuning phase, parameter update: The underlying feature extraction layer of the model was frozen, and only the patient-specific fully connected layer parameters were updated to adapt to the individualized microbiome baseline and drug response characteristics. Small sample optimization, synthetic data augmentation: Synthetic data augmentation technology based on generative adversarial networks (GANs) was used to generate virtual patient trajectories to address the issue of data sparsity in the early stages of individual monitoring. Validation and output, cross-validation comparison, ROC curve: The ROC curves of the GNN / Transformer model and the traditional logistic regression model were compared through cross-validation to evaluate model performance.

[0143] Specific data: The AUC value of the GNN / Transformer model is 0.85, and the AUC value of the traditional logistic regression model is 0.75.

[0144] Visual decision map output and key driver annotation: Output visual decision map and annotate key drivers, such as specific bacterial species, metabolites or clinical indicators.

[0145] Specific example: Key driving factors such as high abundance of pathogenic bacteria, abnormal metabolite levels, and elevated concentrations of inflammatory factors are marked in the visual decision map.

[0146] 204. Ultrasound elastography is used to quantify the hardness of the intestinal wall, combined with optical coherence tomography to detect the microstructural integrity of the mucosal layer. Wearable devices are used to measure the intestinal electric field impedance spectrum to invert the changes in permeability between intestinal epithelial cells, and obtain a set of intestinal wall mechanical-electrophysiological composite parameters. These parameters are input into a personalized prediction model, combined with the real-time impedance spectrum calibration model prediction value, and the corrected disease progression risk level is output.

[0147] Specifically, multimodal biophysical parameters are collected simultaneously: high-frequency ultrasound elastography is used to obtain the intestinal wall shear wave velocity and Young's modulus distribution map to quantify the intestinal wall hardness gradient; frequency-domain optical coherence tomography is used to scan the mucosal layer to extract goblet cell density, mucus layer thickness, and crypt structural integrity index; an alternating electric field is applied through a wearable flexible electrode array to invert the dynamic changes of intestinal epithelial cell resistance and transmembrane capacitance based on the Cole-Cole impedance model; composite parameter set construction and feature fusion: the intestinal wall shear wave velocity, goblet cell density, and intestinal epithelial cell resistance values are spatiotemporally aligned to generate a three-dimensional intestinal segment mechanical-electrophysiological fusion thermal map;

[0148] Define the composite parameter barrier stress index:

[0149]

[0150] Among them, SWV max Indicates the maximum value of the intestinal wall shear wave velocity, which reflects the hardness or elasticity of the intestinal wall tissue and can be measured in real time by ultrasound elastography. min The minimum value of the electrical resistance between intestinal epithelial cells is an important indicator for evaluating intestinal barrier function and reflects the integrity of the tight junctions between epithelial cells. α and β are adaptive weight coefficients based on the patient's historical data.

[0151] Dynamic calibration of the personalized prediction model: Establishing an impedance spectrum-histopathology correlation matrix: By solving the inverse finite element problem, the real-time impedance phase angle is mapped to the opening and closing probability of tight junctions between epithelial cells. The extended Kalman filter algorithm is used to dynamically correct the prediction results of the Transformer model using real-time impedance data as observations.

[0152] The calibrated model outputs the risk correction factor:

[0153]

[0154] Among them, predicted TEER refers to the intestinal epithelial cell resistance value predicted by the personalized prediction model; measured TEER refers to the intestinal epithelial cell resistance value obtained by actual measurement; d(BSI) / dt represents the rate of change of BSI over time, reflecting the dynamic changes in intestinal barrier function;

[0155] Risk level stratification output: According to the RCF value and the BSI threshold range, the disease progression risk is divided into three levels: Level I: RCF ≤ 0.2 and BSI < 30; Level II: 0.2 < RCF ≤ 0.5 or 30 ≤ BSI < 60; Level III: RCF > 0.5 or BSI ≥ 60; The risk level is logically OR-operated with the degree of OCT mucosal structure damage. If more than 50% of the crypts are deformed, it will be automatically upgraded by one level to generate the final corrected risk level.

[0156] It should be noted that in the synchronous acquisition of multimodal biophysical parameters, high-frequency ultrasonic elastography includes data acquisition: obtaining the intestinal wall shear wave velocity (SWV) and Young's modulus distribution map through high-frequency ultrasonic elastography. Quantifying the intestinal wall hardness gradient: The SWV range is 5 - 20 m / s, and the Young's modulus range is 10 - 50 kPa. Specific data: For a certain patient, the maximum value of the intestinal wall SWV is 15 m / s, and the maximum value of the Young's modulus is 30 kPa.

