Immune marker real-time monitoring and early warning system based on biosensor

Abstract

The invention relates to the technical field of biomedical sensing and intelligent health monitoring, and particularly discloses an immune marker real-time monitoring and early warning system based on a biosensor. The system comprises a biosensing acquisition unit, a data preprocessing unit, an individualized reference modeling unit, a nonlinear dynamics analysis unit and a risk early warning output unit. The concentration of immune markers such as interleukin 6 and complement components is continuously collected through an implantable or wearable sensor, a time autocorrelation coefficient and a fluctuation variance are analyzed in real time by combining an individualized immune homeostasis reference model and utilizing a critical moderation theory, an early symptom of immune system restoring force decline is recognized, and the early symptom of immune system restoring force decline is determined. And when it is judged that the phase change critical point is approached, graded early warning is triggered. According to the technical scheme, a disease deterioration early warning window can be advanced, accurate risk prompting is achieved before clinical symptoms appear, and the initiative and individuation level of autoimmune disease management is improved.

Description

Technical Field

[0001] This invention belongs to the field of biomedical sensing and intelligent health monitoring, specifically relating to a real-time monitoring and early warning system for immune biomarkers based on biosensors. Background Technology

[0002] With the rapid development of biosensing technology and precision medicine, real-time monitoring of autoimmune diseases has become a key direction for improving treatment outcomes and patient prognosis in clinical medicine. By integrating highly sensitive biosensors to capture subtle changes in the body's microenvironment, continuous and accurate data support can be provided for the long-term management of chronic diseases. For complex immune system diseases characterized by recurrent and insidious progression, real-time monitoring of the dynamic evolution of immune biomarkers has significant scientific and clinical value for preventing irreversible organ damage and optimizing individualized medication regimens.

[0003] Early identification and risk assessment of immune system state transitions, and the establishment of efficient real-time early warning models, are key research areas in biomedical engineering. This technological direction aims to transform complex biochemical signals into quantifiable physiological risk indicators. By continuously tracking the levels of specific cytokines or proteins, it seeks to capture subtle evolutionary characteristics of disease transitioning from remission to active phases. Constructing a monitoring system with high-frequency dynamic analysis capabilities can improve the system's response speed to individualized disease progression, enabling a shift from passive clinical diagnosis to proactive early intervention.

[0004] Existing methods for detecting immune biomarkers typically rely on fixed reference ranges for logical judgment, ignoring individual differences in immune baseline values ​​among different patients. This makes it difficult to accurately detect diseases in their latent progression stages because they have not yet reached the hard threshold. Traditional analytical models often employ linear evaluation logic, lacking a deep understanding of the immune system as a complex nonlinear dynamic system, making it difficult to identify the state instability signals exhibited by the system when it loses its resilience.

[0005] Current monitoring methods, when processing time-series data, often focus only on the absolute numerical fluctuations of biomarkers, neglecting the underlying autocorrelation and variance evolution. This results in the system's inability to detect critical signs of impending immune system collapse before clinical symptoms fully develop. This technological limitation of delayed intervention directly leads to the loss of the treatment window, increasing the risk of acute clinical events in patients. Therefore, a real-time monitoring and early warning system for immune biomarkers based on biosensors is anticipated. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time monitoring and early warning system for immune biomarkers based on biosensors, which can solve the problem of intervention lag in the above-mentioned background technology.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A real-time monitoring and early warning system for immune biomarkers based on biosensors includes a biosensor acquisition unit, a data preprocessing unit, a nonlinear dynamics analysis unit, a personalized benchmark modeling unit, and a risk warning output unit. The biosensing acquisition unit is configured to continuously acquire time-series concentration data of specific immune markers in the patient's body through implantable or wearable biosensors, the immune markers including interleukin-6 and complement components. The data preprocessing unit is configured to perform noise filtering, signal calibration, and sampling frequency unification on the time series concentration data to generate a standardized dynamic monitoring data stream. The individualized benchmark modeling unit is configured to construct a patient’s own immune homeostasis benchmark model based on the patient’s historical monitoring data. This immune homeostasis benchmark model characterizes the individualized fluctuation range and dynamic balance characteristics of each immune marker during the patient’s clinical remission period. The nonlinear dynamics analysis unit is configured to perform real-time calculation of the time autocorrelation coefficient and fluctuation variance of the standardized dynamic monitoring data stream based on the critical slowing theory, and compare the calculation results with the individualized benchmark model to identify early signs of decreased immune system resilience. The risk warning output unit is configured to generate a graded warning signal and push the warning information to the medical terminal or patient terminal through a secure communication link when the nonlinear dynamics analysis unit determines that the immune system is in a critical unstable state.

[0008] Preferably, the nonlinear dynamics analysis unit is further configured to continuously track the recovery rate of the concentration of immune markers per unit time. When the recovery rate is lower than a preset threshold and accompanied by a continuous increase in the variance of fluctuation, it is determined that the system is approaching the critical point of phase transition.

[0009] Furthermore, the individualized baseline modeling unit is configured to automatically update the patient's immune homeostasis baseline model after the patient enters a new clinical remission period, ensuring that the baseline model can dynamically adapt to the long-term evolution of the patient's immune status.

[0010] Furthermore, the biosensor acquisition unit adopts a multi-channel synchronous sensing architecture, which can simultaneously monitor no fewer than two immune biomarkers with synergistic indicative effects, and fuse multidimensional time series data into the nonlinear dynamics analysis unit to improve the robustness and specificity of phase transition recognition.

