A pressure ulcer prevention early warning and closed-loop intervention method and device based on multi-physical field decoupling, computer equipment and storage medium

Abstract

The application provides a pressure ulcer prevention early warning and closed-loop intervention method and device based on multi-physical field decoupling, computer equipment and a storage medium, relates to the technical field of intelligent medical monitoring data processing, and aims at the problems of existing sensor cross-coupling interference and static early warning rigidity. The application extracts high-dimensional synchronous features through bottom layer flexible isolation and timestamp dynamic registration; extracts individualized steady-state baseline, introduces nonlinear fitting to strip the pseudo-temperature drift caused by humidity sudden change, reconstructs the real temperature and pressure characteristics for dynamic integral to output the risk index; finally, the intelligent matching clinical strategy is issued, the intervention action is adaptively identified based on the pressure gradient, and the prediction model and the intervention knowledge base are driven for long-term evolution. The application effectively eliminates the system false alarm caused by micro-environment drift, realizes the medical full closed loop from accurate decoupling monitoring to individualized dynamic intervention.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical monitoring data processing technology, and in particular to a method, device, computer equipment, and storage medium for pressure ulcer prevention early warning and closed-loop intervention based on multi-physics field decoupling. Background Technology

[0002] Pressure ulcers (pressure injuries) are a major clinical challenge for long-term bedridden and disabled / semi-disabled patients. Their occurrence is closely related to prolonged pressure on local tissues, abnormally elevated temperatures, and excessively high humidity in the skin's microenvironment. Therefore, achieving continuous real-time monitoring of these three parameters is a crucial development direction for current intelligent pressure ulcer prevention and care equipment.

[0003] Current technologies typically employ radio frequency photonic sensors such as fiber Bragg gratings (FBGs) to monitor pressure and temperature, while simultaneously using flexible capacitive sensors to monitor humidity. However, in the complex medical microenvironment where patients frequently turn over, experience pressure, and sweat (drastic temperature and humidity changes), existing technologies generally fall into a systemic technical pitfall: separating the "lower-level physical sensing acquisition" from the "upper-level algorithm logic judgment," failing to resolve the vicious cycle of lower-level data distortion and upper-level warning failure caused by the cross-coupling of multiple physical fields. Specifically, existing solutions simply and densely deploy radio frequency photonic sensor networks (optical signals) and flexible capacitive sensor networks (electrical signals) of the same size and at the same locations. Under dynamic deformation, the stiffness of the upper capacitor array directly compresses the lower optical fiber, causing micro-bending loss in the FBG fiber and thus triggering "pseudo-drift" in the optical signal. The microenvironment of the pressure ulcer pad surface is dynamically coupled. When local sweating causes drastic changes in fabric humidity, the physical changes in its thermal conductivity and specific heat capacity directly lead to temperature response distortion (i.e., temperature drift) in the FBG grating immediately beneath it. In addition, optical and electrical signals belong to different heterogeneous clock domains, and current technology has not achieved microsecond-level alignment at the underlying layer, resulting in inherent spatiotemporal misalignment in the acquired multimodal data. This physical coupling interference and spatiotemporal misalignment mean that the raw data output by the system itself already contains significant environmental drift errors.

[0004] Because existing technologies ignore the cross-sensitivity and baseline drift of the underlying sensors in complex microenvironments, their upper-level warning algorithms incorrectly and directly apply clinical medical standards, such as setting absolute static thresholds like pressure ≥32 mmHg or local humidity ≥65% RH as judgment conditions. This "if-else" logic, which directly codifies medical common sense, not only fails to identify and filter out false data caused by drift in the underlying hardware environment, but also substitutes these "distorted data" as real physiological indicators into the alarm model. This severe disconnect between the algorithm and the underlying physical characteristics results in extremely high false alarm and false negative rates for existing products in complex clinical applications. Summary of the Invention