[0157] Frequency-domain optical coherence tomography (OCT), data acquisition: Scanning the mucosal layer using frequency-domain OCT to extract the goblet cell density, mucus layer thickness, and crypt structure integrity index.

[0158] Specific data: Goblet cell density: 500 - 1500 cells per square millimeter. Mucus layer thickness: 50 - 200 μm. Crypt structure integrity index: 0.7 - 1.0 (dimensionless).

[0159] Example: For a certain patient, the goblet cell density is 1000 cells / mm 2 , the mucus layer thickness is 100 μm, and the crypt structure integrity index is 0.8.

[0160] Wearable flexible electrode array, data acquisition: Applying an alternating electric field through the wearable flexible electrode array and inversely calculating the dynamic changes of the transepithelial electrical resistance (TEER) and transmembrane capacitance based on the Cole-Cole impedance model. Specific data: The TEER range is 100 - 1000 Ω·cm 2 , and the transmembrane capacitance range is 1 - 10 μF / cm 2 . Example: For a certain patient, the minimum value of TEER is 200 Ω·cm 2 , and the transmembrane capacitance is 5 μF / cm 2 .

[0161] Construction of composite parameter set and feature fusion, spatio-temporal registration: Spatially and temporally registering the intestinal wall shear wave velocity, goblet cell density, and transepithelial electrical resistance value to generate a three-dimensional intestinal segment mechanics-electrophysiology fusion heat map. Specific data (example): On the heat map, the area with higher intestinal wall hardness overlaps with the area with lower TEER, indicating impaired intestinal barrier function in this area.

[0162] Define the composite parameter barrier stress index (BSI)

[0163] Formula:

[0164] Adaptive weight coefficients: α = 0.6, β = 0.4 (adjusted based on patient historical data).

[0165] Specific calculation: Taking a certain patient as an example, SWV_max = 15 m / s, TEER_min = 200 Ω·cm 2 , mucus layer thickness = 100 μm (baseline value is 150 μm).

[0166]

[0167] Dynamic calibration of the personalized prediction model, establishing an impedance spectrum - histopathology correlation matrix. Method: Solving the inverse problem of finite elements to map the real - time impedance phase angle to the opening - closing probability of tight junctions between epithelial cells. Specific data (example): The impedance phase angle range is - 45° to 45°, and the opening - closing probability of tight junctions is 0 - 1 (dimensionless).

[0168] Extended Kalman filter algorithm. Method: Using the real - time impedance data as the observation value to dynamically correct the prediction results of the Transformer model. Specific data (example): The Transformer model predicts TEER to be 250 Ω·cm 2 , the measured TEER is 220 Ω·cm 2 .

[0169] The calibrated model outputs a risk correction factor (RCF). Formula:

[0170]

[0171] Specific calculation (example): Assuming d(BSI) / dt = 0.05 (estimated based on historical data), then

[0172]

[0173] Risk level stratified output. Risk level classification: Level I: RCF ≤ 0.2 and BSI < 30; Level II: 0.2 < RCF ≤ 0.5 or 30 ≤ BSI < 60; Level III: RCF > 0.5 or BSI ≥ 60.

[0174] Specific judgment (example): For a certain patient, RCF = 0.07, BSI = 0.375, belonging to Level I risk.

[0175] Logical OR operation for the degree of OCT mucosal structure damage: If the OCT examination shows that more than 50% of the crypts are deformed, the risk level is automatically upgraded by one level. It is set that the OCT examination of this patient does not show crypt deformation, so the final corrected risk level is still Level I.

[0176] 205. Also included is the verification of the combined dynamic regulation of the intestinal flora and metabolome: phage-directed regulation and metabolomics feedback: screening of pathogenic bacteria (such as adherent-invasive Escherichia coli AIEC)-specific phages based on metagenomic sequencing data to construct phage cocktail preparations; real-time monitoring of microbial changes after phage intervention by nanopore sequencing, and calculation of pathogen clearance rate (CR = 1-post-intervention abundance / baseline abundance); combined with mass spectrometry imaging to detect changes in the spatial distribution of metabolites (such as secondary bile acids and tryptophan derivatives) and evaluate the dynamic response of microbiome-metabolite-barrier function;

[0177] Cross-omics causal network construction: Using a Bayesian network-based causal inference algorithm, with phage intervention as an exogenous variable, the causal chain of bacterial species, metabolites, and barrier function parameters (BSI, TEER) was analyzed; the contribution of key regulatory pathways was defined as CRC:

