Emergency rescue immersive visual communication system based on virtual reality

Abstract

The invention provides an emergency rescue immersive visual communication system based on virtual reality, and relates to the technical field of augmented reality. Physical deformation data and tissue metabolism data of a patient are deeply fused by introducing a space-time coupling dynamic model, and the technical span from state monitoring to risk prediction is realized; the space-time coupling perfusion-metabolism kinetic model is configured to fuse the first data representing the physical deformation state and the second data representing the tissue metabolism state so as to generate a predictive indication signal for early warning the metabolic failure risk at the future moment. The conversion from passive response to active prevention is realized, and a precious time window is won for rescue by providing prospective risk early warning; and meanwhile, the decision-making efficiency and accuracy of remote guidance are improved, and the cognitive load and misjudgment of experts are reduced by superposing multi-dimensional dynamic early warning information containing risk levels, deterioration trends and fundamental inducements in an augmented reality view.

Description

Technical Field

[0001] This invention relates to the field of extended reality technology, specifically to an immersive visual communication system for emergency care based on virtual reality. Background Technology

[0002] With the deep integration of information technology and healthcare, extended reality (XR), the Internet of Things (IoT), and artificial intelligence (AI) technologies are gradually reshaping modern healthcare service models. Particularly in emergency care, where time efficiency and decision-making accuracy are paramount, leveraging advanced digital technologies to empower remote experts, overcome geographical limitations, and improve the success rate of treating critically ill patients has become a crucial development direction in smart healthcare. Building next-generation remote communication systems capable of providing deep situational awareness and proactive decision support is a core issue in current technological evolution.

[0003] Despite the progress made by existing telemedicine systems, they still face the following core challenges when dealing with complex emergency rescue scenarios: Existing technologies, including the scheme published in CN115620924A, while capable of collecting multi-source information and constructing a three-dimensional model, essentially present geometric information and vital sign data as two independent dimensions. Remote experts see a three-dimensional human body model and a set of parallel, constantly fluctuating physiological parameters. This method of information presentation lacks inherent coherence; experts must rely on their own experience to manually establish complex causal relationships between physical spatial changes and abnormal physiological indicators. This not only increases cognitive load but also makes it difficult to reveal the dynamic coupling relationship between the two in spatiotemporal evolution, limiting a deep understanding of the true nature of the condition.

[0004] Current mainstream technological paradigms primarily focus on the real-time "monitoring" and "observation" of a patient's current state. The system can accurately inform experts "what is happening," such as swelling of the body surface or a physiological indicator falling below a threshold. However, it cannot answer a more crucial question: "What is most likely to happen next?" Due to the lack of predictive models that integrate physical damage processes (such as continuous deformation caused by internal bleeding) with physiological deterioration trajectories (such as metabolic collapse caused by decreased tissue perfusion), existing technologies inherently lag in risk assessment, making it difficult to provide decision-valuable early warnings before critical situations occur. This often results in remote intervention being a passive response rather than proactive prevention. Summary of the Invention

[0005] The purpose of this invention is to provide an immersive visual communication system for emergency rescue based on virtual reality, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A virtual reality-based immersive visual communication system for emergency resuscitation, specifically including: The geometric model acquisition module is used to acquire a three-dimensional geometric model of a local area of ​​the target patient's body in real time, and generate first data characterizing the physical deformation state of the local area based on the changes of the three-dimensional geometric model within a preset time window. The tissue state acquisition module is used to acquire second data of the tissue metabolic state corresponding to the local area space in real time through near-infrared spectral imaging. The risk prediction processing module is configured to fuse the first data and the second data using a pre-built spatiotemporally coupled perfusion-metabolic kinetic model to generate a predictive indication signal characterizing the risk of metabolic failure in the local region at a future preset time point; and The augmented reality rendering module is used to generate and overlay dynamic augmented reality early warning information on the corresponding high-risk areas on the three-dimensional geometric model in the extended reality view provided to remote experts, based on the predictive indication signal.

[0007] Compared with existing technologies, the advantages of this invention are: by creating a "spatiotemporally coupled perfusion-metabolic kinetic model" as a "translator" and "prediction engine" for fusing two heterogeneous data sets, three-dimensional geometric deformation (first data) and tissue metabolic state (second data) are represented as two different dimensions of the same pathophysiological process, and a quantitative mapping relationship between them is established based on biophysical principles. The model can understand how non-periodic tissue expansion (physical deformation) at a specific rate will affect microcirculation perfusion, ultimately leading to a decrease in tissue blood oxygen saturation (metabolic state) at a specific future time point.

[0008] Building upon this, forward extrapolation using the aforementioned coupled model generates a "predictive indicator signal." This signal no longer represents the current state, but rather the probability and time trajectory of future metabolic failure. This shift from "state description" to "trajectory prediction" opens an unprecedented decision-making window for remote experts, enabling them to anticipate and intervene in impending crises.

[0009] Furthermore, this invention recognizes that an effective early warning should not only inform of "risk," but also explain "how urgent the risk is" and "where the risk originates." To this end, a hierarchical, state-adaptive augmented reality visual coding mechanism is designed. It characterizes the "deterioration trend" of the risk by analyzing the time derivative of predictive indicator signals, and reveals the "primary triggers" of the risk by analyzing the intrinsic composition of the signal (such as the contribution ratio of periodic physiological fluctuations to non-periodic pathological changes). This multidimensional dynamic early warning information, which integrates "risk level," "deterioration trend," and "primary triggers," achieves a synergistic gain in information utility, providing experts with a basis for decision-making. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram showing the three-dimensional deformation velocity field and tissue blood oxygen saturation field superimposed on a three-dimensional geometric model of the human lower limb; Figure 3 This is a schematic diagram of the execution logic of the geometric model acquisition module, the organization status acquisition module, and the risk prediction and processing module of the present invention; Figure 4 This is a schematic diagram of the execution logic of the augmented reality rendering module of the present invention. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0012] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various elements, but unless otherwise stated, these elements are not limited by these terms. These terms are used only to distinguish one element from another.

[0013] Example 1: Please see Figures 1 to 4 The present invention provides a technical solution: An immersive visual communication system for emergency resuscitation based on virtual reality, comprising: The geometric model acquisition module is used to acquire the three-dimensional geometric model of a local area of ​​the target patient's body in real time, and generate first data representing the physical deformation state of the local area based on the changes of the three-dimensional geometric model within a preset time window. The tissue state acquisition module is used to acquire second data on the tissue metabolic state corresponding to a local area in real time through near-infrared spectral imaging. The risk prediction processing module is configured to fuse the first and second data using a pre-built spatiotemporally coupled perfusion-metabolic kinetic model to generate a predictive indication signal characterizing the risk of metabolic failure in a local region at a future preset time point; and The augmented reality rendering module is used to generate and overlay dynamic augmented reality early warning information on high-risk areas on a 3D geometric model in an extended reality view provided to remote experts, based on predictive indication signals.

[0014] The core of this embodiment lies in achieving nonlinear, adaptive intelligent fusion based on working condition identification. Its core idea is to decompose the complex raw data (deformation velocity field) into purer feature components (periodic and non-periodic deformation) with clear physical meaning. Then, these feature components are independently and specifically modeled. Finally, based on the current macroscopic state of the system (characterized by accumulated organizational stress), the relative importance of these independent models in the final decision-making process is dynamically and intelligently determined. This cascaded architecture of "decomposition-independent modeling-dynamic fusion" enables the system to simulate the hierarchical reasoning process of human experts. This provides a fundamental guarantee for achieving the technological leap from "condition monitoring" to "trajectory prediction." The core technical feature of this invention lies in the critical state early warning mechanism. This mechanism includes, but is not limited to, the collaborative work of the following three key modules: The geometric model acquisition module is used to acquire first data characterizing the local physical deformation state of the patient's body. The tissue state acquisition module is used to acquire second data on the tissue metabolic state corresponding to the local space; and The risk prediction and processing module is based on a spatiotemporally coupled perfusion-metabolic kinetic model. This model is configured to fuse first and second data to generate a predictive indicator signal for early warning of the risk of metabolic failure at future times.

