Subcutaneous sensing system for noninvasive prediction of exogenous substance anaphylactic reaction

Abstract

The invention relates to the technical field of anaphylactic reaction monitoring, in particular to a subcutaneous sensing system for noninvasive prediction of exogenous substance anaphylactic reaction, which comprises a distributed sensor module, a central processing module, a grading early warning module and an optical verification module. According to the invention, at least four points of chest and abdomen, two points of four limbs, three points of back and one point of neck of a patient are covered by the distributed flexible patch array, subcutaneous impedance, temperature and capillary pressure are collected in real time, sterile single coverage limitation during operation is broken through, and early recognition of anaphylactic reaction is realized; the risk index is output through multi-modal feature fusion, the misjudgment rate is remarkably reduced through double-threshold early warning, and the first-aid response time is shortened through linkage with medicine infusion.

Description

Technical Field

[0001] The present invention relates to the technical field of allergic reaction monitoring, and in particular to a subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances. Background Art

[0002] During the perioperative infusion of anesthetic drugs or allogeneic blood products, allergic reactions triggered by exogenous allergens are a major clinical hazard that threatens patient safety. This type of reaction is characterized by sudden onset and rapid progression. Its early pathophysiological changes are mainly manifested in the degranulation of mast cells in the dermis of the skin, which releases inflammatory mediators such as histamine, leading to local capillary dilation and increased permeability, and the appearance of maculopapular lesions in areas such as the chest and abdomen, proximal limbs, back, and neck. The specific manifestations are itchy skin, clearly demarcated erythema, protruding wheals on the skin surface, and eczema-like changes. However, in the surgical environment, the patient's body is tightly covered with multiple layers of sterile dressings to achieve aseptic operation requirements. The skin area outside the surgical field is completely invisible, making it impossible for medical staff to detect these early skin symptoms in a timely manner through naked eye observation. When an allergic reaction is not identified early and continues to progress, a large amount of inflammatory mediators entering the blood can trigger a systemic reaction, including persistent hypotension caused by vasodilation and a sudden increase in ventilation resistance caused by bronchial smooth muscle spasm. In severe cases, it can induce anaphylactic shock or even cardiac arrest. At this time, the success rate of treatment will decrease exponentially with the delay time.

[0003] The monitoring principles of currently commonly used intraoperative monitoring equipment, such as multi-parameter ECG monitors, non-invasive blood pressure monitoring devices, and pulse oximeters, focus on changes in macro-physiological parameters of the circulatory and respiratory systems (such as heart rate, blood pressure, and blood oxygen saturation). Although such equipment can capture circulatory failure or oxygenation disorders that occur in the middle and late stages of allergic reactions, it cannot perceive early biophysical changes at the subcutaneous tissue level: including the characteristic decrease in dermal electrical impedance due to tissue edema, the abnormal gradient distribution of local temperature due to inflammatory reactions, and fluctuations in microvascular pressure due to changes in permeability. There is no existing technology that can penetrate the surgical dressing barrier to achieve synchronous dynamic monitoring of the subcutaneous tissue status in multiple locations; there is also a lack of intelligent analysis systems that can analyze the characteristics of histamine release and the spread trend of inflammatory reactions by integrating data from multiple regions such as the chest, abdomen, limbs, back, and neck.

[0004] Therefore, there is an urgent need to develop a new non-invasive monitoring system that covers rash-prone areas through a distributed sensor network, realizes dynamic assessment of subcutaneous tissue status under sterile sheet covering conditions during surgery, provides early warning and severity grading of allergic reactions for clinicians, and fills the gap in existing medical monitoring technology. Summary of the Invention

[0005] The purpose of the present invention is to provide a subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances. Through a distributed flexible patch array covering at least 4 points on the patient's chest and abdomen, 2 points on each of the limbs, 3 points on the back and 1 point on the neck, the system collects subcutaneous impedance, temperature and microvascular pressure in real time, breaking through the sterile single-coverage limitation during surgery and realizing early identification of allergic reactions; the risk index is output through multimodal feature fusion, the dual-threshold warning significantly reduces the misjudgment rate, and the linkage with drug infusion shortens the emergency response time.