[0005] To address the shortcomings of the existing technologies, the technical problem this invention aims to solve is to provide a method, device, computer equipment, and storage medium for pressure ulcer prevention early warning and closed-loop intervention based on multi-physics field decoupling. This invention incorporates a heterogeneous signal timestamp alignment compensation mechanism at the data generation end, achieving precise spatiotemporal alignment of optical / electrical signals at high frequencies. Furthermore, it utilizes synchronously aligned real-time humidity electrical signals as environmental variable constraints, combined with the temperature drift coefficient generated by the fabric heat dissipation rate change under specific humidity conditions on the FBG, to perform real-time reverse calibration in the demodulation algorithm. To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: The present invention provides a method for early warning and closed-loop intervention of pressure ulcer prevention based on multi-physics field decoupling, comprising the following steps: S1. The radio frequency photonic sensing subnet and the flexible capacitive sensing subnet are physically decoupled via a flexible isolation layer, and the high-frequency optical raw feature set in the heterogeneous clock domain is acquired in parallel. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable ; S2. When the monitored microenvironment is determined to have reached a steady state, extract the personalized initial physical baseline matrix. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

[0006] The preferred solution also includes step S3, as follows: S3, Based on the comprehensive pressure ulcer risk index Match the reconstructed feature variables with the optimal clinical intervention strategy matrix. It outputs signals from each level. Integrated intervention and scheduling data packets Tracking real and effective stress characteristic variables Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. And trigger the re-extraction of the personalized initial physical baseline matrix. The steps.

[0007] The preferred solution also includes step S4, as follows: S4. Based on the timestamp of the intervention action Extracting long-period state sequence sets Quantifying individual tolerance degradation factors According to individual tolerance degradation factors Output updated temperature acceleration coefficient With the updated humidity acceleration factor This involves overwriting the system's underlying parameters and updating the clinical intervention case database based on feedback on intervention effectiveness. .

[0008] In a preferred embodiment, step S1 involves physically decoupling the radio frequency photonic sensing subnet from the flexible capacitive sensing subnet via a flexible isolation layer, including the following specific steps: Based on the pre-set relative permittivity within the flexible isolation layer Electromagnetic shielding is performed to reduce the parasitic capacitance deviation between the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet. ; Based on the pre-set mechanical buffer modulus within the flexible isolation layer Perform mechanical low-pass filtering to absorb horizontal shear forces generated by dynamic deformation and block optical micro-bending loss; Based on the established global absolute time reference source Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. ,include: Discrete measurements from the heterogeneous clock domain are extracted, and the original high-frequency optical feature set is derived based on an interpolation-extrapolation mapping model. With low-frequency electrical primitive feature set Align timestamp with target Perform virtual feature registration calculation and output virtual feature reconstruction values; Dimensionality reduction and assembly are performed on the virtual feature reconstruction values ​​to generate a high-dimensional synchronous feature matrix. .

[0009] In the preferred embodiment, in step S2, a personalized initial physical baseline matrix is ​​extracted. And reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Output pressure ulcer comprehensive risk index And generate signals at each level. The specific steps include the following: Obtaining the high-dimensional synchronization feature matrix In the synchronization time window The data sequence within the dataset is used to calculate environmental fluctuation characteristic values ​​based on an environmental fluctuation convergence model. ; When environmental fluctuation characteristics Less than the physical convergence threshold At that time, extract the personalized initial physical baseline matrix. To establish an absolute zero reference for error inversion; Based on a nonlinear thermal conductivity fitting model, the real-time humidity deviation scalar Perform mapping processing to extract pseudo-temperature drift feature scalars ; Based on cross-sensitive reverse compensation logic, utilizing pseudo-temperature drift feature scalar With preset cross sensitivity coefficient For the original temperature variable Compared with the original pressure variable Perform reverse error subtraction to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables To eliminate physical cross-interference caused by drastic changes in the microenvironment; Based on the dynamic damage integral model, the true temperature characteristic variables are... Scalar of deviation from real-time humidity Transformed into microenvironment penalty weight factor And combined with real and effective stress characteristic variables Perform weighted time-series integration to output the pressure ulcer comprehensive risk index. ; Based on the comprehensive risk index of pressure ulcers With dynamic early warning threshold vector Based on the critical alignment results, signals at each level are generated. This will trigger auxiliary intervention actions at the corresponding risk level.