[0011] Preferably, the risk warning output unit is divided into three warning levels according to the degree of instability of the immune system, corresponding to observation suggestions, clinical follow-up reminders and emergency intervention alarms, and the triggering conditions of each warning level are dynamically adjusted according to the patient's past disease characteristics.

[0012] Furthermore, the data preprocessing unit incorporates an adaptive filtering algorithm that can automatically optimize filtering parameters based on the drift characteristics of the biosensor and environmental interference patterns, ensuring data reliability under long-term continuous monitoring.

[0013] Furthermore, the nonlinear dynamics analysis unit is configured to use a sliding time window mechanism when calculating the time autocorrelation coefficient. The length of the sliding time window is set within a specific range based on the physiological half-life of the immunomarker to balance analytical sensitivity and timeliness.

[0014] Preferably, the system further includes a medication record synchronization unit, which is configured to receive and integrate the patient's immunosuppressant or biologic drug administration time and dosage information, and input the immunosuppressant or biologic drug administration time and dosage information as a covariate into the nonlinear dynamics analysis unit to exclude non-pathological fluctuations caused by drug intervention.

[0015] Furthermore, when constructing the initial baseline model, the individualized baseline modeling unit requires the patient to be in a clinically confirmed stable remission period and to be continuously monitored for no less than a predetermined time, so as to ensure the representativeness and stability of the baseline data.

[0016] Furthermore, before generating an early warning signal, the risk warning output unit performs cross-validation logic, which requires at least two different types of immune markers to exhibit critical slowing characteristics before a high-level warning can be triggered, thereby reducing the false alarm rate.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The real-time monitoring and early warning system for immune biomarkers based on biosensors provided by this invention abandons the traditional static judgment logic that relies on a fixed reference value range, and instead takes a complex systems science perspective, regarding the immune system as an adaptive system with nonlinear dynamic characteristics.

[0018] 2. By introducing the critical slowing theory, this invention enables the system to capture signals of declining resilience exhibited by immune markers when their absolute values ​​remain within the normal range, such as increased time autocorrelation and increased variance of fluctuations. This sliding time window mechanism shifts the warning window for disease deterioration forward, providing precise risk alerts to both doctors and patients before clinical symptoms appear or organ damage occurs.

[0019] 3. By constructing an individualized immune homeostasis baseline model, the system overcomes the technical obstacle of significant differences in immune baselines among patients, achieving truly personalized dynamic monitoring. The introduction of multi-biomarker fusion analysis, medication information covariate integration, and multi-level cross-validation mechanisms improves the accuracy and clinical applicability of early warning, providing a new technical approach for proactive intervention and precise management of autoimmune diseases such as systemic lupus erythematosus. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention; Figure 2 This is a schematic diagram of the core principle framework of nonlinear dynamics analysis based on critical slowing-down theory in this invention; Figure 3 This is a logical flowchart of the data preprocessing and individualized immune homeostasis benchmark modeling in this invention; Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow of multi-dimensional time series data fusion and pharmacodynamic covariate integration in this invention. Detailed Implementation

[0021] Example 1: Please refer to the appendix Figure 1 To be continued Figure 4 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0022] The real-time monitoring and early warning system for immune biomarkers based on biosensors includes a biosensor acquisition unit, a data preprocessing unit, a personalized benchmark modeling unit, a nonlinear dynamics analysis unit, a medication record synchronization unit, and a risk warning output unit. The biosensor acquisition unit is configured to continuously acquire time-series concentration data of specific immune markers in the patient's body through implantable or wearable biosensors. The immune markers include interleukin-6 and complement components. The biosensor acquisition unit adopts a multi-channel synchronous sensing architecture, which can simultaneously monitor no less than two immune markers with synergistic indicative effects, and fuses multidimensional time-series data into the nonlinear dynamics analysis unit. The data preprocessing unit is connected to the biosensor acquisition unit and is used to perform noise filtering, signal calibration and sampling frequency unification processing on the time series concentration data to generate a standardized dynamic monitoring data stream. The data preprocessing unit has a built-in adaptive filtering algorithm that can automatically optimize the filtering parameters according to the drift characteristics of the biosensor and the environmental interference mode. The individualized benchmark modeling unit is used to construct a patient’s own immune homeostasis benchmark model based on the patient’s historical monitoring data. The immune homeostasis benchmark model characterizes the individualized fluctuation range and dynamic balance characteristics of each immune marker during the clinical remission period, and automatically updates the patient’s immune homeostasis benchmark model after the patient enters a new clinical remission period. The nonlinear dynamics analysis unit is connected to the data preprocessing unit and the individualized benchmark modeling unit. It is used to perform real-time calculation of the time autocorrelation coefficient and fluctuation variance of the standardized dynamic monitoring data stream based on the critical slowing theory, and compare the calculation results with the individualized benchmark model to identify early signs of decreased immune system resilience. The medication record synchronization unit is configured to receive and integrate the patient's immunosuppressant or biologic drug administration time and dosage information, and input the immunosuppressant or biologic drug administration time and dosage information as a covariate into the nonlinear dynamics analysis unit. The risk warning output unit is used to generate a graded warning signal and push the warning information to the medical terminal or patient terminal through a secure communication link when the nonlinear dynamics analysis unit determines that the immune system is in a critical unstable state.