[0006] In order to achieve the above technical objectives and the above technical effects, the present invention is implemented through the following technical solutions:

[0007] A subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances, comprising:

[0008] The distributed sensor module consists of a reusable patch array made of a flexible polymer substrate. The patch array is configured to simultaneously cover at least four detection points on the patient's chest and abdomen, two detection points on each of the limbs, three detection points on the back, and one detection point on the neck. Each patch has an integrated biocompatible adhesive layer on the bottom;

[0009] The multimodal signal acquisition module is built into each sensor patch and includes a three-electrode impedance detection unit, a miniature digital temperature sensing unit, and a capacitive microvascular pressure sensing unit. The impedance detection unit operates in the range of 1kHz-100kHz to acquire changes in the dielectric properties of subcutaneous tissue. The temperature sensing unit has a sampling accuracy of ±0.1°C, and the microvascular pressure sensing unit has a detection resolution of ≤5Pa.

[0010] The central processing module receives real-time data streams from each acquisition module via Bluetooth Low Energy. Its built-in feature extraction engine uses a wavelet transform algorithm to separate the conductivity fluctuation characteristics associated with histamine release and the local thermal diffusion anomaly characteristics caused by mast cell degranulation (temperature change rate > 0.3°C / min). It then maps the multi-site data into an allergic reaction risk index on a scale of 0-10 using a spatiotemporal fusion analysis model that includes a regional anomaly correlation analysis layer based on a convolutional neural network.

[0011] A graded warning module includes a dual-threshold trigger mechanism: when the risk index reaches 3-5 points, the sound and light alarm unit is activated to output a yellow warning signal; when the risk index is greater than 5 points and the conductivity mutation rate is detected in at least two areas greater than 20% / min, a red warning signal is driven and an emergency priority code is generated;

[0012] The optical verification module, which integrates a 650nm laser diode array and a CMOS image sensor at the edge of the sensor patch, automatically initiates a 10fps subcutaneous microvascular scan upon receiving an early warning signal, and generates a quantitative thermal map of exudate plaques through a dynamic contrast enhancement algorithm.

[0013] Beneficial effects of the present invention:

[0014] The system of the present invention uses distributed flexible microneedle patches to achieve continuous and dynamic monitoring of the patient's subcutaneous multimodal physiological signals, greatly improving the ability to identify allergic reactions to medications at an early stage. The core of the technology lies in the fact that the sensor patch can directly penetrate the stratum corneum of the skin to the superficial dermis, so that the real-time signals of parameters such as electrical impedance, temperature and microvascular pressure are closer to the site of inflammatory changes, and the special molecular imprinting polymers on the surface of the microneedles can highly selectively identify specific molecules such as histamine. It effectively breaks through the limitations of traditional reliance on systemic vital signs and visual inspection of the skin to identify early allergic symptoms, ensuring that in scenarios such as surgery where the skin is covered and vision is limited, the biophysical signal abnormalities of subcutaneous allergic reactions can still be detected in the first place, providing sufficient lead time for medical intervention.

[0015] The present invention adopts multi-region synchronous monitoring combined with spatiotemporal fusion intelligent analysis, which effectively improves the system's ability to perceive and judge the spatial diffusion dynamics of allergic reactions, and significantly reduces the probability of misjudgment and missed judgment. The system comprehensively analyzes signals from multiple areas of the chest, abdomen, back, limbs and neck, and no longer relies on abnormalities in a single area as a criterion. The central processing unit is based on multimodal feature collaboration, regional weight configuration and cross-regional collaborative alarm mechanism, which can distinguish the essential differences between local physiological fluctuations and systemic inflammatory responses. Models such as convolutional neural networks further enhance the discrimination effect of abnormal changes in time and space. It not only avoids false positive and false negative results caused by local mechanical pressure, occasional jitter or signal interference, but also greatly improves the tracking and comprehensive judgment capabilities of the multi-center and multi-stage diffusion process of allergic reactions, making the early warning classification more accurate and the intervention timing more scientific.