[0015] Furthermore, it should be noted that: Figure 2 The three-dimensional geometric model of the injured person's lower limb, obtained through methods such as 3D scanning, is presented as a 3D mesh and forms the basis of all data visualizations. Two data fields are overlaid on the surface of the 3D geometric model. The first is a 3D deformation velocity field, represented by a set of discrete vector arrows. Each vector arrow originates at a vertex of the model, and its direction and length characterize the displacement direction and velocity of that point in 3D space, respectively. As shown in the figure, in the region anterior to the tibia of the lower limb, the length of the vector arrows is greater than in other regions, indicating that rapid physical expansion is occurring in this area.

[0016] The second is the tissue oxygen saturation field, represented by a dotted pattern of varying densities on the surface. The density of the dotted pattern is inversely proportional to the oxygen saturation value of the subcutaneous tissue. In the same anterior lower leg region, the high density of the dotted pattern creates a visually dark area, indicating severe hypoxemia in the subcutaneous tissue of that region. In other areas, the dotted pattern is very sparse, indicating normal tissue oxygen saturation. In this way, Figure 2 It clearly reveals the strong spatial correlation between high-speed physical deformation and severe tissue hypoxia, providing intuitive visual evidence for rapid diagnosis by remote experts.

[0017] Further explanation: The geometric model acquisition module includes a synchronous positioning and mapping unit, which uses data from the depth sensor and the inertial measurement unit to generate a three-dimensional geometric model in real time. Furthermore, the first data is the three-dimensional deformation velocity field of the vertices on the surface of the three-dimensional geometric model within a preset time window. The second set of data includes local tissue oxygen saturation data and changes in total hemoglobin concentration in the local area.

[0018] Further explanation: The spatiotemporal coupled perfusion-metabolic kinetic model is configured to: establish a temporal coupling relationship between the physical deformation represented by the first data and the tissue metabolic state represented by the second data; and based on the temporal coupling relationship, predict the rate of change of tissue blood oxygen saturation in a local area at a future preset time point as a predictive indicator signal.

[0019] Further explanation: The risk prediction processing module is further configured to perform the following steps to generate predictive indication signals: The first data is subjected to deformation mode decomposition to identify and separate at least periodic deformation components and non-periodic deformation components. Secondly, using the first sub-model, the first metabolic influencing factor was calculated based on the periodic deformation component; Using the second sub-model, the second metabolic influencing factor was calculated based on the non-periodic deformation component; Dynamic weighted fusion of the first and second metabolic influencing factors was performed to determine the rate of change in tissue blood oxygen saturation.

[0020] Further explanation: In a preferred embodiment of the present invention, the risk prediction processing module performs in-depth processing on the data provided by the geometric model acquisition module and the tissue state acquisition module to achieve accurate prediction of the future metabolic state of the patient's local tissues; in this embodiment, the following key parameters are involved: The three-dimensional deformation velocity field is denoted as V. fieldThis represents the set of three-dimensional spatial displacement vectors of each vertex on the surface of a three-dimensional geometric model of a local area of ​​the patient's body within a unit of time. This embodiment employs a composite sensor module integrating a time-of-flight (ToF) depth camera and a six-axis inertial measurement unit (IMU). The ToF camera outputs a 640×480 pixel depth image at a rate of 30 frames per second, where each pixel value represents the straight-line distance from that point to the sensor. The IMU outputs three-axis acceleration and three-axis angular velocity data at a rate of 200 Hz.

[0021] The raw depth image is processed by a median filter to remove random salt-and-pepper noise. Subsequently, a bilateral filter is used to smooth the depth data while preserving edge sharpness. The raw IMU data is processed by a Kalman filter to fuse accelerometer and gyroscope data, resulting in a smoother and more accurate attitude estimate. The system runs VINS-Fusion, a visual-inertial SLAM algorithm. This algorithm extracts ORB feature points from the gradient map of the RGB or depth image and combines them with IMU data for attitude calculation and feature point triangulation, thereby generating a 3D point cloud model with precise pose information in each frame. In this embodiment, to determine the 3D deformation velocity field, the system performs the following steps: The system maintains a circular buffer storing the point cloud models of the most recent N frames and their corresponding timestamps. For the point cloud models of the current frame and the previous frame, the Iterative Closest Point (ICP) algorithm is applied. This algorithm uses the previous frame's point cloud as the source and the current frame's point cloud as the target, calculating an optimal rigid transformation matrix that aligns the overlapping areas between the two frames. For each vertex in the previous frame's point cloud, the rigid transformation matrix is ​​applied to its coordinates to obtain its predicted position in the current frame's coordinate system. Then, the three-dimensional Euclidean distance vector between this predicted position and the vertex's actual matching point position in the current frame is calculated; this is the vertex's displacement vector within the time interval Δt. The displacement vector calculated for each vertex is divided by the time interval Δt to obtain the vertex's three-dimensional velocity vector. The set of all vertex velocity vectors constitutes the final output three-dimensional deformation velocity field V, representing the physical deformation state. field .

[0022] The tissue oxygen saturation field is denoted as StO2. field This characterizes the spatial distribution of local tissue oxygen saturation values ​​in the subcutaneous tissue region corresponding to the surface space of the three-dimensional geometric model. The determination is based on the Modified Beer-Lambert Law (MBLL). This law describes the relationship between the attenuation of light intensity and the concentration of absorbing substances as light passes through the scattering medium characterized by biological tissue. The tissue state acquisition module performs the following steps to determine the tissue oxygen saturation values: The NIRS probe inside the module alternately emits near-infrared light of the first and second wavelengths into the tissue, and simultaneously measures the light intensity values ​​received by the detector at a fixed distance after being scattered by the tissue, which are denoted as I(λ1) and I(λ2), respectively; in this embodiment, the first wavelength and the second wavelength are λ1=760nm and λ2=850nm, respectively. The system calculates the optical density at two wavelengths by taking the natural logarithm of the ratio of emitted light intensity to received light intensity. The resulting optical densities at the two wavelengths are OD(λ1) and OD(λ2).

[0023] Subtracting the initial baseline value from the current measured optical density value yields the changes in optical density ΔOD(λ1) and ΔOD(λ2); the initial baseline value is the stable reading at the start of the measurement. According to the MBLL principle, there is a linear relationship between the changes in optical density and the changes in oxyhemoglobin concentration (Δ[HbO2]) and deoxyhemoglobin concentration (Δ[Hb]) in the tissue. These two concentration changes are calculated by solving a system of two linear equations. The calculation logic is as follows: ΔOD(λ1) equals Δ[HbO2] multiplied by the extinction coefficient of HbO2 at λ1, plus Δ[Hb] multiplied by the extinction coefficient of Hb at λ1, and then multiplied by the distance between the light source and the detector and the differential path length factor.

[0024] ΔOD(λ2) equals Δ[HbO2] multiplied by the extinction coefficient of HbO2 at λ2, plus Δ[Hb] multiplied by the extinction coefficient of Hb at λ2, and then multiplied by the distance between the light source and the detector and the differential path length factor.

[0025] Tissue oxygen saturation is defined as the percentage of oxyhemoglobin concentration relative to total hemoglobin concentration. It is calculated using the following logic: the calculated change in oxyhemoglobin concentration (Δ[HbO2]) is divided by the sum of the changes in oxyhemoglobin and deoxyhemoglobin concentrations, and then multiplied by 100% to obtain the final tissue oxygen saturation value.

[0026] In this embodiment, the extinction coefficients of oxyhemoglobin and deoxyhemoglobin at a specific wavelength are publicly available physical constants, which the system obtains from an internal lookup table. The differential path length factor is a dimensionless correction factor used to compensate for the fact that the actual path light travels in tissue due to scattering is much longer than the straight-line distance between the light source and the detector; this value is related to tissue type and age. In this embodiment, the differential path length factor value was determined through the following calibration experiment: To determine the average differential path length factor value applicable to limb tissues in emergency scenarios, 50 healthy volunteers were recruited. Using the NIRS probe of this invention and a gold-standard frequency-domain NIRS device, optical parameters of their forearm muscles were simultaneously measured under identical conditions during rest, transient ischemic attack (touch band insufflation), and reperfusion. The optical density measured by the device of this invention was compared with the absolute optical path measured by the gold-standard device. The differential path length factor value that best matches the two measurements was calculated using least-squares fitting. Finally, the average differential path length factor value of the 50 volunteers was used as the system's preset differential path length factor constant and stored in a spreadsheet file.