[0023] In the biosensing acquisition unit, the multi-channel synchronous sensing architecture includes a first sensing channel and a second sensing channel. The first sensing channel integrates a highly specific biofunctionalized surface for interleukin-6, and the second sensing channel integrates an electrochemical sensor for complement component 3 or complement component 4.

[0024] The biosensor acquisition unit is configured to perform data acquisition tasks at a preset sampling period. This sampling period is dynamically configured based on the physiological half-life of different immune markers to ensure that the collected time-series data can fully cover the dynamic evolution of the immune system. To ensure the real-time performance and continuity of the data, the biosensor acquisition unit is equipped with a local temporary memory for temporarily storing monitoring data when the communication link is unstable and for resuming transmission after the link is restored.

[0025] Furthermore, the biosensor acquisition unit also includes an environmental compensation subunit, which is equipped with a temperature sensor and a hydrogen ion concentration index sensor to monitor the physicochemical parameters of the microenvironment in which the biosensor is located. Since the sensitivity of the biosensor is affected by temperature and pH, the environmental compensation subunit sends the real-time acquired physicochemical parameters as correction factors to the data preprocessing unit to compensate for signal drift caused by environmental fluctuations.

[0026] In the data preprocessing unit, the adaptive filtering algorithm employs a dynamic compensation model based on an improved Kalman filter. The data preprocessing unit is configured to identify and remove outliers caused by physical collisions, electromagnetic interference, or extreme physiological activity. The signal calibration subunit is configured to perform zero-point calibration and range calibration, converting the original current or potential signal into a standardized concentration physical quantity by comparing it with pre-stored sensor characteristic curves. To meet the requirements of nonlinear dynamic analysis for equally spaced time series, the data preprocessing unit also includes a resampling module. This resampling module is configured to use a spline interpolation algorithm to fill in irregular time nodes caused by communication packet loss, outputting a standardized dynamic monitoring data stream with a uniform sampling frequency.

[0027] The core logic of the personalized baseline modeling unit lies in identifying and defining the "immune fingerprint" of a patient in a clinically stable state. This unit enters learning mode at the initial system startup or after a patient's condition is reset, requiring the patient to be in a stable remission period confirmed by a clinician. The personalized baseline modeling unit is configured to extract the distribution characteristics of immune markers under different diurnal rhythms over a continuous period of no less than 14 circadian cycles.

[0028] The immune homeostasis baseline model includes not only the mean and variance of each biomarker, but also potential surface features that reflect the depth of the system's homeostasis. When a patient enters a new remission equilibrium state due to disease progression, the individualized baseline modeling unit incorporates the new homeostasis features into the baseline model through a recursive learning algorithm, achieving dynamic adaptive evolution of the baseline model.

[0029] The nonlinear dynamics analysis unit performs an in-depth assessment of the stability of the immune system using the critical slowing-down theory. Critical slowing-down refers to the physical phenomenon where, when a complex system approaches its critical point or phase transition point, its rate of recovery to its original equilibrium state after being disturbed becomes exceptionally slow. The nonlinear dynamics analysis unit is configured to monitor the evolution of the statistical characteristics of the standardized dynamic monitoring data stream in real time within a sliding time window. The nonlinear dynamics analysis unit includes an autocorrelation calculation module and a variance monitoring module.

[0030] The autocorrelation calculation module is configured to calculate the first-order time autocorrelation coefficient of the monitored data within each sliding time window step. The time autocorrelation coefficient is obtained by calculating the covariance between the current time-to-time data and the previous time-to-time data and then standardizing it. When the immune system is in a highly stable equilibrium state, disturbances are rapidly attenuated due to the system's strong negative feedback regulation capability, resulting in low time autocorrelation. However, when the system approaches the instability threshold, the impact of disturbances accumulates, leading to an increased correlation between the current state and the previous state.

[0031] The variance monitoring module is configured to synchronously calculate the fluctuation variance within the sliding time window. As the immune system shifts towards an unstable state, its potential energy surface gradually flattens, and even minor internal or external random perturbations can trigger large numerical fluctuations, manifested as an abnormal increase in the variance of biomarker concentration fluctuations.

[0032] The nonlinear dynamics analysis unit is also configured to calculate the recovery rate of immune marker concentration per unit time. When the system detects that the absolute value of the slope of the return path of a certain marker after deviating from the baseline value continues to decrease, it determines that the relaxation time of the system has increased. When the recovery rate is lower than a preset rate threshold, and the time autocorrelation coefficient exceeds a first preset warning line and the fluctuation variance exceeds a second preset warning line, the nonlinear dynamics analysis unit determines that the immune system is losing its recovery ability and the condition is approaching the phase transition critical point.

[0033] The medication record synchronization unit is configured to interface with the patient's electronic pillbox or mobile medical application to obtain the administration time, dosage, and pharmacokinetic parameters of various immunomodulatory drugs. The medication record synchronization unit uses the expected concentration fluctuations caused by the drug as a covariate. Upon receiving the covariate, the nonlinear kinetic analysis unit performs residual analysis logic, that is, it isolates the drug-directed components from the total fluctuation signal, analyzing only the fluctuation components caused by changes in the dynamic characteristics of the immune system itself. This mechanism prevents normal physiological responses caused by medication from being misinterpreted by the system as signs of instability.