[0016] The present invention realizes closed-loop management of the entire process from real-time perception, intelligent analysis to emergency intervention and data archiving, significantly improving the efficiency of allergy emergency treatment and the quality of medical care. On the basis of intelligent graded early warning, the system can automatically link sound and light alarms, surface thermal map visualization, emergency drug injection pump preheating and precise titration infusion and other equipment. The central processing module can dynamically adjust the drug infusion plan according to the patient's weight, physiological feedback and other information, and quickly push a three-dimensional distribution map to mark abnormal areas when a high-risk warning is issued, to assist the medical team in making targeted decisions and operations. After the operation, all monitoring data, abnormal events and emergency treatment processes are automatically archived, providing solid data support for clinical review, efficacy evaluation and quality improvement. The closed-loop linkage of the system not only greatly shortens the response delay from abnormality identification to emergency intervention, improves the success rate of treatment of anaphylactic shock and severe allergic reactions, but also promotes the intelligent management of hospitals and the construction of medical big data, and ultimately comprehensively improves the level of medical services and patient life safety.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. DETAILED DESCRIPTION

[0018] The following is a clear and complete description of the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0019] Example 1

[0020] The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances described in this embodiment includes:

[0021] The distributed sensor module consists of a reusable patch array made of a flexible polymer substrate. The patch array is configured to simultaneously cover at least four detection points on the patient's chest and abdomen, two detection points on each of the limbs, three detection points on the back, and one detection point on the neck. Each patch has an integrated biocompatible adhesive layer on the bottom;

[0022] The multimodal signal acquisition module is built into each sensor patch and includes a three-electrode impedance detection unit, a miniature digital temperature sensing unit, and a capacitive microvascular pressure sensing unit. The impedance detection unit operates in the range of 1kHz-100kHz to acquire changes in the dielectric properties of subcutaneous tissue. The temperature sensing unit has a sampling accuracy of ±0.1°C, and the microvascular pressure sensing unit has a detection resolution of ≤5Pa.

[0023] The central processing module receives real-time data streams from each acquisition module via Bluetooth Low Energy. Its built-in feature extraction engine uses a wavelet transform algorithm to separate the conductivity fluctuation characteristics associated with histamine release and the local thermal diffusion anomaly characteristics caused by mast cell degranulation (temperature change rate > 0.3°C / min). It then maps the multi-site data into an allergic reaction risk index on a scale of 0-10 using a spatiotemporal fusion analysis model that includes a regional anomaly correlation analysis layer based on a convolutional neural network.

[0024] A graded warning module includes a dual-threshold trigger mechanism: when the risk index reaches 3-5 points, the sound and light alarm unit is activated to output a yellow warning signal; when the risk index is greater than 5 points and the conductivity mutation rate is detected in at least two areas greater than 20% / min, a red warning signal is driven and an emergency priority code is generated;

[0025] The optical verification module, which integrates a 650nm laser diode array and a CMOS image sensor at the edge of the sensor patch, automatically initiates a 10fps subcutaneous microvascular scan upon receiving an early warning signal, and generates a quantitative thermal map of exudate plaques through a dynamic contrast enhancement algorithm.

[0026] In this embodiment, the patch of the distributed sensor module uses a polydimethylsiloxane (PDMS) substrate with a thickness of 0.8 mm, and microneedle electrode units are arranged in a 15×15 array on its surface. A single microneedle is composed of a 316L medical stainless steel core material and a 20 nm thick gold coating. The microneedle has a truncated cone structure: bottom diameter 80 μm, top diameter 20 μm, height 350±10 μm, and needle tip curvature radius 0.5 μm. A molecular imprinting polymer layer is deposited on the surface of the microneedle by electropolymerization. This layer is formed by methacrylic acid-ethylene glycol dimethacrylate copolymer, has a thickness of 500 nm, and is loaded with 4-vinylphenylboronic acid as a histamine-specific recognition ligand.