[0027] The periodic deformation component is denoted as V. periodic ; Separate the deformation component with stable frequency and amplitude characteristics caused by the patient's spontaneous breathing or heartbeat from the three-dimensional deformation velocity field; The non-periodic deformation component is denoted as V. aperiodic It represents the portion of deformation that is separated from the three-dimensional deformation velocity field, mainly caused by traumatic swelling, external pressure, or changes in body posture, and lacks stable periodicity.

[0028] Dynamic fusion weights, denoted as W dynamic It represents a dynamically changing value between 0 and 1, used to adjust the contribution of periodic and non-periodic deformations to the final metabolic prediction in fusion calculations.

[0029] The limitation of directly using deformation velocity field and tissue blood oxygen saturation field as inputs for regression prediction is that it cannot distinguish the distinct effects of deformation caused by different factors on physiological metabolism. Stable respiratory deformation is associated with stable metabolism, while rapid swelling deformation indicates severe tissue damage and impending metabolic failure. This single mapping relationship is prone to misjudgment. To address this issue, this embodiment designs an innovative, hierarchical fusion mechanism with dynamic memory effect. The core idea of ​​this mechanism originates from time-frequency analysis in signal processing and adaptive control in cybernetics; the internal working logic of this mechanism is detailed below: The risk prediction and processing module processes the input three-dimensional deformation velocity field V. field The data is processed by applying Fast Fourier Transform (FFT) or Wavelet Transform algorithms to the time-series velocity data of each vertex in the field to analyze its spectral characteristics.

[0030] The system reads preset physiological frequency ranges from an external configuration spreadsheet file, specifically defining 0.1 Hz to 0.5 Hz as the "respiratory frequency band" and 1.0 Hz to 1.5 Hz as the "heart rate band." Using a bandpass filter, the signal components falling within these physiological frequency bands are extracted and reconstructed into periodic deformation components V. periodicSubtracting the periodic deformation component from the original three-dimensional deformation velocity field yields the residual signal, which is the non-periodic deformation component V. aperiodic .

[0031] The first sub-model is used to assess the contribution of periodic deformation to metabolic stability. Its computational model is as follows: Calculate the periodic deformation component V. periodic The stability of amplitude and frequency over a 5-second period is calculated, specifically its standard deviation. Next, this stability value is non-linearly mapped using a pre-defined standard logistic function to convert it into a physiological stability index. The calculation logic for this physiological stability index is as follows: Obtain the currently calculated "periodic deformation stability value" and read two preset parameters from the configured spreadsheet file: a stability midpoint parameter and a steepness parameter.

[0032] Calculate the difference between the stability midpoint parameter and the periodic deformation stability value, and then multiply the difference by the steepness parameter.

[0033] Calculate the product obtained in the previous step raised to the power of the natural constant e. Add the power obtained to the value 1, and then divide the sum by the value 1 to obtain the final physiological stability index with the range normalized to the interval [0,1].

[0034] In this embodiment, the preferred value for the stability midpoint parameter is 0.05, which represents the critical value of the clinically acceptable standard deviation of periodic deformation. The preferred value for the kurtosis parameter is 50, which determines the transition sensitivity of the mapping curve near the critical value. These two parameters were determined by receiver operating characteristic curve analysis of clinically collected respiratory pattern data labeled as "stable" and "unstable" by experts, and are the optimal values ​​that maximize classification accuracy; this is the first metabolic influence factor.

[0035] The second sub-model is used to assess the driving effect of aperiodic deformation on metabolic deterioration. It introduces a "tissue stress accumulation" state variable. The calculation model is as follows: at each time step, the square of the amplitude of the current aperiodic deformation component is multiplied by a preset time decay coefficient and added to the "tissue stress accumulation" value from the previous time step to update the current accumulation value. This accumulation value is then normalized to the [0,1] interval, which is the second metabolic influence factor. This design reflects the cumulative effect of damage.

[0036] In this embodiment, the "time decay coefficient" is a dimensionless parameter ranging from 0 to 1, used to simulate the physiological dissipation or repair effect of biological tissues on stress damage in the "tissue stress accumulation" model. A higher coefficient value means that the influence of historical stress decays rapidly, suitable for tissues with good elasticity; a lower coefficient value means that the stress influence persists for a long time, suitable for simulating plastic damage or hematoma accumulation. The preferred value of this time decay coefficient was determined through the following in vitro tissue model calibration experiments: A miniature mechanical testing system capable of applying precisely controlled normal pressure was employed. The experimental sample was a tissue phantom with biomechanical properties similar to human soft tissue, made from a pre-concentrated agarose gel. An array of pressure sensors was placed beneath the tissue phantom to monitor the internal stress distribution in real time.

[0037] The mechanical testing system is controlled to apply a known, step pressure load to the tissue phantom; in this embodiment, it is a pressure of 50 kPa lasting for 5 seconds.

[0038] After applying and removing pressure, the readings of the pressure sensor array below are continuously recorded at a sampling frequency of 10 Hz until the stress is fully recovered or a new equilibrium state is reached. The collected pressure recovery curves are then fitted using a single exponential decay model. The calculation logic is as follows: the pressure values ​​at different times are represented as the initial peak pressure multiplied by an exponential function with the natural constant e as the base. The exponential part of this exponential function is the product of the negative time constant and time. The optimal time constant is obtained by fitting the curve using a nonlinear least squares method.

[0039] The time constant obtained from the above experiments is converted into a discrete-time decay coefficient corresponding to the time step of the algorithm in this embodiment. In this embodiment, if the algorithm's operating frequency is 10 Hz, the preferred value of the time decay coefficient is determined to be 0.98. In practical applications, the reasonable range for this time decay coefficient is between 0.95 and 0.995. This range is an effective interval determined by repeating the above experiments on tissue phantoms of different hardness, taking into account measurement noise and model errors. This parameter is pre-configured and stored in a spreadsheet file.

[0040] In this embodiment, to determine the final prediction result, the first and second metabolic influencing factors are fused. Specifically, a dynamic weighting mechanism is used, the core of which lies in the dynamic fusion weight W. dynamic .

[0041] The dynamic fusion weights are determined as follows: the goal is to intelligently and non-linearly adjust the contribution of periodic and non-periodic deformations in the final risk prediction model based on the current "stress state" of the tissue, in order to characterize the decision-making logic of clinicians: in a stable state, attention is paid to the stability of respiratory rhythm; in the event of acute injury, attention is completely shifted to the development of the injury itself.

[0042] This embodiment uses a fuzzy logic-based reasoning method to determine the weight, which consists of three parts: fuzzification, rule base reasoning, and defuzzification.

[0043] Input variable: The only input is the "second metabolic influence factor" defined in the previous embodiment, which is the normalized "tissue stress accumulation" value, with a value range of [0,1].

[0044] Fuzzification: Define three fuzzy language sets for the input variables: "low stress", "medium stress" and "high stress", and design membership functions for them (respectively, descending semi-trapezoidal, triangular and increasing semi-trapezoidal functions).

[0045] Rule base construction: The rule base is built from the knowledge of clinical experts and includes the following rules: Rule 1: IF "Accumulated Tissue Stress" IS "Low Stress" THEN "Fusion Weight" IS "Bias Towards Periodicity". Where "IF" means if; "IS" means yes; and "THEN" means then. Rule 2: IF "Accumulated stress" IS "Medium stress" THEN "Fusion weight" IS "Equilibrium".

[0046] Rule 3: IF “Stress Accumulation” IS “High Stress” THEN “Integration Weight” IS “Bias Towards Non-Periodicity”

[0047] The output variable "fusion weight" defines three fuzzy language sets: "biased to periodicity" (center value 0.8), "balanced" (center value 0.5), and "biased to non-periodicity" (center value 0.1). The system uses the centroid method for defuzzification to calculate the precise dynamic fusion weight W within the range [0,1]. dynamic The membership function parameters and rule base of the fuzzy logic system are fully encoded and stored in a spreadsheet file. The system loads these configurations at runtime to instantiate the inference engine.