[0178]

[0179] in, It indicates the degree of influence of changes in metabolite concentration on BSI; Indicates the degree of influence of changes in bacterial abundance on metabolite concentrations;

[0180] Screening microbiome-metabolite pairs with CRC > 0.5 as therapeutic targets (e.g., Faecalibacterium → butyrate → occludin expression ↑);

[0181] Microbiome-host metabolic model prediction: Genome-scale metabolic network models (GEMs) are constructed to integrate the metabolic flux of patient-specific microbiota and host intestinal cells; flux balance analysis (FBA) is used to predict the short-chain fatty acid (SCFA) synthesis rate and mucin secretion after targeted regulation; if the predicted SCFA synthesis rate increases by ≥30% and the error with the measured value is <15%, the microbiome intervention plan is considered effective.

[0182] It is important to note that phage-directed regulation and metabolomics feedback, phage cocktail preparation construction, and metagenomic sequencing data revealed a high abundance of adherent-invasive Escherichia coli (AIEC) in the patient's intestine, at 10^7 CFU / g feces. Phage screening: Phages specific for AIEC were screened from the phage library to construct a phage cocktail preparation.

[0183] Nanopore sequencing monitored changes in the microbiome. Baseline abundance: Before intervention, the abundance of AIEC was 10^7 CFU / g feces. Post-intervention monitoring: Real-time nanopore sequencing monitored changes in the microbiome after phage intervention, revealing that the abundance of AIEC decreased to 10^5 CFU / g feces. Pathogen clearance rate (CR): CR = 1 - post-intervention abundance / baseline abundance = 1 - 10^7 = 0.99, indicating 99% AIEC clearance.

[0184] Mass spectrometry imaging was used to detect metabolite changes. Metabolite detection: Combined mass spectrometry imaging was used to detect changes in the spatial distribution of metabolites such as secondary bile acids and tryptophan derivatives in intestinal tissue. Specific data: Before intervention, the average concentration of secondary bile acids was 5 μM, and the average concentration of tryptophan derivatives was 2 μM; after intervention, the average concentration of secondary bile acids increased to 8 μM, and the average concentration of tryptophan derivatives increased to 3.5 μM. Evaluation: Changes in metabolite concentrations indicate that phage intervention has positively altered the intestinal metabolic environment, contributing to improved intestinal barrier function.

[0185] Cross-omics causal network construction and Bayesian network causal inference of exogenous variables: phage intervention was used as the exogenous variable. Causal chain analysis: A Bayesian network-based causal inference algorithm was used to analyze the causal chain between bacterial species (such as Faecalibacterium), metabolites (such as butyrate), and barrier function parameters (BSI, TEER).

[0186] Calculation of key regulatory pathway contribution (CRC), formula application:

[0187]

[0188] Specific data: Taking Faecalibacterium→Butyrate→Occludin expression↑ as an example, the calculated CRC=0.6, indicating that this pathway has a high contribution to barrier function.

[0189] Screening therapeutic targets: Screening microbiota-metabolite pairs with CRC>0.5 as therapeutic targets, such as Faecalibacterium→butyrate.

[0190] Microbiome-host metabolic model prediction, genome-scale metabolic network model (GEMs) construction, data integration: integration of patient-specific microbiome and host intestinal cell metabolic flux data. Model construction: construction of genome-scale metabolic network models (GEMs).

[0191] Flux balance analysis (FBA) prediction, prediction goal: predict the short-chain fatty acid (SCFA) synthesis rate and mucin secretion after targeted regulation. Specific data: Before intervention, the SCFA synthesis rate was 5mmol / L / h, and the mucin secretion was 10mg / L / h. Through FBA prediction, the SCFA synthesis rate increased to 7mmol / L / h and the mucin secretion increased to 12mg / L / h after targeted regulation. Verification: The measured SCFA synthesis rate was 6.8mmol / L / h, and the error with the predicted value was 2.86%, which was less than 15%; the measured value of mucin secretion was 11.8mg / L / h, which was close to the predicted value. Judgment: Since the predicted SCFA synthesis rate increased by ≥30% and the error with the measured value was <15%, the microbiome intervention plan was judged to be effective.