[0027] In this embodiment, when establishing a baseline model before surgery, the adaptive baseline calibration unit of the central processing module continuously collects the impedance value Z0, temperature T0, and microvascular pressure P0 of each detection point 10 times at an interval of 1 minute, and takes the sliding average as the baseline matrix; during surgery, the current parameter deviation ΔZ / Z0, ΔT / T0, and ΔP / P0 are calculated every 30 seconds. When the operating room temperature fluctuates by ≥±1.5°C, the dynamic compensation subunit is activated. This subunit uses the Kalman filter algorithm to predict the real physiological signal. Its state equation is expressed as follows:

[0028] X k =AX k-1 +BU k +W k

[0029] Z k =HX k +V k

[0030] Where A is the temperature-impedance coupling coefficient matrix, B is the ambient temperature influence weight, H is the observation matrix, W k / V k is the system / observation noise covariance.

[0031] In this embodiment, the regional weighted calculation unit of the spatiotemporal fusion analysis model performs the following operations:

[0032] The weight coefficients of the chest and abdomen detection points were set as α = 0.45 ± 0.05, the back α = 0.30 ± 0.03, the limbs α = 0.15 ± 0.02, and the neck α = 0.10 ± 0.01;

[0033] When the cross-regional coordinated alarm conditions are met: the neck ΔZ / Z0 mean value is greater than 18% and the limbs ΔZ / Z0 variation coefficient is less than 0.2 or ≥3 regions simultaneously detect ΔT greater than 1.0℃ and the temperature gradient is greater than 0.5℃ / cm 2 It overrides the regular risk index calculation and directly outputs the highest level warning.

[0034] In this embodiment, the emergency linkage unit of the hierarchical warning module includes:

[0035] Drug preload control subunit: sends instructions to the syringe pump via the CAN bus, driving the epinephrine storage chamber to heat up to a constant temperature of 25±0.5℃ within 15 seconds;

[0036] Intelligent drug delivery subunit: When the risk index is greater than 7 minutes for 120 seconds, titration infusion is automatically started according to the formula: infusion rate = 0.05 × risk index (μg / kg / min);

[0037] 3D navigation subunit: Generates a 3D grid map of the patient's body surface based on the sensor coordinates and marks high-risk areas in the form of a heat map (areas with ΔZ / Z0>25% are displayed as flashing red).

[0038] In this embodiment, the microneedle electrode unit is connected to an integrated microfluidic system, including:

[0039] A microchannel network with a depth-to-width ratio of 3:1 and a heparin-agarose composite coating on the inner wall;

[0040] SERS detection chamber, whose gold nanorod substrate surface is immobilized with anti-histamine IgE antibodies;

[0041] Capillary pump drive unit: Generates -15kPa negative pressure through hydrophilic modified PDMS, achieving continuous sampling of interstitial fluid at 5nL / min.

[0042] In this embodiment, the dual-wavelength spectral analysis unit of the optical verification module performs:

[0043] Synchronously emit 650nm and 850nm lasers and calculate the oxygenated / deoxygenated hemoglobin concentration using the Beer-Lambert law:

[0044] ΔC hβ =(ε1·OD2-ε2·OD1) / (ε1·ε2'-ε2·ε1')

[0045] Where ε is the extinction coefficient and OD is the optical density;

[0046] When ΔC hβoxy >15 μmol / L or ΔC hβ When the vascular leakage index is >10 μmol / L, it is generated and superimposed on the heat map.

[0047] In this embodiment, the anti-motion interference module includes:

[0048] A three-axis MEMS accelerometer embedded in each patch;

[0049] Motion artifact elimination algorithm: When acceleration > 1.5g and frequency > 2Hz are detected, a three-level processing flow is initiated:

[0050] S1: Freeze current data collection;

[0051] S2: Call the valid data from the previous 10 seconds to train the ARIMA forecast model;

[0052] S3: Output compensated signal:

[0053] In this embodiment, the microfluidic system includes an online quality control unit, which:

[0054] Electrochemical impedance spectroscopy (frequency 100 Hz-1 MHz) was used to monitor the microchannel blockage rate in real time;

[0055] SERS signal stability detection: When 1350cm -1 The self-cleaning process is triggered when the Raman peak intensity fluctuation is greater than ±15%.