[0048] The overall calculation process of the spatiotemporally coupled perfusion-metabolic kinetics model executed by the risk prediction and processing module in this embodiment is as follows: At the beginning of each calculation cycle, the risk prediction processing module receives the latest three-dimensional deformation velocity field V from the geometric model acquisition module. fieldThe module receives the latest tissue oxygen saturation field (StO2) from the tissue status acquisition module. field .

[0049] Perform "deformation mode decomposition" to obtain the periodic deformation component V. periodic Aperiodic deformation component V aperiodic .

[0050] Two sub-models are executed in parallel: the first sub-model is used to process the periodic deformation component V. periodic The first metabolic influencing factor was obtained; the aperiodic deformation component V was processed using the second sub-model. aperiodic The second metabolic influencing factor was obtained.

[0051] Based on the current value of the second metabolic influence factor, perform "dynamic weighted fusion" processing to determine the current dynamic fusion weight W. dynamic .

[0052] The first metabolic influencing factor is multiplied by the dynamic fusion weight to obtain the first weighted result; then, the second metabolic influencing factor is multiplied by the result of (1 minus the dynamic fusion weight) to obtain the second weighted result; then, the first weighted result and the second weighted result are added to obtain the comprehensive risk score; the comprehensive risk score is multiplied by the preset maximum failure rate benchmark value to obtain the final predicted rate of change of tissue blood oxygen saturation as a predictive indicator signal.

[0053] In this embodiment, the "maximum rate of exhaustion baseline" is a constant with a defined physical unit of "% / second," providing an upper limit for the predicted rate of change in tissue oxygen saturation that aligns with clinical reality. This parameter converts the normalized "comprehensive risk score" into a quantifiable predictive value with clinical diagnostic significance. The optimal value for this parameter was derived through statistical analysis of a retrospective clinical database containing monitoring data of trauma patients, specifically using an intensive care database containing over 1000 patients with severe limb crush injuries or major blood vessel ruptures. The screening criteria were: within 2 hours of admission, the tissue oxygen saturation monitoring data at the injured site showed a rapid and sustained downward trend.

[0054] For each selected case, the time series of its tissue oxygen saturation monitoring data was extracted.

[0055] A sliding window with a width of 10 seconds was applied to the time series data of tissue oxygen saturation, and the slope of the linear regression of the data within the window was calculated as the instantaneous rate of change at that time point. The maximum instantaneous rate of decline in each case was identified. The maximum rate of decline values ​​for all cases were statistically analyzed, and the 95th percentile was selected as the final "maximum rate of failure baseline value". The 95th percentile was selected to exclude extreme outliers while ensuring that the baseline value covers the vast majority of critical situations. Based on the above analysis, in this embodiment, the preferred value for the maximum rate of failure baseline value was determined to be -2.5% / second, with negative values ​​indicating a decline. In practical applications, depending on different tissue sites and types of trauma, the reasonable range for this baseline value is set between -1.5% / second and -4.0% / second.

[0056] The predictive indication signal is then transmitted to the augmented reality rendering module to generate corresponding early warning information. In this embodiment, the final output of the risk prediction processing module is a predictive indication signal, specifically in the form of the rate of change of tissue blood oxygen saturation, expressed in "% / second". The trend and technical meaning of this output value are as follows: When the rate of change of tissue oxygen saturation approaches 0, it indicates that the current physical deformation and metabolic state of the tissue are in a dynamic equilibrium or stable state, predicting that the tissue's oxygen supply and demand will remain stable within a predetermined time point in the future, and the risk of metabolic failure is lower. When the rate of change of tissue oxygen saturation is negative, it indicates that the current physical deformation pattern of the tissue is having a serious negative impact on tissue perfusion and metabolism. The larger the absolute value of the negative value, the greater the predicted trend of metabolic failure, the higher the risk level, and the stronger the urgency of intervention. When the rate of change of tissue blood oxygen saturation is positive, it indicates that the current tissue perfusion status is improving and predicts that the future metabolic status of the tissue will tend to improve.

[0057] The final output of this invention, the rate of change in tissue oxygen saturation, is the result of a dynamic fusion of multiple internal influencing factors. The influence trend analysis of its core input parameters is as follows: The stability of the periodic deformation component is negatively correlated with the final predicted risk (i.e., the negative degree of change in tissue oxygen saturation); when other parameters remain constant, the periodic deformation component V periodic An increase in the stability value, indicating more regular breathing or heartbeat, will cause the first metabolic influencing factor to tend towards a stable value, thereby reducing the overall risk score and making the final output tissue oxygen saturation change rate approach 0. This design reflects the fact that patients with stable vital signs will necessarily have regular breathing and heartbeat. By quantifying this regularity and using it as an important positive criterion for judging system stability, the system can effectively avoid misinterpreting normal physiological movements (such as respiratory fluctuations) as risk signals.

[0058] Non-periodic deformation component V aperiodic The magnitude of the deformation component V is positively correlated with the final predicted risk. When other parameters remain constant, the aperiodic deformation component V... aperiodic An increase in the magnitude of stress will lead to a rise in the value of the second metabolic influencing factor through the "tissue stress accumulation" model, thereby directly increasing the comprehensive risk score and causing the final output rate of change in tissue oxygen saturation to increase negatively. Non-periodic deformation, including persistent swelling caused by internal bleeding or tissue edema, or instantaneous impact caused by external blunt injury, is the direct physical cause of tissue microcirculation disruption, decreased perfusion, and ultimately metabolic failure. This invention uses such deformation as the main driver of risk, ensuring that the model is highly sensitive to truly dangerous pathological processes.

[0059] Dynamic fusion weight W dynamic It is a nonlinear regulator determined by the second metabolic influencing factor. When the non-periodic deformation component V aperiodic When the cumulative effect is small, the dynamic fusion weight W dynamic The weighting is biased towards the first metabolic factor; as the cumulative effect increases, the dynamic fusion weight W... dynamic It then rapidly and non-linearly shifts its focus to secondary metabolic factors. This design creates a "state switch" effect. During stable periods, the system primarily focuses on the stability of physiological rhythms, exhibiting high specificity (low false alarms); once it detects the accumulation of non-periodic deformation, the system immediately shifts its decision-making focus to risk assessment, demonstrating highly sensitive early warning.

[0060] This dynamic adjustment mechanism characterizes the attention allocation mechanism of experienced clinicians, enabling the model of this invention to intelligently and smoothly switch between "routine monitoring" and "crisis response" modes, thereby achieving both low false alarm rate and high early warning timeliness in complex scenarios. To decouple the core algorithm of this invention from specific application strategies and to ensure the configurability and ease of debugging of the technical solution, all configurable operating parameters are predefined and stored in a structured external data carrier in the specific implementation path of this invention. Preferably, this is a local data file. In a preferred embodiment, this data file is a spreadsheet file, including Microsoft Excel format.

[0061] Furthermore, the core technical feature of this embodiment lies in providing an augmented reality rendering module. This module not only generates and overlays dynamic augmented reality early warning information in high-risk areas based on predictive indicator signals, but also solves the technical problem that existing static visual mapping schemes cannot reveal the "urgency" and "root cause" of risks by introducing in-depth analysis of the dynamic characteristics and inherent causal structure of the signal's temporal dimension. This invention, by constructing a hierarchical, state-adaptive visual encoding mechanism, elevates the display of a single risk level to a multi-dimensional information visualization containing "risk level," "deterioration trend," and "main triggers," thereby improving the efficiency and accuracy of remote expert decision-making.

[0062] Further explanation: Dynamic augmented reality early warning information includes: displaying pulsating halos with different colors or flashing frequencies on high-risk areas based on the strength of predictive indication signals.

[0063] Further explanation: The augmented reality rendering module is further configured as follows: Acquire historical data of the predictive indicator signal within a preset time window, and calculate the signal time derivative, which characterizes the trend of change of the predictive indicator signal, based on the historical data. Furthermore, based on the intensity and time derivative of the predictive indication signal, the visual characteristics of the pulsating halo are determined together, so that the predictive indication signal with a higher negative time derivative can generate a more visually urgent warning message.

[0064] Further explanation: The augmented reality rendering module is further configured as follows: Obtain the first and second metabolic influence factors that contribute to predictive indicator signals; Calculate the causal composition ratio that characterizes the contribution of the second metabolic factor; In addition, the dynamic augmented reality early warning information also includes causal indicators superimposed on the pulsating halo. The shape of the causal indicators is selected according to the causal composition ratio to visually distinguish the risk sources dominated by periodic deformation instability or non-periodic deformation accumulation.