[0010] In the preferred embodiment, in step S3, the optimal clinical intervention strategy matrix is ​​matched. And track the real and effective stress characteristic variables. Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. The specific steps include the following: Pressure ulcer comprehensive risk index With reconstructing the effective feature matrix Dimensionality reduction and stitching are performed to construct a real-time risk state vector representing the local tissue deterioration state. ; Based on the Mahalanobis distance mapping model, the real-time risk state vector With standard clinical intervention case database The similarity of each historical feature vector is calculated, and the scalar of the intervention plan matching degree is output. To eliminate the interference of linear correlation between multidimensional physical variables; Based on intervention protocol matching scalar The sorting results are extracted and processed to output the optimal clinical intervention strategy matrix. ; Based on time-series differential logic, the true effective pressure characteristic variables are analyzed. Gradient calculations are performed to extract the rate of change of local pressure temporal gradient between adjacent synchronization cycles. ; When the local pressure time gradient change rate Meets the preset threshold for intervention action recognition When conditions are met, perform a log marking operation and output the timestamp of the intervention action. To identify physical intervention events at the software layer; Output intervention action timestamp Subsequently, based on the comprehensive pressure ulcer risk index The rate of change before and after the intervention was calculated to generate a scalar for assessing the effectiveness of the intervention. ; When the scale for evaluating the effectiveness of intervention Meeting the preset effective intervention threshold conditions and real-time humidity deviation scalar When the safe fallback condition is met, execute the underlying reset control flow to trigger the re-extraction of the personalized initial physical baseline matrix. The steps.

[0011] In the preferred embodiment, in step S4, the individual tolerance degradation factor is quantified. And update the database of evolving clinical intervention cases. The specific steps include the following: Based on the time stamp of the intervention action Perform segmentation and concatenation processing on the time series data to generate a long-period state sequence set. ; Based on long-period state sequence sets The cumulative slope of the risk index during the resting period is extracted, and a degradation quantification model is used to map the cumulative slope of the risk index to output the individual tolerance degradation factor. To quantitatively characterize the decline state of individual skin microcirculation; Based on the logarithmic optimization model, using individual tolerance degradation factors The system's preset base acceleration coefficient is nonlinearly amplified to output the updated temperature acceleration coefficient. With the updated humidity acceleration factor To adaptively tighten the warning tolerance of the system; Based on a reinforcement learning reward model, and utilizing a scalar for evaluating intervention effectiveness. Optimal clinical intervention strategy matrix Associated policy confidence reward scalar Perform weight update processing; Scalar of reward based on policy confidence The dynamic ranking results output evolves into a clinical intervention case database. This allows for the self-iteration of the clinical intervention strategy library.

[0012] In a preferred embodiment, the present invention also provides a pressure ulcer prevention early warning and closed-loop intervention device based on multi-physics field decoupling, comprising: The heterogeneous multimodal sensing signal underlying physical acquisition and spatiotemporal synchronization fusion module is used to physically decouple the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet via a flexible isolation layer, and to acquire in parallel the high-frequency optical raw feature set in the heterogeneous clock domain. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable ; A baseline dynamic drift reverse compensation and risk calculation module based on a multiphysics coupling model is used to extract a personalized initial physical baseline matrix when the monitored microenvironment is determined to have reached a steady state. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

[0013] The intelligent matching and closed-loop verification module for clinical intervention strategies is used to verify the effectiveness of interventions based on the comprehensive pressure ulcer risk index. Match the reconstructed feature variables with the optimal clinical intervention strategy matrix. It outputs signals from each level. Integrated intervention and scheduling data packets The intelligent matching and closed-loop verification module for clinical intervention strategies is also used to track real and effective stress characteristic variables. Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. And trigger the re-extraction of the personalized initial physical baseline matrix. Steps; The module for dynamic updating of the personalized prediction model evolution and intervention knowledge base is used to update the knowledge base based on the timestamps of intervention actions. Extracting long-period state sequence sets Quantifying individual tolerance degradation factors The module for dynamic updating of the individualized prediction model evolution and intervention knowledge base is also used to analyze individual tolerance degradation factors. Output updated temperature acceleration coefficient With the updated humidity acceleration factor This involves overwriting the system's underlying parameters and updating the clinical intervention case database based on feedback on intervention effectiveness. .