[0056] In this embodiment, the intelligent drug delivery subunit is further configured as follows:

[0057] When the patient's weight is >90 kg, adjust the infusion according to the formula: Modified rate = basal rate × [1 + 0.01 × (weight - 70)];

[0058] Blood pressure feedback signals were collected every 30 seconds. If the systolic blood pressure was <90 mmHg, the rate was increased by 0.02 μg / kg / min until the blood pressure returned to normal.

[0059] Example 2

[0060] A subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances as described in this embodiment includes:

[0061] The distributed sensor network synchronously collects multimodal physiological signals from 12 monitoring points at a frequency of 10 times per second. The impedance detection unit measures changes in subcutaneous tissue conductivity at a characteristic frequency of 10 kHz. The raw data is filtered through a 0.1-50 Hz bandpass filter and outputs the conductivity change rate ΔG / G0. The temperature sensing unit uses a differential measurement method to eliminate environmental interference and calculate the temperature gradient ▽T (in °C / cm) between adjacent monitoring points.

[0062] The microvascular pressure sensing unit identifies permeability abnormalities by analyzing the variance of pressure fluctuations. 2 In order to prevent limb movement interference during surgery, the system performs real-time motion artifact elimination: when the triaxial accelerometer detects high-frequency vibrations greater than 1.5g, it immediately freezes the current data buffer and calls the previous 10 seconds of valid data to build a prediction model. The output compensated value replaces the contaminated signal. This process ensures signal reliability during electrosurgical operation or body position adjustment.

[0063] The system separates the 0.5-2 Hz low-frequency components through wavelet transform and calculates the characteristic energy integral E of histamine release w =∫|W(0.5~2Hz)| 2 df. When E w A positive marker for histamine release is triggered when the value exceeds 200% of the baseline value for 60 seconds.

[0064] The inflammatory response diffusion analysis utilizes a dual-temporal and spatial verification mechanism: after detecting ΔG / G0 > 15% in the chest and abdomen, the system automatically checks for synchronization signals in the back and extremities. If ΔT > 0.8°C on the back and the temperature gradient ▽T > 0.5°C / cm on the extremities, the analysis is considered positive for inflammatory diffusion. A temporal correlation constraint is also established: the time delay between the histamine release signal and the temperature rise must be less than 90 seconds; otherwise, the event is considered unrelated.

[0065] The risk index RI is synthesized through dynamic weighting:

[0066] Chest and abdomen signal weight 45% (S c =min(100,50×ΔG / G0));

[0067] Back 30% (S b =min(100,40×ΔT));

[0068] Limbs 20% (S l =min(100,30×▽T));

[0069] Neck 5% (S n =min(100, 60×ΔP / P0))

[0070] Final RI = 0.45S c +0.30S b +0.20S l +0.05S n This model quantifies the complexity of pathological processes as a linear index with a score of 0-10.

[0071] The early warning system implements a three-level response logic: when 3≤RI<5, a yellow warning is activated, driving the buzzer to output a 1kHz pulse tone and recording the event log; when RI≥5 and the conductivity mutation rate dG / dt in at least two areas is greater than 20% / min, a red warning is triggered and three first aid preparations are simultaneously executed: 650 / 850nm dual-wavelength laser scanning is initiated to generate a leakage thermal map, epinephrine is preheated to a constant temperature of 25±0.5℃, and a first aid priority code based on body surface coordinates is generated.

[0072] If RI>7 minutes for 120 seconds, the system will automatically calculate the following formula:

[0073] R=0.05×RI×[1+0.01×(W t -70)]μg / kg / min

[0074] Start epinephrine infusion (W t is the patient's weight in kg). During the infusion, blood pressure feedback is monitored every 30 seconds. When the systolic blood pressure is less than 90 mmHg, dynamic rate adjustment is performed: R new =R+0.02×t (t is the number of minutes of low blood pressure), until the blood pressure returns to the safe threshold.