[0065] Further explanation: The augmented reality rendering module is further configured as follows: Obtain display capability parameters that characterize the display quality of the extended reality view; In addition, when the display capability parameters are lower than a preset quality threshold, the dynamic presentation of the augmented reality warning information is adaptively adjusted, including replacing high-frequency flicker with enhanced color saturation.

[0066] The innovation of this embodiment stems from questioning the core transformation unit of "mapping the predicted signal into visual information": Traditional mapping is instantaneous and ignores the evolution of risk. This invention captures the "acceleration" of risk by introducing the signal time derivative, thereby distinguishing between two distinct clinical scenarios: "severe but stable" and "moderate but rapidly deteriorating," thus solving the problem of "loss of temporal information" in static mapping.

[0067] Traditional mapping is singular and fails to reveal the intrinsic composition of risk. This invention constructs a causal composition ratio by analyzing the signal sources (first and second metabolic influencing factors) and designs a hierarchical visual structure of "pulsating halo + causal indicator". This structure decomposes abstract risk into two levels: "appearance (how severe)" and "root cause (why)," solving the problem of "fuzzy causal information" caused by single indicators.

[0068] In this embodiment, the key parameters processed by the augmented reality rendering module are defined as follows: The predictive indicator signal, which serves as the core input of this augmented reality rendering module, is a data field that corresponds one-to-one with the vertices of the three-dimensional geometric model surface and characterizes the predicted rate of change in future tissue blood oxygen saturation.

[0069] The time derivative of the signal, denoted as d 2 / dt 2 StO2 This parameter characterizes the rate of change of the "predictive indicator signal," i.e., the "acceleration" of risk. Larger negative values ​​indicate that the risk is rapidly deteriorating, while values ​​approaching zero indicate that the risk state is stabilizing. Specifically, the determination method is as follows: the augmented reality rendering module maintains a time-series buffer storing historical data of the "predictive indicator signal" for the most recent K=5 time steps. To determine the signal's time derivative at the current moment, the following steps are performed: a weighted linear regression algorithm is applied to the time-series data, where the time step is the independent variable, the historical predictive indicator signal value is the dependent variable, and data points closer to the current moment are assigned higher weights; secondly, the slope of the linear regression fitted line is calculated; this slope is used as the signal's time derivative at the current moment. The calculation of this parameter is based on the time-series data of the predictive indicator signal. This input signal is periodically output by the upstream "risk prediction processing module" at a fixed time step. This signal stream contains valuable dynamic trend information, but is also accompanied by high-frequency computational noise; in this embodiment, the fixed time step is set to 100 milliseconds. The input time series data is preprocessed. The core step involves passing the K most recent data points stored in the time series buffer through a one-dimensional Gaussian smoothing filter. The standard deviation (Sigma) of this filter is set as a configurable parameter to effectively filter out noise with periods less than three time steps, while preserving key decreasing or increasing trends.

[0070] After smoothing, a weighted linear regression model is used to calculate the instantaneous derivative of the signal. More recent data points are more indicative of the current trend. The detailed calculation steps are as follows: 2.1) Obtain the K most recent data points from the time series buffer after Gaussian smoothing. Each data point contains a timestamp and a signal value.

[0071] 2.2) Calculate the weights for each of the K data points according to a preset exponential decay function. The calculation logic is as follows: using the preset time decay constant as the base, and the difference between the current timestamp and the data point's timestamp as the exponent, perform a power operation to obtain the weight of each data point. This design ensures that the newest data point has the highest weight.

[0072] 2.3) Using timestamps as independent variables and signal values ​​as dependent variables, perform weighted least squares linear regression.

[0073] 2.4) The slope of the straight line obtained by the weighted linear regression fitting is used as the time derivative of the final output signal. The unit of this slope is "(% / second) / second", which quantifies the acceleration of risk change.

[0074] It should be noted that processing is performed using a one-dimensional Gaussian smoothing filter. The standard deviation (Sigma) controls the degree of smoothing; a larger Sigma value produces a smoother but potentially delayed signal, while a smaller value results in a more sensitive response but is more susceptible to noise. In a preferred embodiment of the invention, the Sigma value is set to 1.5 time steps. Within a reasonable implementation range, this value is adjusted between 1.0 and 2.5 time steps. This range is determined by conducting parameter scanning experiments on different Sigma values ​​using a synthetic dataset containing typical physiological noise and simulated risk events (including sudden drops), and selecting a numerical range that achieves the best balance between signal-to-noise ratio and signal response delay.

[0075] The time decay constant is a value between 0 and 1 that determines the rate at which the weights of historical data decay. The closer the value is to 1, the longer the "memory" of historical data is retained. In this preferred embodiment, the time decay constant is set to 0.85. This value is determined to ensure that, with a buffer size of K=5, the weight of the oldest data point decays to less than 20% of the weight of the newest data point, thereby ensuring regression stability while focusing on reflecting the latest signal change trend.

[0076] The causal ratio, denoted as R causalThis parameter is a normalized value in the range [0,1], used to quantitatively describe the contribution of non-periodic deformation (including trauma and swelling) to the current predicted risk. A causal composition ratio value close to 1 indicates that the risk is almost entirely caused by non-periodic factors, while a value close to 0 indicates that the risk mainly stems from the instability of periodic physiological rhythms; the specific determination method is as follows: The first and second metabolic influencing factors are obtained from the risk prediction and processing module. The calculation logic is as follows: obtain the absolute value of the second metabolic influencing factor; calculate the sum of the absolute values ​​of the first and second metabolic influencing factors; divide the absolute value of the second metabolic influencing factor by the sum of the absolute values ​​obtained in the second step; the quotient is the causal composition ratio R. causal To prevent the extreme case where the denominator is zero, when the sum of absolute values ​​is less than the preset constant 1e-6, the final causal composition ratio output value is set to 0.5.

[0077] The visual urgency factor, denoted as F urgency This is an internal composite index designed to integrate the "current severity" and "future deterioration trend" of risk, generating a driving indicator that more accurately reflects clinical urgency than a single risk value. The calculation logic of this factor can be summarized as follows: its core idea originates from the proportional-differential (PD) control concept in cybernetics. The calculation steps for "jointly determining the visual characteristics of pulsating halos" are as follows: The predictive indicator signal and the signal time derivative are mapped to a unified [0,1] interval through their respective normalization functions to obtain normalized intensity and normalized trend values; secondly, the "trend weight coefficient W" is read from the configuration file. trend ; Calculate the normalized trend value and the trend weight coefficient W trend The product of the values ​​is then added to the normalized intensity value to obtain the final visual urgency factor F. urgency .

[0078] The specific determination method is as follows: In the proportional-derivative (PD) controller concept, the system output depends not only on the current error (proportional term) but also on the rate of change of the error (derivative term), thereby achieving a faster and more stable response. This invention applies this concept to the encoding of visual information, treating the current intensity of the predictive indicator signal as the "proportional term" and its signal time derivative as the "derivative term," thereby constructing a comprehensive indicator that can predictively reflect the clinical urgency.

[0079] To achieve dimensional consistency, the two core inputs are normalized.

[0080] Intensity normalization: The absolute value of the predictive indication signal is obtained and mapped to the [0,1] interval using a predefined nonlinear mapping function to obtain the "normalized intensity value". The inflection point parameter of this mapping function is stored in a configuration file to define what constitutes "low", "medium", and "high" risk. In this embodiment, the nonlinear mapping function is a piecewise linear function; Trend normalization: The time derivative of the signal is obtained and mapped to the interval [-1, 1] using an independent nonlinear mapping function to obtain the "normalized trend value". Negative values ​​represent a deteriorating trend, while positive values ​​represent an improving trend.