[0014] In a preferred embodiment, the present invention also provides a computer device, wherein the computer device includes at least one processor coupled to at least one memory, the memory storing at least one computer program or instruction, characterized in that the computer program or instruction is loaded and executed by the processor to implement the steps of the aforementioned method for pressure ulcer prevention early warning and closed-loop intervention based on multi-physics field decoupling. In a preferred embodiment, the present invention further provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program or instructions, which, when executed by a processor, implement the steps of the pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling described above.

[0015] This invention provides a method, device, computer equipment, and storage medium for pressure ulcer prevention early warning and closed-loop intervention based on multi-physics field decoupling. Through the coordination of the above structures, compared with existing methods, it has the following beneficial effects: First, it overcomes the signal crosstalk and timing misalignment caused by the simple superposition of heterogeneous sensors, reshaping the foundation of high-fidelity data. Existing technologies, when integrating multimodal sensors, are prone to mechanical compression interference and spatiotemporal mismatch between sensors due to the lack of physical isolation and clock synchronization under the dynamic deformation of the patient's body. This invention effectively absorbs the lateral shear force generated by the patient's turning over and other movements by introducing a flexible mechanical buffer and electromagnetic isolation mechanism between the sensor networks, thus curbing the false jump phenomenon of sensors caused by mechanical compression from the physical source. At the same time, by combining a timing registration algorithm of heterogeneous clock domain at the data generation end, the inherent sampling time difference between different hardware devices is eliminated.

[0016] Secondly, it effectively avoids systemic false alarms. Existing pressure ulcer prevention devices suffer from a severe disconnect between their alarm logic and the underlying physical sensing characteristics. They are unable to recognize changes in fabric thermal conductivity caused by sudden changes in local humidity, such as patient sweating, leading to "false temperature drift" and frequent false alarms. Through a personalized baseline dynamic compensation mechanism based on underlying physical characteristics, it can keenly sense changes in the microenvironment, quantitatively calculate, and reverse-engineer the false temperature drift and pressure cross-sensitivity errors caused by drastic humidity changes. This correction mechanism successfully decouples the environmental drift errors of the underlying hardware from the patient's actual surface physiological indicators, ensuring that the final warning judgment is entirely based on the absolute physiological load, effectively suppressing false alarms caused by complex and changing clinical microenvironments.

[0017] Thirdly, recognizing the clinical challenge of underreporting or underestimating risks in standardized risk assessment models due to the continuous decline in physiological functions of long-term bedridden patients, a long-term data-driven evolutionary mechanism has been established. By continuously mining and aggregating intervention history and risk accumulation characteristics over a long timeframe, the system can accurately quantify the physical rate of skin tolerance degradation in individual patients. Based on this degradation assessment, the system can adaptively and dynamically optimize and tighten the warning tolerance for each patient. Simultaneously, by incorporating feedback on the effectiveness of actual nursing actions, the system continuously updates its clinical auxiliary intervention strategy library, allowing it to learn and evolve autonomously as patients' physical conditions gradually change. This truly achieves personalized care and long-term, precise protection throughout the entire lifespan in the field of pressure ulcer prevention. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a main view structural diagram of the process of this invention; Figure 2 This is a flowchart of the microenvironment pseudo-temperature drift inversion compensation algorithm based on nonlinear thermal conductivity in Embodiment 1 of the present invention; Figure 3 This is a simulation comparison curve of the pseudo-temperature drift removal effect in an extreme high humidity microenvironment in Embodiment 1 of the present invention; Figure 4 This is the thermodynamic deduction diagram of physical intervention verification based on abrupt changes in time-series pressure gradient in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the structure of the computer device of the present invention. Detailed Implementation