[0075] Heatmap generation uses hemoglobin leakage quantification analysis using the Beer-Lambert law:

[0076] ΔC HbO2 =(ε 850 ·OD 650 -ε 650 ·OD 850 ) / (ε 650 ·ε' 850 -ε 850 ·ε' 650 )

[0077] Calculate the concentration change (ε 650 =3.5×10 4 M -1 cm -1 , ε 850 =1.2×10 4 M -1 cm -1 ), the area >15 μmol / L was marked in red, 10-15 μmol / L was marked in orange, and <10 μmol / L was marked in green.

[0078] The microfluidic channel implements online health monitoring: when the 1MHz electrochemical impedance spectroscopy detects a phase angle θ < -45°, the channel is determined to be blocked and a three-stage self-cleaning procedure is automatically initiated: first, 20μL of 0.1M NaOH solution is injected to dissolve protein deposits, then 50μL of PBS buffer is rinsed, and finally the baseline impedance value is recalibrated. The SERS detection unit is equipped with a signal stability check: when the histamine characteristic peak (1350cm -1 ) When the intensity fluctuation is greater than ±15% for 10 seconds, perform hardware self-test and update the baseline intensity I0 = (I max +I min ) / 2. All fault events are recorded in real time to the system log and fed back to the central monitoring platform through status codes. The system operation sequence strictly follows the four-stage protocol: baseline parameter collection is completed 10 minutes before surgery (G0=Σ1 0 Gi / 10); synchronize multimodal signals every 100ms during surgery; perform feature extraction and risk calculation every 30 seconds; and trigger emergency response with early warning signals in real time. This design ensures a closed-loop response latency of less than 3 seconds from signal perception to drug administration intervention.

[0079] Example 3

[0080] Thirty minutes before the procedure, medical staff deploy the sensor array on key areas of the patient's skin. Clean, dry skin surfaces are required for this procedure. Chest and abdominal patches are precisely positioned 1 cm below the xiphoid process, along the lower costal margins on both sides, and at the umbilicus. Extremity sensors are located at the midpoint of the biceps brachii and rectus femoris muscles on the upper arms and thighs. Back patches are distributed along the spine, focusing on the midpoint of the line connecting the inferior angles of the scapulae and the area adjacent to the spinous processes of the thoracic vertebrae. Neck sensors are placed at the posterior edge of the sternocleidomastoid muscle at the level of the fourth cervical vertebra. Each patch requires 10 seconds of pressure to allow the microneedles to penetrate the stratum corneum, until the LED indicator turns solid green, indicating a successful conductive circuit. The system then automatically performs a 10-minute baseline parameter acquisition: The median impedance is measured at intervals of 10 times as a baseline value. Temperature and microvascular pressure are simultaneously recorded at steady-state levels. Any abnormal temperature fluctuations exceeding ±0.5°C are automatically rejected.

[0081] When the operation begins, the system synchronously collects multimodal signals from 12 detection points at a frequency of 10 times per second. The impedance detection unit adopts a dual-frequency scanning strategy (10kHz / 100kHz), focusing on the conductivity change rate (ΔG / G0) in the 0.5-2Hz frequency band; the temperature sensing module calculates the temperature gradient between adjacent sensors in real time, and triggers a preliminary warning when the local temperature change rate exceeds the 0.3℃ / min threshold; the microvascular pressure sensor identifies permeability abnormalities by analyzing the pulsatile pressure variance, 40Pa 2 The above fluctuations are judged as pathological changes. If strong interference such as electrosurgery is encountered during surgery, the triaxial accelerometer will immediately freeze the data stream when detecting high-frequency vibration exceeding 1.5g, and call the ARIMA prediction model to generate a compensation signal (formula: The "motion artifact compensation" status is marked on the monitoring screen to ensure data reliability.