[0081] The following steps are performed to determine the "normalized intensity value": First, the absolute value of the predictive indication signal is obtained as the input intensity value. Second, two intensity thresholds are read from the configuration file: a low-risk intensity threshold Th... lowS and high risk intensity threshold Th highS In this embodiment, the preferred values ​​are 0.5 and 2.5, respectively. Then, a conditional judgment is performed: If the input intensity value is less than or equal to the low-risk intensity threshold, the output "normalized intensity value" is 0; if the input intensity value is greater than or equal to the high-risk intensity threshold, the output "normalized intensity value" is 1; if the input intensity value is between the two, the module performs linear interpolation calculation, the logic of which is: subtract the low-risk intensity threshold from the input intensity value to obtain the difference, and then divide the difference by the difference between the high-risk intensity threshold and the low-risk intensity threshold. The quotient is the final "normalized intensity value".

[0082] For trend normalization, the following steps are performed to determine the "normalized trend value": First, obtain the raw value of the signal's time derivative as the input trend value. Second, read two trend thresholds from the configuration file: the deteriorating trend threshold Th. negT And the improvement trend threshold Th posT In this embodiment, the preferred values ​​are -0.5 and 0.5, respectively. Then, a conditional judgment is performed: If the input trend value is less than or equal to the deterioration trend threshold, the output "normalized trend value" is -1; if the input trend value is greater than or equal to the improvement trend threshold, the output "normalized trend value" is 1; if the input trend value is between the two, the module directly divides the input trend value by the improvement trend threshold, and the quotient is the final "normalized trend value".

[0083] Read the "trend weight coefficient W" from the configuration file. trend Multiplying the "normalized trend value" by the "trend weight coefficient" yields the "weighted trend component".

[0084] Adding the "normalized intensity value" to the "weighted trend component" yields a temporary fusion value.

[0085] To ensure the final output value remains stable within the [0,1] range, the obtained temporary fusion value is saturated. That is, if the value is greater than 1, output 1; if the value is less than 0, output 0; otherwise, output the value itself. This final result is the visual urgency factor F. urgency .

[0086] Display capability parameter, denoted as Q display This represents a normalized value within the range [0,1], characterizing the real-time display quality of the extended reality view on the remote expert terminal. A higher value indicates better display quality. Through a specifically defined fuzzy logic reasoning mechanism, the three preprocessed indicators are merged into a single display capability parameter, Q. display The specific structure of this fuzzy logic reasoning mechanism is as follows: Set the rendering frame rate (FPS): Define two fuzzy sets, "poor" and "superior". "Poor" is a trapezoidal membership function, which is 1 in the interval [0,25] and linearly decreases to 0 in the interval [25,45]. "Superior" is a trapezoidal membership function, which linearly increases to 1 in the interval [45,60] and is 1 in the interval [60,∞).

[0087] Set Round-Trip Time (RTT): Define two fuzzy sets, "High" and "Low". "High" is a trapezoidal membership function, with a value of 1 in the [200,∞) millisecond interval and linearly increasing to 1 in the [100,200] millisecond interval. "Low" is a trapezoidal membership function, with a value of 1 in the [0,50] millisecond interval and linearly decreasing to 0 in the [50,100] millisecond interval.

[0088] Jitter setting: Define two fuzzy sets, "large" and "small". "Large" is a trapezoidal membership function, with a value of 1 in the [50,∞) millisecond interval and a linear increase to 1 in the [20,50] millisecond interval. "Small" is a trapezoidal membership function, with a value of 1 in the [0,10] millisecond interval and a linear decrease to 0 in the [10,20] millisecond interval.

[0089] Define a fuzzy rule base: The output variable's display capability is also defined using two fuzzy sets, "poor" and "excellent". The rule base contains the following three core rules: Rule 1: IF (Poor FPSis) OR (High RTTis) THEN (Display capability is poor); Rule 2: IF (High Jitter) THEN (Display capability is poor); Rule 3: IF (Excellent FPSis) AND (Low RTTis) AND (Small Jitter) THEN (Display capability is excellent); In the above, "IF" means "if"; "OR" means "or"; "THEN" means "then"; "is" means "is"; and "AND" means "and".

[0090] The centroid-method is used for defuzzification. The fuzzy set of the output variable, display capability, is defined as follows: "Poor" is a triangle in the range [0, 0.5] with a vertex at 0.25; "Excellent" is a triangle in the range [0.5, 1.0] with a vertex at 0.75. The activation intensity is calculated according to these rules, and the output fuzzy set is synthesized. Finally, its centroid is calculated to obtain a precise value in the interval [0, 1]. This value is the display capability parameter Q. display .

[0091] To address the limitations of static threshold mapping in existing technologies, namely its inability to effectively convey the dynamic evolution information and inherent causal relationships of risks, this invention designs the following hierarchical state adaptive visual coding engine.

[0092] Traditional methods that map risk values ​​to specific colors based on fixed thresholds can identify high-risk areas, but their output information is static and singular. They cannot distinguish between a chronically ischemic area that remains stable at -1.5% / second and an acutely hemorrhagic area that deteriorates from -0.5% / second to -1.5% / second within seconds, even though the latter has a much higher clinical urgency. Furthermore, they cannot inform experts whether the risk stems from respiratory distress or internal swelling.

[0093] The encoding engine of this invention constructs a two-layer visual information structure and enables it to adaptively adjust according to the external environment.

[0094] The first layer is the basic risk layer: used for generating the pulsating halo. The goal of this layer is to visually display the overall urgency of the risk. Its working mechanism is as follows: The visual urgency factor F obtained using the aforementioned calculation urgency As input to the mapping table.

[0095] Load a "visual mapping table" from an external configuration file. This table defines the visual urgency factor F. urgency Different numerical ranges of the light halo and their corresponding halo colors and flashing frequencies; in this embodiment, the halo color is defined by RGBA values; the flashing frequency is defined by Hz. When the visual urgency factor F urgency When it is in the range [0.5, 0.7), it is mapped to yellow and flashes at 1Hz; when it is in the range [0.7, 0.9), it is mapped to red and flashes at 3Hz; when it is greater than or equal to 0.9, it is mapped to dark red and flashes at a high frequency of 5Hz.

[0096] The second layer is the causal insight layer: causal indicator generation. The goal of this layer is to reveal the dominant causes of risk. Its working mechanism is as follows: Using the causal composition ratio R causalAs a basis for decision-making, an internally preset "indicator library" contains various forms of 3D icons or textures, each corresponding to a source of risk. A dynamic texture resembling a wave represents "dominant periodic instability," while an inward-contracting arrow icon represents "dominant non-periodic trauma."

[0097] According to the causal composition ratio R causal The value of R is used to select the corresponding indicator through a logic gating unit. causal If R < 0.3, select "Wave Texture"; if R causal If the value is greater than 0.7, select "Shrink Arrow"; otherwise, the indicator will not be displayed. The selected indicator will be rendered and overlaid on the center area of ​​the first layer of pulsating halo.

[0098] Continuous monitoring display capability parameter Q display Define a "quality threshold Q" in the configuration file. threshold "and a set of downgrade strategies".

[0099] IF display capability parameter Q display <quality threshold Q threshold The THEN engine activates a degradation strategy. The strategy stipulates that all flicker frequencies greater than 2Hz are forcibly set to 0Hz (i.e., become constantly lit), while simultaneously increasing the saturation channel value of their corresponding colors by 50%. This ensures that warning information remains clear and undistorted even at low frame rates or with poor network conditions, avoiding misinterpretation caused by high-frequency flickering turning into random jumps.

[0100] This mechanism shortens the expert's cognitive process from "observation of data - reasoning and judgment" to "direct insight." Experts can not only see "where the danger is," but also understand "the degree of urgency represented by the halo driven by the visual urgency factor" and "the cause of danger represented by the causal indicator," reducing cognitive load and providing unprecedented information support for making correct decisions within the golden rescue time. At the same time, the display adaptive mechanism ensures the robustness of information transmission under non-ideal conditions.

[0101] The complete calculation process of the augmented reality rendering module is as follows: 3.1) Receives in parallel a predictive indication signal data field, a first metabolic influence factor, and a second metabolic influence factor from the risk prediction processing module; and receives the display capability parameter Q from the network communication module or the head-mounted display device. display 3.2) For each vertex in the received predictive indication signal data field, calculate the corresponding signal time derivative and causality ratio R. causal and visual urgency factor F urgency 3.3) Traverse all vertices, for the visual urgency factor F urgencyFor vertices in high-risk areas that exceed a preset display threshold, a two-layer encoding mechanism is implemented: based on the visual urgency factor F. urgency The value is used to query the "visual mapping table" to determine the color and original flicker frequency of the pulsating halo.