[0192] 206. It also includes a multimodal monitoring device collaboration and real-time decision-making system: wearable-implantable device collaborative architecture: stretchable electronic skin: integrated with flexible ultrasound transducers for detecting intestinal wall hardness, impedance sensors for TEER monitoring, and optical microlens arrays for OCT signal acquisition, which adheres to the abdominal skin for continuous monitoring; smart capsule endoscope: loaded with nanoprobes and micro-metabolite sensors to detect intraluminal butyrate concentration, retained in the intestinal lesion site for ≥ 72 hours, and powered by a biofuel cell; subcutaneous implantable microprocessor: receives data from multiple devices, executes a lightweight version of the personalized prediction model, predicts the risk level and triggers the alarm threshold;

[0193] Edge-cloud hybrid computing framework: On the edge, a lightweight GNN model based on knowledge distillation is deployed in an implantable microprocessor to calculate the barrier function trend index in real time. On the cloud, a high-precision Transformer model is reconstructed after receiving full data, generating weekly optimization recommendations for barrier repair pathways, including microbiome transplantation timing and drug dosage adjustments. Blockchain smart contracts enable secure multi-center data sharing and incremental model learning.

[0194] Closed-loop feedback treatment system: When BFTI exceeds the threshold for three consecutive hours, at least one of the following interventions is automatically activated: sustained-release nanoprobes loaded with anti-TNF-α antibodies are released through smart capsules; transcutaneous electrical stimulation of the wearable device at a frequency of 5Hz is triggered to promote goblet cell regeneration; a signal is sent to the associated drug pump to adjust the drug release rate of the mesalazine microneedle patch; and a metabolic flux prediction model is automatically started after the intervention to evaluate the efficacy and optimize the cycle.

[0195] It should be noted that in the wearable-implantable device collaborative architecture, the stretchable electronic skin includes a flexible ultrasonic transducer: integrated into the stretchable electronic skin, it is used to detect the hardness of the intestinal wall. After the patient's abdominal skin is attached, continuous monitoring shows that the average hardness of the intestinal wall is 20kPa, with a standard deviation of 2kPa. Impedance sensor: used for TEER (transepithelial electrical resistance) monitoring to reflect the integrity of the intestinal barrier. Continuous monitoring data shows that the average TEER is 30Ω·cm 2 , with a standard deviation of 3Ω·cm 2 Optical microlens array: Used for OCT (optical coherence tomography) signal acquisition to observe intestinal mucosal structure. OCT images show a thickness of 0.5 mm with a standard deviation of 0.05 mm.

[0196] The smart capsule endoscope includes a nanoprobe and a micro-metabolite sensor, which are installed within the capsule to monitor intraluminal butyrate concentrations. After 72 hours of retention at the site of intestinal lesions, the average butyrate concentration was 1 mmol / L with a standard deviation of 0.1 mmol / L. A biofuel cell provides continuous power to the capsule, ensuring its prolonged operation within the intestine.

[0197] Subcutaneous implantable microprocessor, data reception: Receives data from stretchable electronic skin and smart capsule endoscopy. Personalized prediction model: Executes a lightweight version of the personalized prediction model to predict the risk level based on historical data and real-time data. When the TEER value is lower than 25Ω·cm 2 Or when the butyrate concentration is lower than 0.8mmol / L, the alarm threshold is triggered.

[0198] Edge-cloud hybrid computing framework: On the edge, lightweight GNN model: A lightweight GNN (graph neural network) model after knowledge distillation is deployed in an implantable microprocessor to calculate the barrier function trend index (BFTI) in real time. The average value of BFTI is 0.8 and the standard deviation is 0.05. On the cloud, high-precision Transformer model: After receiving the full amount of data, the high-precision Transformer model is reconstructed to generate barrier repair path optimization recommendations every week. Recommendations include the timing of microbiota transplantation (such as when the abundance of beneficial bacteria in the intestine is below the threshold) and drug dosage adjustments (such as adjusting the mesalazine dosage according to inflammatory indicators). Blockchain smart contracts: Realize multi-center secure data sharing and incremental model learning to ensure data privacy and model updates.

[0199] Closed-loop feedback therapy system: Intervention activation, BFTI threshold: When the BFTI exceeds 0.9 for three consecutive hours, intervention measures are automatically activated. Intervention: Smart capsules release sustained-release nanoprobes loaded with anti-TNF-α antibodies to suppress inflammatory responses. Within 24 hours of release, intestinal TNF-α concentrations decreased by 30%. Transcutaneous electrical stimulation: Triggering transcutaneous electrical stimulation at a frequency of 5Hz from the wearable device promotes goblet cell regeneration. Within seven days of electrical stimulation, intestinal mucus secretion increased by 20%. Adjusting the release rate of the mesalazine microneedle patch: Sending a signal to the associated drug pump to adjust the release rate based on inflammatory indicators. After the adjustment, mesalazine blood concentrations stabilize within the effective range. Efficacy evaluation and loop optimization, metabolic flux prediction model: After the intervention, the metabolic flux prediction model is automatically activated to evaluate efficacy. The model predicted a 20% increase in butyrate synthesis rate, while the measured increase was 18%, with an error of 10%. Loop optimization: Based on the efficacy evaluation results, intervention measures and parameters are adjusted to achieve continuous optimization of the closed-loop feedback therapy system.