[0034] The risk warning output unit executes multi-level warning logic. The three warning levels include: Level 1 observation suggestion, corresponding to a slight increase in the autocorrelation of a single biomarker, prompting the user to maintain routine monitoring; Level 2 clinical follow-up suggestion, corresponding to two or more biomarkers simultaneously showing obvious critical slowing characteristics, and the recovery rate decreasing by more than a preset proportion, indicating a possible risk of hidden progression; Level 3 emergency intervention alarm, corresponding to biomarker values ​​starting to deviate rapidly from the baseline model, and the nonlinear dynamics indicator showing the system has crossed a critical threshold.

[0035] To reduce false alarm rates, the risk warning output unit executes cross-validation logic before triggering a high-level warning. This cross-validation logic requires that interleukin-6 and complement components simultaneously exhibit statistically significant abnormal shifts within a preset time window. If only a single biomarker is abnormal, the system will extend the observation time window or increase the detection frequency, rather than immediately triggering an emergency alarm. The risk warning output unit is also configured to dynamically adjust the trigger sensitivity of each level of warning based on the patient's historical disease characteristics, such as the frequency and severity of previous relapses, achieving truly personalized risk management.

[0036] Example 2: Based on the real-time monitoring and early warning system for immune biomarkers based on biosensors described in Example 1, this example provides a distributed implementation scheme based on edge computing and cloud collaborative architecture, which aims to solve the problem of storage of ultra-large-scale long time series data and computational load balancing of complex dynamic models.

[0037] In this embodiment, the system is divided into an edge processing layer and a cloud analytics layer. The biosensor acquisition unit and the data preprocessing unit are deployed at an edge node close to the patient. The edge node includes a high-performance, low-power embedded processor configured to perform high-frequency data sampling and preliminary filtering calibration.

[0038] The nonlinear dynamics analysis unit is further divided into an edge real-time analysis submodule and a cloud-based deep mining submodule. The edge real-time analysis submodule is configured to perform basic autocorrelation and variance calculations based on a short sliding window to ensure sub-second response to acute instability symptoms. Due to the limited computing resources of edge nodes, the edge real-time analysis submodule adopts a fixed-point arithmetic optimization algorithm to avoid the high energy consumption and latency caused by floating-point operations.

[0039] The cloud-based analytics layer consists of a high-performance server cluster, and the individualized benchmark modeling unit is fully deployed in the cloud. Edge nodes periodically upload standardized dynamic monitoring data streams to the cloud-based analytics layer via encrypted wireless communication links. The cloud-based deep mining submodule within the cloud-based analytics layer is configured to perform system dynamics trend analysis over a longer time horizon.

[0040] The cloud-based analytics layer also includes a population baseline database, which aggregates a large amount of immune biomarker evolution data from patients with similar disease types. When constructing a baseline model for a specific patient, the individualized baseline modeling unit is configured to utilize transfer learning algorithms, referencing common features in the population baseline database, to establish a preliminary model with high confidence even in the initial stage with limited data. As monitoring time increases, the model weights gradually shift towards the patient's own data, completing a smooth transition from a population baseline to a purely individualized baseline.

[0041] In the distributed architecture, the data preprocessing unit further includes a data compression submodule. This data compression submodule is configured to process redundant stationary data using a lossless compression algorithm or a lossy compression algorithm based on feature preservation. When the system detects severe signal fluctuations or nonlinear dynamic indicators approaching warning thresholds, the data compression submodule automatically switches to the original data pass-through mode to ensure that the cloud can obtain the most complete signal details for in-depth analysis.

[0042] The cloud-based deep mining submodule within the cloud analytics layer is also configured to perform phase space reconstruction analysis. By calculating the delayed embedding dimension and the Lyapunov index, the overall degree of chaos in the patient's immune system is assessed from a macroscopic perspective. If the Lyapunov index changes from negative to positive or increases in value, it is determined that the nonlinear interactions within the system are intensifying, and the system is evolving towards an unstable chaotic state. This is used as one of the in-depth criteria for the risk warning output unit to trigger advanced warnings.

[0043] The medication record synchronization unit integrates multi-source data within a cloud-based architecture. It not only receives input from the patient but is also configured to automatically interface with the hospital's information system via a standard medical interface to obtain the patient's latest laboratory test reports and medical orders. The system is configured to compare real-time monitoring values ​​from biosensors with laboratory gold standard test values. If a systematic deviation occurs, the data preprocessing unit will automatically initiate a remote recalibration process to eliminate long-term zero-point drift or sensitivity decay of the sensor by correcting model parameters.

[0044] The risk warning output unit offers richer presentation options in a cloud environment. It not only sends text notifications to patient terminals but is also configured to generate dynamic risk evolution charts, displaying the trajectory of the immune system in a multi-dimensional feature space. The risk warning output unit is configured to calculate the "critical distance," i.e., the geometric distance from the current system state point to the instability boundary, and visually presents the disease's evolution to medical personnel in the form of a countdown or percentage.

[0045] The system also includes a security audit unit for anonymizing and encrypting all personal physiological data transmitted between edge nodes and the cloud using national cryptographic standards. In the secure communication link, a zero-trust architecture-based identity authentication mechanism is employed to ensure that only authorized medical personnel can access the in-depth analysis reports.