[0019] To better understand the purpose, system architecture, and functional implementation of this embodiment, the embodiments and features described herein can be combined with each other without conflict. The exemplary embodiments disclosed herein will be described below with reference to the accompanying drawings, including specific technical details disclosed to aid understanding; however, these details should be considered exemplary rather than restrictive. Therefore, those skilled in the art should understand that various improvements and adjustments can be made to the embodiments described herein without departing from the scope and core ideas of the invention. Similarly, for clarity, detailed descriptions of well-known technologies, functions, and structures are omitted in the following description.

[0020] Before delving into the specific technical steps, it's crucial to first clarify the technical challenge addressed by this invention: pressure ulcers (pressure injuries) pose a significant clinical problem for long-term bedridden and disabled / semi-disabled patients. Their occurrence is closely related to prolonged pressure on local tissues, abnormally elevated temperatures, and excessively high humidity in the skin's microenvironment. Therefore, achieving continuous real-time monitoring of these three parameters—pressure, temperature, and humidity—is a vital development direction for current intelligent pressure ulcer prevention and care devices.

[0021] Current technologies typically employ radio frequency photonic sensors such as fiber Bragg gratings (FBGs) to monitor pressure and temperature, while simultaneously using flexible capacitive sensors to monitor humidity. However, in the complex medical microenvironment where patients frequently turn over, experience pressure, and sweat (drastic temperature and humidity changes), existing technologies generally fall into a systemic technical pitfall: separating the "lower-level physical sensing acquisition" from the "upper-level algorithm logic judgment," failing to resolve the vicious cycle of lower-level data distortion and upper-level warning failure caused by the cross-coupling of multiple physical fields.

[0022] Specifically, existing solutions involve a simple, dense deployment of radio frequency photonic sensing networks (optical signals) and flexible capacitive sensing networks (electrical signals) at the same size and location. Under dynamic deformation, the rigidity of the upper capacitive array directly compresses the lower optical fiber, causing micro-bending loss in the FBG fiber, which in turn triggers "pseudo-drift" in the optical signal. Simultaneously, the microenvironment of the pressure ulcer pad surface is dynamically coupled. When local sweating causes drastic changes in fabric humidity, the physical changes in its thermal conductivity and specific heat capacity directly lead to temperature response distortion (i.e., temperature drift) in the FBG grating directly beneath it. Furthermore, optical and electrical signals belong to different heterogeneous clock domains, and existing technologies have not achieved microsecond-level alignment at the underlying layer, resulting in inherent spatiotemporal misalignment in the acquired multimodal data. This physical coupling interference and spatiotemporal misalignment mean that the raw data output by the system already contains significant environmental drift errors.

[0023] Because existing technologies ignore the cross-sensitivity and baseline drift of the underlying sensors in complex microenvironments, their upper-level warning algorithms incorrectly and directly apply clinical medical standards, such as setting absolute static thresholds like pressure ≥32 mmHg or local humidity ≥65% RH as judgment conditions. This "if-else" logic, which directly codifies medical common sense, not only fails to identify and filter out false data caused by drift in the underlying hardware environment, but also substitutes these "distorted data" as real physiological indicators into the alarm model. This severe disconnect between the algorithm and the underlying physical characteristics results in extremely high false alarm and false negative rates for existing products in complex clinical applications.