[0200] In an embodiment of the present invention, simultaneous multimodal signal acquisition and analysis enable real-time imaging and molecular-level dynamic tracking of intestinal barrier function, facilitating early detection of intestinal barrier impairment and improving the early diagnosis rate of ulcerative colitis. A multidimensional database is constructed and advanced artificial intelligence models are used to analyze the data, outputting personalized intestinal barrier impairment scores and disease progression predictions, providing physicians with accurate treatment decision support. Non-invasive monitoring methods such as swallowable capsule endoscopes and wearable devices are used to reduce patient pain and improve monitoring comfort and acceptance. Through dynamic validation of combined intestinal microbiota-metabolome regulation, key regulatory pathways are screened as therapeutic targets, enabling precise intervention. A closed-loop feedback treatment system is constructed to automatically adjust intervention measures based on real-time monitoring data and evaluate efficacy to achieve cyclic optimization and improve treatment effectiveness. Blockchain smart contracts are used to enable secure multi-center data sharing and incremental model learning, ensuring data privacy while promoting continuous model optimization and updating. In summary, this technology provides a comprehensive, accurate, and personalized solution for the monitoring, diagnosis, treatment, and efficacy evaluation of ulcerative colitis, with significant clinical application value and promising prospects for widespread application.

[0201] The above describes the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis in an embodiment of the present invention. The following describes the multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis in an embodiment of the present invention. Figure 3In one embodiment of the present invention, a multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis includes: an acquisition module 301 for designing a nanoprobe with a surface-modified specific binding element, the specific binding element targeting the tight junction protein Occludin and / or ZO-1, and simultaneously loading a catalytically active nanozyme, and using the nanoprobe to synchronously acquire multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals, and combining with a swallowable capsule endoscope to achieve real-time imaging and molecular-level dynamic tracking of the intestinal barrier, thereby obtaining a probe-based multimodal data set; a processing module 302 for constructing a genetically engineered bacterium, which carries a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio, and a sensing module for the inflammatory factor TNF-α or IL-6, and detects the fluorescence reporter signal or electrical signal output of the genetically engineered bacterium through a fecal sample to obtain engineered bacterium detection data and quantify intestinal permeability abnormalities. and local inflammation levels to obtain quantitative data on intestinal permeability and inflammation; an analysis module 303 is used to construct a multidimensional database based on a probe-based multimodal data set, engineered bacteria detection data, microbial composition data from metagenomic sequencing, and clinical biochemical indicators; a graph neural network or a Transformer model is used to analyze the nonlinear associations of the multidimensional database, and output an intestinal barrier function damage score and a disease progression prediction result; the pre-trained model is adapted to individual patient characteristics through transfer learning to achieve personalized monitoring and obtain a personalized prediction model; a calibration module 304 is used to quantify intestinal wall hardness using ultrasound elastography, and to detect the microstructural integrity of the mucosal layer using optical coherence tomography, and to measure the intestinal electric field impedance spectrum through wearable devices to invert the permeability changes between intestinal epithelial cells, thereby obtaining a composite parameter set of intestinal wall mechanics and electrophysiology, which is input into a personalized prediction model, and combined with the real-time impedance spectrum calibration model prediction value to output a corrected disease progression risk level.

[0202] above Figure 3 The multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis in an embodiment of the present invention is described in detail from the perspective of modular functional entities. The multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis in an embodiment of the present invention is described in detail from the perspective of hardware processing.

[0203] Figure 4: This is a schematic diagram of the structure of a multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis provided by an embodiment of the present invention. The multi-parameter monitoring device 400 for evaluating the intestinal barrier function of ulcerative colitis may vary greatly due to different configurations or performances, and may include one or more processors (central processing units, CPU) 410 (for example, one or more processors) and a memory 420, and one or more storage media 430 (for example, one or more mass storage devices) for storing application programs 433 or data 432. Among them, the memory 420 and the storage medium 430 can be temporary storage or permanent storage. The program stored in the storage medium 430 may include one or more modules (not shown in the figure), each module may include a series of instruction operations in the multi-parameter monitoring device 400 for evaluating the intestinal barrier function of ulcerative colitis. Furthermore, the processor 410 can be configured to communicate with the storage medium 430 to execute a series of instruction operations in the storage medium 430 on the multi-parameter monitoring device 400 for evaluating the intestinal barrier function of ulcerative colitis.