[0046] Example 3: In this example, the hardware physical implementation details, sensor material characteristics, and robust design scheme for long-term continuous monitoring of the real-time monitoring and early warning system for immune biomarkers based on biosensors are described in detail.

[0047] The biosensor acquisition unit employs a highly integrated biochip architecture. Its first and second sensing channels share a miniaturized potentiostat circuit. The potentiostat is configured to precisely lock the potential of the working electrode relative to the reference electrode at the peak potential of a specific immune marker's redox reaction using a closed-loop control algorithm. To suppress ambient background current, the biosensor acquisition unit employs a three-electrode system, comprising a gold working electrode, a platinum counter electrode, and a reference electrode made of saturated calomel or silver chloride.

[0048] For the detection of interleukin-6, the working electrode surface is modified with directionally aligned polyclonal antibodies. These polyclonal antibodies are anchored to the electrode surface using a self-assembled monolayer technique, forming a highly oriented trapping layer. To improve the sensor's linear range and response speed, carbon nanotubes or graphene nanocomposites are also introduced onto the electrode surface to enhance electron transport rates and increase specific surface area.

[0049] For the monitoring of complement components, the sensor employs an impedance spectroscopy-based principle. The biosensor acquisition unit is configured to apply a weak, multi-frequency sinusoidal AC perturbation signal to the electrode and measure the phase and amplitude of the feedback current in real time. When complement molecules bind to specific receptors on the electrode surface, a regular change in the interfacial charge transfer resistance occurs. The data preprocessing unit extracts resistance parameters characterizing complement concentration through equivalent circuit fitting.

[0050] At the hardware protection level, since implantable or wearable devices are in a biofluid environment for extended periods, the biosensor acquisition unit employs a packaging design with excellent biocompatibility. Its outer shell is made of medical-grade titanium alloy or polyetheretherketone (PEEK) material, coated with a polyethylene glycol coating that resists non-specific protein adsorption. This coating is configured to prevent fibrin from forming a membrane in the sensor's sensitive area, ensuring that immune markers can diffuse smoothly to the sensing surface and guaranteeing the long-term authenticity of the time-series data.

[0051] The data preprocessing unit is based on an ultra-low-power system-on-a-chip (SoC). This SoC integrates a hardware-accelerated vector processing unit, configured to perform Fast Fourier Transform (FFT) and wavelet denoising operations in parallel. The filtering parameters in the adaptive filtering algorithm are stored in on-chip non-volatile memory and dynamically optimized based on the real-time calculated signal-to-noise ratio (SNR).

[0052] The personalized baseline modeling unit supports local encrypted storage of data at the hardware level. To protect patient privacy, the original physiological voltage signals are immediately assigned a unique anonymous identifier after conversion. The personalized baseline modeling unit is configured to generate encryption keys using an on-chip random number generator, ensuring that even if the device is lost, the locally stored historical baseline data cannot be illegally deciphered.

[0053] The physical implementation of the nonlinear dynamics analysis unit employs a dual-core collaborative architecture. The first core is responsible for basic data flow management and logic scheduling, while the second core, as a dedicated digital signal processing core, is responsible for performing high-intensity matrix operations to solve for the time autocorrelation coefficient in real time. The size of the sliding time window is programmable, ranging from 10 minutes to 4 hours, with the specific value automatically switched by the cloud analysis layer based on the patient's current condition stability.

[0054] The medication record synchronization unit supports multiple near-field communication protocols. It includes a near-field communication module and a low-power Bluetooth module. When a patient uses the smart drug delivery device, the drug delivery information is synchronized seamlessly at the millisecond level via near-field communication technology. The medication record synchronization unit also has a manual record verification mechanism. If the system detects fluctuations in monitoring data that conform to the pharmacodynamic characteristics of the drug but no drug delivery record has been received, it will automatically send a verification request to the patient to ensure the integrity of covariate information.

[0055] The risk warning output unit includes multi-mode alert hardware. In addition to terminal software push notifications, the wearable device integrates a miniature vibration motor and multi-colored LEDs. When a Level 3 emergency intervention alarm is detected, the device will activate a specific vibration sequence and a high-frequency red flashing light, using physical senses to remind the patient to take immediate action. The risk warning output unit is configured to automatically activate an emergency communication link, attempting to dial a preset emergency contact number or send a distress text message with geolocation information.

[0056] The system also features self-diagnostic and fault-tolerant logic. The data preprocessing unit is configured to continuously monitor the impedance characteristics of the sensors. If the electrode impedance is found to exceed the normal aging range, it is determined that the sensor may have failed or detached. The risk warning output unit will output a system fault indication, rather than an erroneous warning signal. If the nonlinear dynamics analysis unit detects a non-physical abrupt change in the variance of the input data during the calculation process, it will automatically trigger a data verification procedure to determine whether there is any sudden environmental interference by comparing historical trends.

[0057] Example 4: This example focuses on describing the specific working logic of the real-time monitoring and early warning system for immune biomarkers based on biosensors in clinical application scenarios, especially how to handle the nonlinear coupling relationship between multiple biomarkers and how to use critical slowing indicators for preventive intervention guidance.

[0058] In actual clinical management, immune system instability is often not driven by a single indicator, but rather is the result of a coordinated imbalance across multiple systems. The nonlinear dynamics analysis unit is configured to perform high-dimensional state space reconstruction. This unit uses interleukin-6, complement component-3, and complement component-4 as three independent dimensions to construct a vector space representing the immune state. The nonlinear dynamics analysis unit is configured to calculate the eigenvalues ​​of the system's trajectory within this vector space.