[0082] The system achieves early diagnosis through multi-dimensional feature cross-validation: when the conductivity mutation rate of the chest and abdomen detection points exceeds 15% / min, and the low-frequency energy integral is higher than 200% of the baseline for 60 seconds, it is determined to be a histamine release feature. At this time, cross-regional collaborative verification is automatically initiated - if the back sensor simultaneously detects a temperature increase of more than 0.8℃, or the temperature gradient in the limbs exceeds 0.5℃ / cm 2If the inflammatory response is confirmed to be spreading, it will directly override the conventional calculation logic and trigger a red alert. The risk index (RI) is generated using a dynamic weighting algorithm: chest and abdominal signals account for 45% (based on conductivity changes), back temperature accounts for 30%, limb temperature gradient accounts for 20%, and neck microvascular pressure accounts for 5%. The final output is a quantitative assessment of 0-10 points.

[0083] When the risk index rises to 3-5 (yellow warning), the system activates a 1kHz pulse buzzer alarm, highlights abnormal areas on the monitor screen, and recommends suspending the suspected drug infusion. If the RI exceeds 5 and the conductivity mutation rate in two or more areas exceeds 20% / min, a triple emergency response is initiated: a 650 / 850nm dual-wavelength laser scan immediately generates a subcutaneous leakage heat map; epinephrine is automatically preheated to a constant temperature of 25±0.5°C for use; and an emergency code (such as "A3" for severe thoracic or abdominal leakage) is generated. If the RI exceeds 7 for 120 seconds, the system automatically infuses epinephrine according to a weight-adjusted formula (basal rate 0.05 × RI μg / kg / min, increased by 1% for each kg of excess weight). Blood pressure is monitored every 30 seconds during the infusion. If systolic blood pressure falls below 90 mmHg, an additional 0.02 μg / kg is administered every minute until it returns to normal.

[0084] After the warning is triggered, the laser scanning unit calculates the change in hemoglobin concentration using the Beer-Lambert law: when the leakage of oxyhemoglobin is greater than 15μmol / L, it is marked as a red flashing area on the three-dimensional grid map of the patient's body surface. The microfluidic channel has an intelligent maintenance mechanism - if the 1MHz impedance spectrum detects a phase angle of less than -45°, it is determined to be blocked, and sodium hydroxide is automatically perfused to dissolve protein deposits and then flush; the histamine SERS characteristic peak (1350cm -1 ) When the intensity fluctuation exceeds 15%, the system recalibrates the baseline to ensure detection accuracy.

[0085] For obese patients (BMI>30), sensors are added at the T3-T12 vertebral levels on the back; for burn patients, the system switches to pure optical monitoring mode. When the electrosurgical unit is in use, the system automatically reduces the sampling frequency to 5Hz and enhances filtering. After surgery, an allergic reaction trend chart is automatically generated, including a timeline of key parameters and disposal records. Actual clinical verification shows that during liver transplantation, the system detected a surge in chest and abdominal conductivity of 22% / min (RI=5.2) 8 minutes after blood transfusion, and issued an alarm 12 minutes earlier than the blood pressure dropped, winning a critical time window for rescue. All data is encrypted and stored and compatible with the hospital's electronic medical record system, forming a complete closed-loop monitoring chain.

[0086] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances, characterized in that: include: A distributed sensor module comprising a patch array configured to simultaneously cover at least four detection points on the patient's chest and abdomen, two detection points on each of the four limbs, three detection points on the back, and one detection point on the neck. Each patch has a biocompatible adhesive layer integrated into its bottom. The multimodal signal acquisition module is built into each sensor patch and includes a three-electrode impedance detection unit, a miniature digital temperature sensing unit, and a capacitive microvascular pressure sensing unit. The impedance detection unit operates in the range of 1kHz-100kHz to acquire changes in the dielectric properties of subcutaneous tissue. The temperature sensing unit has a sampling accuracy of ±0.1°C, and the microvascular pressure sensing unit has a detection resolution of ≤5Pa. The central processing module receives real-time data streams from each acquisition module via Bluetooth. Its built-in feature extraction engine uses a wavelet transform algorithm to separate the conductivity fluctuation characteristics associated with histamine release and the local thermal diffusion abnormalities caused by mast cell degranulation. It then maps the multi-site data into a 0-10 allergic reaction risk index using a spatiotemporal fusion analysis model that includes a regional abnormality correlation analysis layer based on a convolutional neural network. A graded warning module includes a dual-threshold trigger mechanism: when the risk index reaches 3-5 points, the sound and light alarm unit is activated to output a yellow warning signal; when the risk index is greater than 5 points and the conductivity mutation rate is detected in at least two areas greater than 20% / min, a red warning signal is driven and an emergency priority code is generated; The optical verification module, which integrates a 650nm laser diode array and a CMOS image sensor at the edge of the sensor patch, automatically initiates a 10fps subcutaneous microvascular scan upon receiving an early warning signal, and generates a quantitative thermal map of exudate plaques through a dynamic contrast enhancement algorithm.