[0204] The multi-parameter monitoring device 400 for evaluating the intestinal barrier function of ulcerative colitis may further include one or more power supplies 440, one or more wired or wireless network interfaces 450, one or more input and output interfaces 460, and / or one or more operating systems 431, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 4 The structure of the multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis shown does not constitute a limitation of the multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0205] The present invention also provides a multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis. The multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis in the above-mentioned embodiments.

[0206] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis.

[0207] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0208] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0209] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis, characterized in that: The multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis includes: Design a nanoprobe with a surface-modified specific binding element that targets the tight junction proteins occludin and / or ZO-1 and is loaded with catalytically active nanozymes. Use the nanoprobe to synchronously collect multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals. Combined with a swallowable capsule endoscope, real-time imaging and molecular-level dynamic tracking of the intestinal barrier are achieved, resulting in a probe-based multimodal dataset. Constructing genetically engineered bacteria that carry a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio and a sensing module for the inflammatory factors TNF-α or IL-6. The fluorescent reporter signal or electrical signal output of the genetically engineered bacteria is detected in fecal samples to obtain engineered bacteria detection data, quantify intestinal permeability abnormalities and local inflammation levels, and obtain quantitative intestinal permeability and inflammation data. A multidimensional database is constructed based on probe-based multimodal datasets, engineered bacteria detection data, microbial composition data from metagenomic sequencing, and clinical biochemical indicators. Graph neural networks or Transformer models are used to analyze the nonlinear associations in the multidimensional database, outputting intestinal barrier function damage scores and disease progression prediction results. Transfer learning is used to adapt the pre-trained model to individual patient characteristics, enabling personalized monitoring and obtaining a personalized prediction model. Ultrasound elastography is used to quantify intestinal wall hardness, combined with optical coherence tomography to detect the microstructural integrity of the mucosal layer. Wearable devices are used to measure the intestinal electric field impedance spectrum to invert the changes in permeability between intestinal epithelial cells, and obtain a set of intestinal wall mechanical-electrophysiological composite parameters. These parameters are input into a personalized prediction model, combined with the real-time impedance spectrum calibration model prediction value, and the corrected disease progression risk level is output.

2. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: The nanoprobe is designed with a surface-modified specific binding element, which targets the tight junction protein Occludin and / or ZO-1 and is loaded with catalytically active nanozymes. The nanoprobe is used to synchronously collect multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals. In combination with a swallowable capsule endoscopy, real-time imaging and molecular-level dynamic tracking of the intestinal barrier are achieved, resulting in a probe-based multimodal dataset, including: The nanoprobe adopts a core-shell structure, with the core being a SrDy2O4 nanozyme with CT / MRI dual-modality imaging capabilities, the shell modified with polyethylene glycol and a single-chain antibody targeting Occludin / ZO-1, and the surface of the nanozyme loaded with the electrochemically active molecule methylene blue; The swallowable capsule endoscope integrates a micro-spectrometer and a wireless transmission module, triggers probe signal acquisition at preset time intervals in the intestine, and transmits multimodal data in real time to an in vitro analysis terminal via near-field communication.

3. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: The genetically engineered bacteria are constructed, which carry a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio and a sensing module for inflammatory factors TNF-α or IL-6. The fluorescent reporter signal or electrical signal output of the genetically engineered bacteria is detected through fecal samples to obtain engineered bacteria detection data, quantify intestinal permeability abnormalities and local inflammation levels, and obtain quantitative intestinal permeability and inflammation data, including: The method for synchronously detecting fluorescent reporter signals and electrical signals uses time-resolved fluorescence detection technology to distinguish the lactulose / mannitol ratio from inflammatory factor signals through dual fluorescence channels; uses screen-printed electrodes to detect impedance changes in the electroactive biofilm of engineered bacteria, and combines differential pulse voltammetry to quantify the electrical signal intensity; The stool sample processing method comprises: enriching the engineered bacteria by centrifugation, collecting signals using a portable smartphone fluorescence imaging module or a micro-electrochemical workstation, and separating the bacterial colony from the host cells using a microfluidic chip; The quantification method includes: calculating the intestinal permeability index based on a standard curve of lactulose / mannitol ratio and fluorescence intensity; and outputting an inflammatory activity score through a nonlinear regression model of electrical signal intensity and TNF-α / IL-6 concentration.

4. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: The method constructs a multidimensional database based on a probe-based multimodal dataset, engineered bacteria detection data, metagenomic sequencing bacterial composition data, and clinical biochemical indicators; uses a graph neural network or a Transformer model to analyze the nonlinear associations of the multidimensional database and outputs an intestinal barrier function damage score and disease progression prediction results; Through transfer learning, the pre-trained model is adapted to individual patient characteristics to achieve personalized monitoring and obtain a personalized prediction model, including: Multidimensional database construction: Temporally and spatially align the multimodal dataset of the probe to extract the characteristic values of fluorescence signal decay rate, electrochemical catalytic efficiency and SERS spectrum peak shift; K-mer frequency encoding and functional pathway enrichment analysis were performed on the metagenomic sequencing data to construct a microbiota-metabolic function association matrix; and time series normalization was performed on clinical biochemical indicators, including the fluctuation curve of fecal calprotectin concentration and the dynamic changes of serum CRP / IL-6 ratio.

5. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: Ultrasound elastography is used to quantify intestinal wall hardness, combined with optical coherence tomography to detect the microstructural integrity of the mucosal layer. Wearable devices are used to measure intestinal electric field impedance spectroscopy to invert changes in intestinal epithelial cell permeability. A composite set of intestinal wall mechanical and electrophysiological parameters is obtained, which is input into a personalized prediction model. Combined with the real-time impedance spectroscopy calibration model prediction value, a revised disease progression risk level is output, including: Synchronous acquisition of multimodal biophysical parameters: High-frequency ultrasound elastography is used to obtain the distribution of intestinal wall shear wave velocity and Young's modulus, quantifying the intestinal wall hardness gradient; frequency-domain optical coherence tomography is used to scan the mucosal layer to extract goblet cell density, mucus layer thickness, and crypt structural integrity index; an alternating electric field is applied through a wearable flexible electrode array, and the dynamic changes of intestinal epithelial intercellular resistance and transmembrane capacitance are inverted based on the Cole-Cole impedance model; Construction of composite parameter set and feature fusion: Spatiotemporally register the intestinal wall shear wave velocity, goblet cell density, and intestinal epithelial cell - to - cell resistance value to generate a three - dimensional mechanical - electrophysiological fusion thermogram of the intestinal segment; Define the composite parameter barrier stress index: Among them, SWV max Indicates the maximum value of the intestinal wall shear wave velocity, which reflects the hardness or elasticity of the intestinal wall tissue and can be measured in real time by ultrasound elastography. min The minimum value of the electrical resistance between intestinal epithelial cells is an important indicator for evaluating intestinal barrier function and reflects the integrity of the tight junctions between epithelial cells. α and β are adaptive weight coefficients based on the patient's historical data. Dynamic calibration of personalized prediction model: Establish an impedance spectrum - histopathology correlation matrix: Solve the inverse problem of finite element to map the real - time impedance phase angle to the opening and closing probability of tight junctions between epithelial cells; Use the extended Kalman filter algorithm to dynamically correct the prediction results of the Transformer model with real - time impedance data as the observation value; The calibrated model outputs a risk correction factor: Among them, the predicted TEER refers to the intestinal epithelial cell - to - cell resistance value predicted by the personalized prediction model; The measured TEER refers to the intestinal epithelial cell - to - cell resistance value obtained by actual measurement; d(BSI) / dt represents the change rate of BSI over time, reflecting the dynamic change of intestinal barrier function; Risk level stratification output: According to the RCF value and the BSI threshold interval, the disease progression risk is divided into three levels: Level I: RCF ≤ 0.2 and BSI < 30; Level II: 0.2 < RCF ≤ 0.5 or 30 ≤ BSI < 60; Level III: RCF > 0.5 or BSI ≥ 60; The risk level is subjected to a logical OR operation with the degree of OCT mucosal structure damage. If more than 50% of the crypts are deformed, the risk level will be automatically upgraded by one level to generate the final corrected risk level.

6. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: It also includes the verification of the combined dynamic regulation of intestinal flora - metabolome: Adopt a causal inference algorithm based on Bayesian network, with phage intervention as an exogenous variable, to analyze the causal chain of flora species, metabolites, and barrier function parameters; Define the contribution degree of the key regulatory pathway as CRC: in, It indicates the degree of influence of changes in metabolite concentration on BSI; Indicates the degree of influence of changes in bacterial abundance on metabolite concentrations; Screen the flora - metabolite pairs with CRC > 0.5 as treatment targets; Prediction of the microbiota - host metabolism model: Construct a genome - scale metabolic network model, integrating patient - specific flora and host intestinal cell metabolic fluxes; Predict the short - chain fatty acid synthesis rate and mucin secretion amount after targeted regulation through flux balance analysis; If the predicted SCFA synthesis rate increases by ≥ 30% and the error from the measured value is < 15%, it is determined that the flora intervention plan is effective.