[0059] When the system is in a healthy remission phase, its trajectory is confined within a stable attractor region, and the mutual evolution of data in each dimension exhibits clear linear or low-dimensional nonlinear characteristics. When the system tends towards instability, the nonlinear dynamics analysis unit detects that the system trajectory begins to shift towards the attractor edge. At this point, the time autocorrelation calculation module is configured not only to calculate the autocorrelation of a single dimension but also to calculate the cross-correlation coefficients between each dimension.

[0060] If the hysteretic cross-correlation between interleukin-6 and complement components increases, it indicates that the cascade effect of the immune response is strengthening, and the feedback regulation mechanism is failing. The nonlinear dynamics analysis unit uses this enhanced coupling phenomenon between dimensions as an important auxiliary criterion for critical slowing down.

[0061] The decision engine of the risk warning output unit integrates a database of clinical guidelines experts. When the system identifies that the immune system is in a critical state of instability, the risk warning output unit is configured to retrieve clinical recommendations that match the current risk level. For example, under a Level 2 warning, the system outputs warning information that includes guidance suggesting increased hydration, reduced strenuous exercise, and increased frequency of manual testing for immune markers.

[0062] The individualized baseline modeling unit is also configured to identify "spurious instability" phenomena. For example, immune indicators may fluctuate when a patient is in a non-disease-specific infection state such as a cold. By comparing data characteristics from historical infection periods, the individualized baseline modeling unit guides the nonlinear dynamics analysis unit to perform differential diagnosis, avoiding unnecessary clinical intervention.

[0063] In this embodiment, the data preprocessing unit further integrates a deep learning-based missing value prediction model. During long-term monitoring, short-term sensor interruptions inevitably occur due to bathing, turning over in sleep, or other reasons. The data preprocessing unit utilizes a long short-term memory neural network model to perform high-fidelity prediction and filling of missing segments based on the dynamic trends before the interruption, ensuring the continuity of nonlinear dynamic analysis as the sliding window moves.

[0064] The risk warning output unit is also configured to integrate with a closed-loop drug delivery system. With physician authorization and patient consent, when the system determines that the condition has definitively crossed the critical point and entered the early stage of an outbreak, the risk warning output unit sends a control command to an external automated drug delivery pump to execute a pilot drug booster, attempting to push the system back into the stable attractor region before clinical symptoms appear. This proactive intervention mode based on phase transition prediction represents a core technological leap forward compared to traditional threshold alarms.

[0065] Example 5: This example describes the integration method of the system of the present invention in a multi-device networking and telemedicine ecosystem.

[0066] In the distributed system, the cloud analytics layer is configured to support simultaneous online monitoring of tens of thousands of patients. To achieve efficient data retrieval and analysis, the cloud analytics layer employs a distributed storage architecture based on a non-relational database. The monitoring data for each patient is divided into multiple data shards, each with a timestamp index.

[0067] The individualized baseline modeling unit utilizes generative adversarial networks for data augmentation in the cloud. For newly enrolled patients with insufficient data accumulation, the individualized baseline modeling unit is configured to extract features from desensitized data of patients with similar pathological characteristics to generate simulated physiological fluctuation baselines, serving as an initial "virtual baseline model" for these patients. As real patient monitoring data is injected, the virtual baseline model is gradually replaced by the real individualized model through a Bayesian update mechanism.

[0068] The nonlinear dynamics analysis unit achieves large-scale parallel computing in the cloud. For each patient, the cloud analysis layer simultaneously runs multiple analysis engines with different parameter settings. One engine is configured to focus on capturing very short-term (minute-level) sudden risks, while another engine is configured to focus on analyzing the evolution of immune tension in the medium to long term (weekly). The risk warning output unit aggregates the evaluation results from multiple engines and outputs the final risk decision using a weighted voting mechanism.

[0069] The system also includes a medical collaboration interface subunit. This subunit is configured to provide attending physicians with a dedicated monitoring screen interface. The interface displays the immune homeostasis distribution of all patients under their management in the form of a heatmap. Patients in a critical state of slowed immune progression are automatically highlighted, and the evolution curves of their various nonlinear dynamic indicators are displayed.

[0070] The data preprocessing unit also performs cross-device collaborative calibration logic in the cloud. If the same patient wears both wearable and implantable sensors, the data preprocessing unit is configured to analyze the correlation between their signals. If outlier fluctuations are detected in one device, the system will automatically adjust the weights of the cross-device calibration performed in the cloud and compensate for the analysis process using data from the other device.

[0071] The risk warning output unit is also configured to include "sensitivity factor analysis" when generating reports. This analysis, through regression analysis of historical data, identifies which lifestyle factors are highly correlated with immune system instability. This feature upgrades the warning system from a simple disease monitoring tool to a decision support platform that assists patients in self-health management.

[0072] Example 6: This example illustrates the reliability assurance mechanism of the system under extreme environments and the intelligent strategy for managing the lifespan of biosensors.

[0073] The biosensor acquisition unit integrates an impedance monitoring circuit for real-time assessment of the surface activity of the sensor electrodes. During long-term continuous acquisition, biofouling or chemical passivation may occur on the electrode surface. The data preprocessing unit is configured to periodically apply specific electrocleaning pulse signals to the electrodes, removing adsorbed impurity proteins through a small electrolytic reaction.