[0024] Example 1 like Figures 1-4 As shown, to completely solve the technical problems of multi-physics field cross-coupling and environmental baseline drift, this embodiment 1 proposes a physical blocking and clock synchronization mechanism to curb the source of drift, and constructs a personalized dynamic compensation algorithm based on underlying physical characteristics to achieve closed-loop de-biasing. The following are the structured details of the core operating logic of this system: S1: Low-level physical acquisition and spatiotemporal synchronous fusion of heterogeneous multimodal sensing signals The core objective of this stage is to cut off the mechanical and electromagnetic crosstalk between the optical and electrical sensor networks at the source of the physical architecture, and to eliminate the timestamp misalignment caused by differences in hardware crystal oscillators at the data generation end, so as to provide a clean and strictly frequency-aligned basic data matrix for subsequent algorithm decoupling.

[0025] S11: Construction of the Dual Physical Isolation Structure at the Bottom Layer of the Sensor Network To overcome the micro-bending loss and parasitic capacitance interference caused by dense deployment at the same location, this invention designs a sandwich-style stacked architecture inside the sensing layer of the modular care pad body. The upper layer is configured with an MXene / bamboo fiber (BCFs) porous composite membrane humidity sensing subnet, and the lower layer is configured with a radio frequency photonic sensing subnet based on fiber optic gratings (FBG).

[0026] Between the two sub-mesh layers, this invention embeds a soft silicone insulating layer with specific material properties. The thickness variation of this soft silicone insulating layer... The setting is 1mm, which contains two key physical constraint parameters: one is the relative permittivity. Secondly, the mechanical buffer modulus .

[0027] Through low relative permittivity The design incorporates a soft silicone insulating layer that significantly increases the electrode spacing between the two heterogeneous sensing networks. Based on the principle of parallel-plate capacitors, this effectively mitigates the parasitic capacitance deviation caused by the upper capacitor network on the lower fiber optic metal encapsulation structure. Reduced to microfarads ( Below the [level], absolute signal insulation and shielding are achieved. Simultaneously, the mechanical buffer modulus... Precisely calibrated, this ensures that when the patient's weight generates macroscopic positive pressure, the strain force can be transmitted to the underlying FBG grating without loss; while when the patient's body slides or rolls over and generates lateral shear force, the silicone layer can absorb and buffer the local uneven compression caused by the upper MXene rigid electrode on the underlying optical fiber, thus blocking the microbending loss phenomenon that causes spurious drift of the FBG center wavelength from the physical source.

[0028] S12: Independent parallel acquisition of heterogeneous signals Based on the physical isolation structure, the radio frequency photonic demodulation module and the capacitor demodulation module inside the integrated signal processing unit are started separately to perform high-frequency independent sampling.

[0029] On one hand, the radio frequency photonic demodulation module modulates the radio frequency signal onto a broadband optical carrier. After reflection by the FBG array, the original high-frequency optical feature set is extracted by monitoring the notch frequency change of the interference spectrum. This set directly contains optical wavelength mapping values ​​under the combined effects of compressive strain and temperature, and its default hardware sampling frequency... Up to 1000Hz, A high-frequency timestamp used to mark the internal clock of an optical demodulator.

[0030] On the other hand, the capacitor demodulation module uses an analog-to-digital converter (ADC) to capture in real time the changes in resistance and capacitance caused by water vapor adsorption on the upper MXene composite film, and extracts the original set of low-frequency electrical features. Due to the hysteresis physical characteristics of the charge and discharge response of flexible thin films, their default hardware sampling frequency is limited. Typically 50Hz, A low-frequency timestamp used to mark the internal clock of a microcontroller.

[0031] S13: Dynamic timestamp alignment compensation for heterogeneous clock domains Because the optical demodulation module and the capacitor demodulation module are completely different heterogeneous hardware, their internal crystal oscillator frequencies, instruction processing cycles, and instruction response delays are all different. This results in different sets of high-frequency optical raw features being acquired. With low-frequency electrical primitive feature set There is a natural random temporal misalignment between them. Directly fusing misaligned data will lead to severe "spatiotemporal mismatch".

[0032] Therefore, this invention introduces a time synchronization and registration algorithm for a multi-sensor software layer. First, a global absolute time reference source is established at the system layer with integrated signal processing units. And set the reference synchronization frequency for the entire pressure ulcer prevention system to output downstream. 10Hz (i.e., the constant of each synchronization cycle) (100ms).