[0102] Based on the causal composition ratio R causal The value, through the logic gating unit, determines whether to display the causal indicator and what type of indicator to select.

[0103] It should be noted that the rules for determining the preset display threshold are as follows: an expert group composed of multiple authoritative experts in the field is organized; a series of typical, anonymized clinical data streams (including predictive indicator signals and their time derivative change curves) are presented to the expert group, and they are asked to independently determine at which time point or within which numerical range they would consider "the situation to have become critical and requires close attention".

[0104] Collect all expert-marked "critical points". Calculate the corresponding visual urgency factor F at these points. urgency Value. For all experts, the visual urgency factor F... urgency Perform statistical analysis on the values ​​to calculate their mean, median, or confidence interval.

[0105] The statistically derived consensus value (median or 75th percentile) is used as the initial preset display threshold. This initial preset display threshold is applied to a validation dataset and submitted again to the expert panel for review. Based on their feedback, the preset display threshold is iteratively fine-tuned until it reaches the general consensus of the expert panel.

[0106] In this embodiment, the preset display threshold is set to 0.7; an example of the content to be judged is as follows: Model vertex A1: Its predictive indicator signal has high intensity but stable trend, and the calculated visual urgency factor F urgency The value is 0.65. Judgment: 0.65 < 0.7, vertex A1 does not belong to the high-risk area.

[0107] Model vertex B1: Its predictive indication signal is of moderate strength but is deteriorating rapidly, with a calculated visual urgency factor F. urgency The value is 0.82. Judgment: 0.82 > 0.7, vertex B1 is marked as a high-risk vertex.

[0108] 3.4) Module check display capability parameter Q display If it is below the quality threshold Q threshold Then, according to the degradation strategy, the original flashing frequency and color of the determined halo are adjusted.

[0109] 3.5) The visual parameters (position, color, final flashing frequency, indicator type, etc.) of each high-risk area vertex are encapsulated into a rendering instruction set and sent to the graphics rendering pipeline of the extended reality view. Finally, dynamic and information-rich augmented reality early warning information is accurately superimposed on the three-dimensional geometric model in the view of remote experts.

[0110] The following detailed embodiments illustrate the above content: The core algorithm of this embodiment outputs two key pieces of information: a continuous visual urgency factor F. urgency And a discrete causal indicator.

[0111] Visual urgency factor F urgency The value range of this parameter is strictly limited to the interval [0,1]. A higher visual urgency factor indicates a more severe overall assessment of the current high-risk area. When the visual urgency factor is closer to 1, it indicates that the current high-risk area is more severe and / or is experiencing a more rapid negative deterioration; this corresponds to a clinical condition that requires immediate, highest-priority attention from remote experts.

[0112] When the visual urgency factor is closer to 0, it indicates that the metabolic state of the current area is stable and the risk is lower.

[0113] Causal indicator: This is a visual element with a variable form. Changes in its form represent the tendency of the underlying physical cause of the risk to fall between two different categories: "periodic physiological instability" and "non-periodic tissue damage".

[0114] The intensity of the predictive indicator signal: When other parameters remain constant, an increase in the absolute value of the predictive indicator signal intensity will lead to a monotonically increasing visual urgency factor. This is because signal intensity directly reflects the rate of metabolic failure and is a fundamental proportional term in risk assessment. This positive correlation design accurately maps the basic physical reality of risk severity.

[0115] Signal time derivative: When other parameters remain constant, a negative increase in the signal time derivative (i.e., a smaller value) will lead to a monotonically increasing visual urgency factor. A large negative value indicates that the risk is rapidly deteriorating, and even if the current risk intensity is not high, it foreshadows a high clinical urgency derivative. This design addresses the fundamental deficiency of existing technologies that cannot distinguish between "slowly developing severe illness" and "suddenly deteriorating acute illness" based solely on current value snapshots, and is the core of this invention's "predictive" early warning system.

[0116] Causal composition ratio R causal The numerical value of this parameter does not affect the calculation of the visual urgency factor. It independently and orthogonally controls the morphological selection of the causal indicator.

[0117] When the causal composition ratio R causal When the value approaches 1, the indicator will appear as a representation of "non-periodic damage". This design decouples the two information dimensions of risk "severity / urgency" and "root cause", allowing experts to simultaneously obtain two key decision-making information within a single view without interference.

[0118] Display capability parameter Q display This parameter does not affect the visual urgency factor F, which is characterized by the core risk factor. urgency It is not calculated directly, but rather serves as a "post-processing" regulator in the final visual presentation stage. When the display capability parameter Q... display When the display quality is reduced, a preset visual degradation strategy will be triggered. This is a negative correlation adjustment, meaning that the worse the display capability, the lower the complexity of visual elements (including high-frequency flicker), in order to ensure the effective transmission of information in low-quality channels.

[0119] To quantitatively verify the beneficial effects of the technical solution of this invention, a comparative experiment covering four typical clinical scenarios was designed. The experiment generated high-fidelity data by driving a fully validated biophysical model, which was then processed by the augmented reality rendering module of this invention. The experiment aimed to demonstrate the significant advancements of this invention in terms of information expression accuracy and urgency differentiation compared to existing technologies that rely solely on risk intensity (predictive indicator signals), as detailed in Table 1 below: Table 1: Comparison of visual warning effects of the present invention and existing technologies in different clinical scenarios Parameter name Scenario one: chronic ischemia Scenario two: acute hemorrhage Scenario three: recovery phase Scenario four: weak network acute hemorrhage Predictive indicator signal (% / sec) -2 -1.5 -2 -1.5 Signal time derivative (% / sec 2 )]]> -0.1 -1 0.8 -1 Causal composition ratio 0.2 0.9 0.9 0.9 Display capability parameters 0.9 0.9 0.9 0.4 Prior art visual warning Red high frequency flicker Orange medium frequency flicker Red high frequency flicker Orange medium frequency flicker Visual urgency factor 0.72 0.91 0.35 0.91 Halo color / frequency Orange / medium frequency Deep red / very high frequency Yellow / low frequency Deep red / constant high intensity Causal indicator Wavy texture Contracting arrow Contracting arrow Contracting arrow Final visual presentation (invention) Orange halo + wavy texture Deep red halo + contracting arrow Yellow fading halo + contracting arrow Deep red high saturation constant halo + contracting arrow Precise differentiation of urgency (Scenario 1 vs. Scenario 2): Existing technologies rely solely on the intensity of predictive indicator signals, thus classifying Scenario 1 (with a higher intensity signal of -2.0) as more dangerous than Scenario 2 (with a signal of -1.5) as a "high-frequency red flashing." However, this is severely inconsistent with clinical reality. Scenario 2 involves sudden massive hemorrhage, whose urgency is far greater than chronic ischemia. This invention, by introducing the signal time derivative, correctly identifies the significant deterioration trend (-1.0) in Scenario 2, calculating a visual urgency factor as high as 0.91, while Scenario 1, due to its stable state, has a visual urgency factor of only 0.72. In this crucial distinction, the visual urgency factor of this invention is 26.4% higher than that of Scenario 1; the calculation method is (0.91-0.72) / 0.72×100%, thereby generating the highest level "deep red / extremely high frequency" warning, avoiding the serious error of existing technologies misclassifying acute fatal risks as secondary risks. This demonstrates the decisive advantage of this invention in differentiating the urgency of risks. This significant quantitative difference drove the system to generate the highest level of "deep red / extremely high frequency" warning, avoiding the serious error of existing technologies misclassifying acute fatal risks as secondary risks. This demonstrates the decisive advantage of this invention in distinguishing the urgency of risks.

[0120] Accurate identification of recovery trends (Scenario 1 and Scenario 3): Scenario 1 and Scenario 3 have the exact same risk intensity (-2.0), therefore existing technologies provide identical, indistinguishable "high-frequency red flashing" warnings, which can cause continuous and unnecessary tension for experts, and even lead to "alarm fatigue." This invention, however, identifies a clear improvement trend in Scenario 3 (+0.8) through the signal time derivative, significantly reducing the visual urgency factor to 0.35, and outputting a "yellow / low-frequency" halo representing low risk. This proves that this invention can not only report "danger" but also "safety," dynamically reflecting the true evolution of the condition, providing experts with complete and accurate situational awareness, and avoiding interference from invalid information.