2. The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances according to claim 1, characterized in that: When establishing a benchmark model before surgery, the adaptive baseline calibration unit of the central processing module continuously collects the impedance value Z0, temperature T0, and microvascular pressure P0 of each detection point 10 times at an interval of 1 minute, and takes the sliding average as the benchmark matrix; during surgery, the current parameter deviation ΔZ / Z0, ΔT / T0, and ΔP / P0 are calculated every 30 seconds. When the operating room temperature fluctuates by ≥±1.5°C, the dynamic compensation subunit is activated. This subunit uses the Kalman filter algorithm to predict the real physiological signal. Its state equation is expressed as follows: X k =AX k-1 +BU k +W k Z k =HX k +V k Where A is the temperature-impedance coupling coefficient matrix, B is the ambient temperature influence weight, H is the observation matrix, W k / V k is the system / observation noise covariance.

3. The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances according to claim 1, characterized in that: The regional weighted calculation unit of the spatiotemporal fusion analysis model performs the following operations: The weight coefficients of the chest and abdomen detection points were set as α = 0.45 ± 0.05, the back α = 0.30 ± 0.03, the limbs α = 0.15 ± 0.02, and the neck α = 0.10 ± 0.01; When the cross-regional coordinated alarm conditions are met: the neck ΔZ / Z0 mean value is greater than 18% and the limbs ΔZ / Z0 variation coefficient is less than 0.2 or ≥3 regions simultaneously detect ΔT greater than 1.0℃ and the temperature gradient is greater than 0.5℃ / cm 2 It overrides the regular risk index calculation and directly outputs the highest level warning.

4. The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances according to claim 1, wherein: The emergency linkage unit of the hierarchical warning module includes: Drug preload control subunit: sends instructions to the syringe pump via the CAN bus, driving the epinephrine storage chamber to heat up to a constant temperature of 25±0.5℃ within 15 seconds; Intelligent drug delivery subunit: When the risk index is greater than 7 minutes for 120 seconds, the titration infusion is automatically started according to the formula: infusion rate = 0.05 × risk index; 3D navigation subunit: Generates a 3D grid map of the patient's body surface based on the sensor coordinates and marks high-risk areas in the form of a heat map.

5. The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances according to claim 1, wherein: The dual-wavelength spectral analysis unit of the optical verification module performs: Synchronously emit 650nm and 850nm lasers and calculate the oxygenated / deoxygenated hemoglobin concentration using the Beer-Lambert law: ΔC hβ =(ε1·OD2-ε2·OD1) / (ε1·ε2'-ε2·ε1') Where ε is the extinction coefficient and OD is the optical density; When ΔC hβoxy >15 μmol / L or ΔC hβ When the vascular leakage index is >10 μmol / L, it is generated and superimposed on the heat map.

6. The subcutaneous sensing system for non-invasively predicting allergic reactions to exogenous substances according to claim 1, characterized in that: The intelligent drug delivery subunit is further configured as follows: When the patient's weight is >90 kg, adjust the infusion according to the formula: Modified rate = basal rate × [1 + 0.01 × (weight - 70)]; Blood pressure feedback signals were collected every 30 seconds. If the systolic blood pressure was <90 mmHg, the rate was increased by 0.02 μg / kg / min until the blood pressure returned to normal.