[0033] Subsequently, discrete measurement values ​​from the heterogeneous clock domains of each sensor are extracted, and virtual feature fusion calculations are performed using an interpolation / extrapolation criterion based on a motion / state change model. Specifically, for the... A number of integer synchronization trigger cycles (i.e., target alignment timestamps) The virtual time registration formula for optical data is defined as follows: ; in, Strictly align timestamps to Virtual optical feature reconstruction values ​​at time; and They are respectively in The two real high-frequency optical sampling timestamps that are closest in distance before and after the time (satisfying) ); and These are specific sampled values ​​from the original high-frequency optical feature set acquired at the corresponding real timestamp. This interpolation algorithm utilizes the time margin of high-frequency sampling to accurately fit the delay-free optical wave characteristics at the target frequency.

[0034] Similarly, for low-frequency electrical humidity signals, interpolation compensation calculations based on the same reference time scale are performed: ; in, Strictly align timestamps to The reconstructed virtual electrical characteristics at time; and In order to be in The two closest real electrical sampling timestamps; and This corresponds to the electrical sampling value.

[0035] Through the aforementioned underlying spatiotemporal alignment compensation, the system completely eliminates the data asynchrony problem caused by hardware clock crystal oscillator deviations in multi-source sensors. This is achieved at each target alignment timestamp. The system will then reconstruct the corresponding virtual optical feature values. Interpreted as original pressure variables Compared with the original temperature variable Reconstructing virtual electrical characteristics Interpreted as raw humidity variable These three parameters are then assembled and fused to output a high-dimensional synchronization feature matrix with uniform dimensions and absolute temporal consistency. The expression for this matrix is: ; The generated high-dimensional synchronization feature matrix This data serves as a crucial and pristine source of correlation constraint data, which is then directly fed into the downstream dynamic baseline construction and environmental parameter drift calibration steps to support the decoupling computation of complex physical field cross-interference.

[0036] S2: Baseline Dynamic Drift Reverse Compensation and Risk Calculation Based on Multiphysics Coupling Model Following the previous stage, in stage S1, the system successfully outputs a high-dimensional synchronization feature matrix with uniform dimension and absolute time consistency through underlying physical isolation and clock synchronization interpolation algorithms. (It contains the aligned original pressure variables) Original temperature variable and the original humidity variable The core objective of this stage is to completely break away from the algorithmic misconception of blindly applying "static medical absolute thresholds (such as 32 mmHg)" in existing technologies. Addressing the deep-seated cross-coupling interference in complex medical microenvironments—where "drastic humidity changes alter thermal conductivity, leading to false temperature drift and pressure distortion in FBG sensors"—we will construct a dynamic compensation algorithm based on hardware physical characteristics to achieve closed-loop debiasing and high-precision risk assessment.

[0037] S21: Adaptive Extraction and Convergence Determination of Personalized Initial Physical Baseline Matrix When a patient first lies on the care mat, their body surface temperature and local microenvironment require a certain amount of time to reach thermodynamic equilibrium. Using the instantaneous power-on data as a baseline would introduce significant initial errors into the system. Therefore, the early warning module first utilizes the high-dimensional synchronization feature matrix output in phase S1. Steady-state baseline extraction is performed.

[0038] The specific operation is as follows: The system sets a synchronization time window on the timeline. (Typically set to a sliding window covering 3000 aligned timestamps, lasting 300 seconds). During each window slide, the system extracts all high-dimensional synchronization feature matrices within the window. Calculate the variance sets for pressure, temperature, and humidity respectively.

[0039] To determine whether a microenvironment has reached both thermodynamic and mechanical equilibrium, a convergence criterion formula is defined: ; in, This represents the environmental fluctuation characteristic value of the current window. For synchronization time window The total number of samples within; and These are the original temperature and humidity variables corresponding to each timestamp within the window; and This represents the window mean of the corresponding variable; This is the dimensional normalization weighting coefficient, with a value of 0.5.