[0121] Robustness assurance for information transmission (Scenario 2 and Scenario 4): Scenario 4 simulates the exact same clinical risks as Scenario 2, but in a poor network environment (display capability parameter Q). display =0.4). At this point, the core risk assessment (visual urgency factor F) is... urgency While the frame rate remains unchanged at 0.91, the adaptive mechanism of this invention is triggered, automatically replacing the distorted or lost "extremely high-frequency flickering" at low frame rates with a more stable and prominent "high-saturation constant light" display. This demonstrates the high robustness of the design of this invention in complex real-world network environments, ensuring that under any circumstances, the most critical early warning information can be conveyed to experts unambiguously and reliably, safeguarding the "lifeline" of remote diagnosis and treatment.

[0122] Furthermore, in order to quantify the visual urgency factor F of this invention... urgency The output is transformed into standardized, executable clinical procedures. This invention defines the following three-level response intervals, as detailed in Table 2 below: Table 2: Practical Application Division of Output Range Response level Visual urgency factor interval Visual presentation characteristics System automated operation Suggested expert operation Level one: surveillance zone [0,0.5] No or low intensity static green halo 1. Continuous data logging. 2. No active alarms. Routine observation, no immediate intervention needed. Level two: alert zone (0.5,0.8] Yellow / orange, medium-low frequency flickering halo 1. Trigger a sound alarm (medium priority). 2. Mark as "level two alert" in the event log. 3. Automatically center the area in the view. High attention needed, assess risk evolution trends, prepare intervention plan. Level three: critical zone (0.8,1.0] Red / deep red, high frequency flickering or high saturation constant halo, with superimposed causal indicator 1. Trigger a sound alarm (highest priority, continuous). 2. Automatically place the case at the top of all pending queues. 3. Push an emergency notification to all online experts of the team. Immediate intervention! Follow the causal indicator to guide on-site personnel to perform targeted first aid measures. By employing a spatiotemporal coupled perfusion-metabolic kinetics model, a "prediction-driven" linkage control center is established. Abstract risk prediction is transformed into an intuitive and dynamic "pulsating halo," enabling experts to instantly perceive the "location" and "urgency" of risks, achieving precise delivery of decision-making information. Detection resources are automatically focused on areas predicted to be high-risk, achieving intelligent scheduling and closed-loop verification of detection resources. This linkage effect, based on core prediction and simultaneously driving optimization of both "human (expert cognition)" and "machine (equipment behavior)," elevates the entire telemedicine system from a passive information conduit to a mechanism with forward-looking insight and autonomous optimization capabilities.

[0123] The computational logic involved in this application can be constructed using algorithms such as regression analysis in machine learning, establishing a mathematical model by analyzing the inherent trends and interrelationships of the collected parameters. This process can be implemented using specialized computational tools (such as Python's Scikit-learn library or the R language environment). Throughout all calculations, to eliminate the influence of different physical dimensions and ensure that data is compared and analyzed on the same scale, the input parameters in each formula are dimensionless. The dimensionless techniques used include, but are not limited to, max-min normalization or Z-score standardization.

[0124] The algorithm of this invention is implemented as a Python script. Before executing the core logic, the program first executes a data loading module (e.g., using the widely used pandas library in Python) configured to read the aforementioned spreadsheet file and load its contents into the program's working memory (e.g., a DataFrame data structure). Subsequent algorithm steps will directly query and retrieve the required configuration parameters from this in-memory data structure.

[0125] It should be emphasized that the foregoing embodiments are merely illustrative of preferred implementations of the present invention and are not intended to limit the scope of protection of the present invention. This application also provides a computer-readable storage medium having computer program instructions stored thereon.

Claims

1. An immersive visual communication system for emergency resuscitation based on virtual reality, characterized in that: Specifically, it includes: The geometric model acquisition module is used to acquire a three-dimensional geometric model of a local area of ​​the target patient's body in real time, and generate first data characterizing the physical deformation state of the local area based on the changes of the three-dimensional geometric model within a preset time window. The tissue state acquisition module is used to acquire second data of the tissue metabolic state corresponding to the local area space in real time through near-infrared spectral imaging. The risk prediction processing module is configured to fuse the first data and the second data using a pre-built spatiotemporally coupled perfusion-metabolic kinetic model to generate a predictive indication signal characterizing the risk of metabolic failure in the local region at a future preset time point; and The augmented reality rendering module is used to generate and overlay dynamic augmented reality early warning information on the corresponding high-risk areas on the three-dimensional geometric model in the extended reality view provided to remote experts, based on the predictive indication signal.

2. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 1, characterized in that: The geometric model acquisition module includes a synchronous positioning and mapping unit, which uses data from a depth sensor and an inertial measurement unit to generate a three-dimensional geometric model in real time. Furthermore, the first data is the three-dimensional deformation velocity field of the vertices on the surface of the three-dimensional geometric model within a preset time window. The second data includes local tissue oxygen saturation data and total hemoglobin concentration change data in the local area.

3. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 2, characterized in that: The spatiotemporal coupled perfusion-metabolic kinetic model is configured to establish a temporal coupling relationship between the physical deformation represented by the first data and the tissue metabolic state represented by the second data. Based on the temporal coupling relationship, the rate of change of tissue blood oxygen saturation in the local area at a future preset time point is predicted as a predictive indication signal.

4. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 3, characterized in that: The risk prediction processing module is further configured to perform the following steps to generate the predictive indication signal: The first data is subjected to deformation mode decomposition to identify and separate at least periodic deformation components and non-periodic deformation components. Secondly, using the first sub-model, the first metabolic influencing factor is calculated based on the periodic deformation component; Using the second sub-model, the second metabolic influencing factor is calculated based on the non-periodic deformation component; The first metabolic influencing factor and the second metabolic influencing factor are dynamically weighted and fused to determine the rate of change of tissue blood oxygen saturation.

5. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 4, characterized in that: The smaller the rate of change in tissue oxygen saturation, the more dynamic the physical deformation and metabolic state of the current tissue are in equilibrium or stable. This indicates that the tissue's oxygen supply and demand will be more stable at a predetermined time point in the future, and the lower the risk of metabolic failure. When the rate of change of tissue oxygen saturation is negative, it indicates that the current physical deformation pattern of the tissue is having a negative impact on tissue perfusion and metabolism. The larger the absolute value of the negative value, the greater the predicted trend of metabolic failure, the higher the risk level, and the greater the urgency of intervention. When the rate of change of tissue blood oxygen saturation is positive, it indicates that the current tissue perfusion status is improving and predicts that the future metabolic status of the tissue will tend to improve.

6. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 5, characterized in that: Dynamic augmented reality warning information includes: displaying pulsating halos with different colors or flashing frequencies on high-risk areas based on the strength of the predictive indication signal.

7. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 6, characterized in that: The augmented reality rendering module is further configured as follows: The historical data of the predictive indication signal within a preset time window are obtained, and the signal time derivative characterizing the trend of change of the predictive indication signal is calculated based on the historical data. Furthermore, based on the intensity of the predictive indication signal and the time derivative of the signal, the visual characteristics of the pulsating halo are jointly determined so that the predictive indication signal with a higher negative time derivative can generate a more visually urgent warning message.

8. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 7, characterized in that: The augmented reality rendering module is further configured as follows: Obtain the first metabolic influence factor and the second metabolic influence factor that contribute to the predictive indication signal; Calculate the causal composition ratio, which characterizes the contribution of the second metabolic factor; Furthermore, the dynamic augmented reality warning information also includes causal indicators superimposed on the pulsating halo, the shape of which is selected according to the causal composition ratio to visually distinguish risk sources dominated by periodic deformation instability or non-periodic deformation accumulation.

9. The immersive visual communication system for emergency resuscitation based on virtual reality according to claim 8, characterized in that: The augmented reality rendering module is further configured as follows: Obtain display capability parameters that characterize the display quality of the extended reality view; Furthermore, when the display capability parameter is lower than a preset quality threshold, the dynamic augmented reality warning information is adaptively adjusted, including replacing high-frequency flicker with enhanced color saturation.

Images