[0074] The nonlinear dynamics analysis unit features automatic gain control when identifying instability signals. When the signal-to-noise ratio of the monitored signal decreases due to sensor aging, the system automatically increases the length of the sliding time window. By trading time for space, it utilizes a longer data sequence to extract a stable autocorrelation coefficient, ensuring that early warning accuracy is maintained even at the end of the sensor's lifespan.

[0075] The individualized baseline modeling unit is configured to record sensor replacement cycles. Whenever a patient replaces a new sensor consumable, the individualized baseline modeling unit automatically initiates a 24-hour rapid calibration process. During this period, the system uses the end-of-cycle data from the previous sensor and the initial data from the new sensor to perform feature matching, enabling a seamless switch in monitoring logic.

[0076] The medication record synchronization unit also has pharmacokinetic simulation capabilities. It internally stores pharmacodynamic models of commonly used immunomodulatory drugs. When the system detects signs of rebound instability in the immune system due to drug concentration decay, the risk warning output unit will send an early reminder to the patient, prompting them to take their medication promptly and preventing systemic phase transitions caused by missed doses.

[0077] The risk warning output unit has local decision-making priority during communication link interruptions. Edge nodes are configured to store the most simplified nonlinear dynamic discrimination logic. If an edge node locally determines that a patient is in a critical condition and cannot contact the cloud, it will skip the cloud verification process and directly connect to nearby mobile devices via local Bluetooth or send a distress signal through the device's local alarm device.

[0078] The real-time monitoring and early warning system for immune biomarkers based on biosensors provided in the above embodiments of the present invention achieves prospective interpretation of complex immune system state transformations by combining the critical slowing theory in nonlinear dynamics with high-sensitivity biosensing technology. The system can not only identify early instability signals that are difficult to detect using conventional threshold methods, but also improve the clinical specificity and reliability of the early warning through individualized modeling, drug efficacy covariate stripping, and multi-level cross-validation, providing a solid technical guarantee for the precise and proactive health management of patients with chronic diseases.

[0079] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A real-time monitoring and early warning system for immune biomarkers based on biosensors, characterized in that, include: A biosensing acquisition unit is configured to continuously acquire time-series concentration data of immune markers in a patient's body via implantable or wearable biosensors, the immune markers including interleukin-6 and complement components. A data preprocessing unit, connected to the biosensor acquisition unit, is configured to perform noise filtering, signal calibration, and sampling frequency unification processing on the time series concentration data to generate a standardized dynamic monitoring data stream. The individualized benchmark modeling unit is configured to construct an individualized immune homeostasis benchmark model based on the patient's historical monitoring data. The individualized immune homeostasis benchmark model characterizes the individualized fluctuation range and dynamic balance characteristics of each immune marker in the patient during the clinical remission period. The nonlinear dynamics analysis unit is connected to the data preprocessing unit and the individualized benchmark modeling unit, respectively. It is configured to perform real-time calculation of the time autocorrelation coefficient and fluctuation variance of the dynamic monitoring data stream based on the critical slowing theory, and compare the calculation results with the individualized immune homeostasis benchmark model to identify early signs of decreased immune system resilience. A risk warning output unit is connected to the nonlinear dynamics analysis unit and is configured to generate a graded warning signal and push it through a secure communication link when the nonlinear dynamics analysis unit determines that the immune system is in a critical unstable state.

2. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The biosensor acquisition unit adopts a multi-channel synchronous sensing architecture, which includes a first sensing channel and a second sensing channel. The first sensing channel integrates a biofunctionalized surface targeting interleukin-6. The biofunctionalized surface is modified with a polyclonal antibody capture layer that is oriented and anchored by self-assembled monolayer technology. Below the polyclonal antibody capture layer is a conductive reinforcement layer containing carbon nanotubes or graphene nanocomposite materials. The second sensing channel integrates an impedance sensor for complement component III or complement component IV; the biosensing acquisition unit also includes a miniaturized potentiostat circuit, which locks the potential of the working electrode relative to the reference electrode at the peak potential of the redox reaction of the immune marker through a closed-loop control algorithm. The biosensor acquisition unit adopts a three-electrode system structure, specifically including a gold working electrode, a platinum counter electrode, and a reference electrode made of saturated calomel or silver chloride.

3. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The data preprocessing unit has a built-in adaptive filtering algorithm. The adaptive filtering algorithm adopts a dynamic compensation model based on Kalman filtering and is configured to automatically optimize the filtering parameters according to the drift characteristics of the biosensor and the environmental interference mode. The data preprocessing unit also includes a signal calibration subunit and a resampling module; the signal calibration subunit is configured to perform zero-point calibration and range calibration, and converts the original signal into a standardized concentration physical quantity by comparing the original current signal or potential signal with the pre-stored sensor characteristic curve. The resampling module is configured to use a spline interpolation algorithm to fill in the data at irregular time points caused by communication packet loss, and output the dynamic monitoring data stream with a uniform sampling frequency.

4. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The biosensing acquisition unit also includes an environmental compensation subunit, which is equipped with a temperature sensor and a hydrogen ion concentration index sensor to monitor the temperature and pH parameters of the microenvironment in which the biosensor is located in real time. The environmental compensation subunit sends the temperature parameter and the pH parameter as correction factors to the data preprocessing unit. The data preprocessing unit is configured to compensate the time series concentration data according to the correction factor in order to eliminate signal drift caused by microenvironmental fluctuations.

5. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The individualized baseline modeling unit is configured to enter learning mode at the initial stage of system startup or after the patient's condition is reset, and to extract the distribution characteristics of immune markers of the patient over a continuous period of no less than 14 days and nights. The individualized immune homeostasis benchmark model includes the numerical mean, numerical variance, and potential surface features that reflect the depth of the system's homeostasis for each of the immune biomarkers. The individualized baseline modeling unit has a built-in recursive learning algorithm, which is configured to achieve dynamic adaptive evolution of the individualized immune homeostasis baseline model by incorporating new homeostasis features into the existing model after the patient enters a new clinical remission period. The individualized baseline modeling unit is also configured to use a generative adversarial network to extract features from desensitized data of patients with similar pathological characteristics, generate a virtual baseline model in the early stage of patient monitoring, and replace the virtual baseline model with the real individualized model through a Bayesian update mechanism as real monitoring data is injected.

6. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The nonlinear dynamics analysis unit includes an autocorrelation calculation module, a variance monitoring module, and a recovery rate evaluation module. The autocorrelation calculation module is configured to calculate the first-order time autocorrelation coefficient of the monitoring data within the current window when the sliding time window is stepped. The first-order time autocorrelation coefficient is obtained by calculating the covariance between the data at the current time and the data at the previous time and performing standardization processing. The variance monitoring module is configured to synchronously calculate the fluctuation variance within the sliding time window; The recovery rate assessment module is configured to calculate the recovery rate of the immune marker concentration per unit time and determine whether the system relaxation time has increased. The length of the sliding time window is set within the range of 10 minutes to 4 hours based on the physiological half-life of the immunomarker. When the recovery rate is lower than a preset rate threshold, and the first-order time autocorrelation coefficient exceeds a first preset warning line and the fluctuation variance exceeds a second preset warning line, the nonlinear dynamics analysis unit determines that the immune system is losing its resilience and approaching the phase transition critical point.

7. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The system also includes a medication record synchronization unit, which is configured to receive and integrate the patient's immunosuppressant or biologic medication time information, dosage information, and pharmacokinetic parameters. The medication record synchronization unit inputs drug-induced concentration fluctuations as covariates into the nonlinear kinetic analysis unit. The nonlinear dynamics analysis unit is configured to perform residual analysis logic, separating the drug-directed fluctuation components from the total fluctuation signal and analyzing only the fluctuation components caused by changes in the dynamic characteristics of the immune system itself.

8. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The risk warning output unit executes multi-level warning logic, and the graded warning signal includes three levels: The first level of warning is an observation recommendation, corresponding to an increase in the first-order time autocorrelation coefficient of the appearance of a single immune marker; The second level of warning is a clinical follow-up visit reminder, which corresponds to two or more of the aforementioned immune markers simultaneously exhibiting critical slowing characteristics, and the decrease in the recovery rate exceeding a preset proportion. The third level of warning is an emergency intervention alert, corresponding to the value of the immune marker deviating from the individualized immune homeostasis benchmark model, and the nonlinear dynamic index indicating that the system has crossed the preset critical threshold. Before triggering the third-level warning, the risk warning output unit executes cross-validation logic, requiring interleukin-6 and complement components to synchronously exhibit statistically significant abnormal shifts within a preset time window.

9. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The system adopts a distributed architecture based on edge computing and cloud collaboration, and is divided into an edge processing layer and a cloud analysis layer. The edge processing layer is deployed at the patient end and includes a system-on-a-chip with a hardware-accelerated vector processing unit, configured to perform high-frequency sampling, filtering, calibration, and basic autocorrelation calculations based on a short sliding window. The cloud analytics layer consists of a server cluster, and the individualized baseline modeling unit is deployed in the cloud analytics layer. The cloud-based analysis layer is also configured to perform phase space reconstruction analysis, which assesses the overall degree of chaos in the patient's immune system from a macroscopic perspective by calculating the delayed embedding dimension and the Lyapunov index. When the Lyapunov exponent changes from negative to positive or its value increases, the system is determined to be evolving into a positive, unstable, chaotic state. The edge processing layer also includes a data compression submodule, configured to process data using a lossless compression algorithm during signal stability periods, and to automatically switch to the original data pass-through mode when the nonlinear dynamics index is detected to be near a warning threshold.

10. The real-time monitoring and early warning system for immune biomarkers based on biosensors according to claim 1, characterized in that, The biosensor acquisition unit adopts a biocompatible packaging structure, and its shell is made of medical-grade titanium alloy or polyether ether ketone material. The surface of the shell is coated with a polyethylene glycol coating that resists non-specific protein adsorption. The polyethylene glycol coating is configured to prevent fibrin from forming a membrane in the sensitive area of ​​the biosensor. The biosensor acquisition unit integrates an impedance monitoring circuit for real-time evaluation of the electrode's surface activity. The data preprocessing unit is configured to periodically apply an electro-cleaning pulse signal to the working electrode to remove adsorbed impurity proteins from the surface through an electrolytic reaction. The system is also equipped with self-diagnostic logic. When the impedance monitoring circuit detects that the electrode impedance exceeds the preset aging range, the risk warning output unit outputs a system fault prompt.

Images