7. The multi-parameter monitoring method for evaluating intestinal barrier function in ulcerative colitis according to claim 1, characterized in that: It also includes the coordination of multi - modal monitoring devices and a real - time decision - making system: Collaborative architecture of wearable - implantable devices: Stretchable electronic skin: Integrate flexible ultrasonic transducers for detecting intestinal wall hardness, impedance sensors for TEER monitoring, and optical microlens arrays for OCT signal acquisition, and adhere to the abdominal skin for continuous monitoring; Smart capsule endoscope: Load the nanosensor and micro - metabolite sensor described in claim 2 to detect the intraluminal butyrate concentration, stay in the intestinal lesion site for ≥ 72 hours, and be powered by a bio - fuel cell; Subcutaneous implantable microprocessor: Receive data from multiple devices, execute a lightweight version of the personalized prediction model described in claim 4, predict the risk level, and trigger the alarm threshold; Edge - cloud hybrid computing framework: Edge side: Deploy a lightweight GNN model after knowledge distillation in the implantable microprocessor to calculate the barrier function trend index in real time; Cloud: After receiving the full data, a high-precision Transformer model is reconstructed to generate weekly recommendations for optimizing barrier repair pathways, including microbiome transplantation timing and drug dosage adjustments. Blockchain smart contracts are used to enable secure multi-center data sharing and incremental model learning. Closed-loop feedback therapy system: When the BFTI exceeds the threshold for 3 consecutive hours, at least one of the following interventions is automatically activated: sustained-release nanoprobes via smart capsules; Transcutaneous electrical stimulation to trigger wearable devices; sending a signal to an associated drug pump; After the intervention, the metabolic flux prediction model described in claim 6 is automatically started to evaluate the therapeutic effect and perform cycle optimization.

8. A multi-parameter monitoring system for evaluating intestinal barrier function in ulcerative colitis, characterized in that: The multi-parameter monitoring system for evaluating the intestinal barrier function of ulcerative colitis includes: An acquisition module is used to design nanoprobes with surface-modified specific binding elements that target tight junction proteins occludin and / or ZO-1 and are loaded with catalytically active nanozymes. The nanoprobes are used to synchronously acquire multimodal signals, including fluorescence signals, electrochemical signals, and oxidative stress catalytic reaction signals. In combination with a swallowable capsule endoscope, real-time imaging and molecular-level dynamic tracking of the intestinal barrier are achieved to obtain a probe-based multimodal dataset. A processing module is used to construct a genetically engineered bacterium that carries a biosensor module that responds to the intestinal permeability marker lactulose / mannitol ratio and a sensing module that detects inflammatory factors such as TNF-α or IL-6. The module detects the fluorescent reporter signal or electrical signal output of the genetically engineered bacterium through a fecal sample to obtain engineered bacterium detection data, quantify intestinal permeability abnormalities and local inflammation levels, and obtain quantitative intestinal permeability and inflammation data. The parsing module is used to construct a multidimensional database based on probe-based multimodal datasets, engineered bacteria detection data, microbial composition data from metagenomic sequencing, and clinical biochemical indicators. Graph neural networks or Transformer models are used to parse the nonlinear associations in the multidimensional database to output intestinal barrier function damage scores and disease progression prediction results. Transfer learning is used to adapt the pre-trained model to individual patient characteristics, enabling personalized monitoring and obtaining a personalized prediction model. The calibration module is used to quantify intestinal wall hardness using ultrasound elastography, detect the microstructural integrity of the mucosal layer using optical coherence tomography, and measure the intestinal electric field impedance spectrum through wearable devices to invert the changes in permeability between intestinal epithelial cells. The intestinal wall mechanical-electrophysiological composite parameter set is input into the personalized prediction model, and the predicted value of the real-time impedance spectrum calibration model is combined to output the corrected disease progression risk level.

9. A multi-parameter monitoring device for evaluating intestinal barrier function in ulcerative colitis, characterized in that: The multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the multi-parameter monitoring device for evaluating the intestinal barrier function of ulcerative colitis to execute the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the multi-parameter monitoring method for evaluating the intestinal barrier function of ulcerative colitis according to any one of claims 1 to 7 is implemented.

Images