[0040] When the environmental fluctuation characteristic value of three consecutive sliding windows All are less than the set physical convergence threshold. When the value range is set to 0.05~0.1 based on clinical trials, the system determines that the thermodynamic baseline has stabilized, extracts the mean combination of the current window, and formally generates and solidifies the personalized initial physical baseline matrix. This personalized initial physical baseline matrix It will be transmitted to steps S22 and S23 as the absolute zero reference for subsequent calculations of the relative drift.

[0041] S22: Nonlinear Modeling of Thermal Conductivity and Extraction of Pseudo-Temperature Drift Factor Caused by Sudden Changes in Microenvironment Humidity like Figure 2 As shown, in practical applications, when a patient sweats locally, increasing the humidity of the fabric, the introduction of moisture instantly alters the specific heat capacity and thermal conductivity of the outermost layer of Atlas silk fabric and the supporting layer. At this time, the originally constant body temperature is accelerated downward, causing the FBG grating directly beneath it to sense an abnormal temperature jump. This distortion in the hardware temperature response caused by humidity changes is called "pseudo-temperature drift".

[0042] To remove this interference, the system calls the high-dimensional synchronization feature matrix output from stage S1. The original humidity variable in Combined with the personalized initial physical baseline matrix output by S21 Initial humidity baseline First, calculate the real-time humidity deviation scalar. : ; Subsequently, based on the nonlinear polynomial fitting decoupling theory, the real-time humidity deviation scalar was used. As an environmental variable constraint, a low-level error compensation model is established to calculate the pseudo-temperature drift characteristic scalar. : ; in, This represents the spurious temperature drift caused by changes in humidity in the FBG grating. , , The elements of the fabric heat dissipation coupling coefficient matrix, pre-calibrated in a constant temperature and humidity laboratory (corresponding to the quadratic fitting parabola parameters in the literature), are... The mechanism, in its physical sense, characterizes the nonlinear growth slope of the thermal conductivity of a specific fabric after water absorption and curing.

[0043] Generated pseudo-temperature drift characteristic scalar It is a key intermediate variable for achieving software and hardware collaboration and decoupling, and will be directly fed into step S23 for reverse inversion of real physical data.

[0044] S23: Relatively Effective Eigenvalue Reconstruction Based on Multiphysics Coupling Compensation like Figure 3 As shown, the FBG grating is simultaneously modulated by axial strain (pressure) and thermal expansion (temperature) in its working mechanism, and the drift of its interference spectrum notch frequency is a superposition of pressure and temperature parameters. Therefore, after removing the pseudo-temperature drift caused by sudden humidity changes, the system must perform closed-loop depolarization reconstruction.

[0045] The system synchronously receives a high-dimensional synchronization feature matrix. Personalized initial physical baseline matrix and the newly generated pseudo-temperature drift characteristic scalar .

[0046] First, by subtracting the spurious temperature drift in reverse, false data is removed, and the true temperature characteristic variables are reconstructed. : ; Next, because fluctuations in actual temperature will still cause thermal expansion of the FBG grating, thus affecting the original pressure variable... To mitigate the impact of temperature deformation errors, the system utilizes feedback from the temperature-sensing grating to perform cross-sensitivity calibration of the pressure, thereby calculating the true and effective pressure characteristic variable. : ; in, This is the purified pressure value after dynamic reverse compensation of wet bleaching and warm bleaching. The inherent cross sensitivity coefficient of the FBG grating (unit: kPa / °C, representing the amount of pseudo-pressure drift generated per 1°C change; this parameter is factory calibrated).

[0047] Finally, the system repackages the variables after removing the errors and outputs the reconstructed effective feature matrix. : ; The reconstructed effective feature matrix It completely eliminates distorted raw hardware readings and completely removes false alarm sources caused by water vapor evaporation and changes in thermal conductivity in the clinical microenvironment, ensuring that the data delivered to the top-level early warning algorithm has a medical-grade high signal-to-noise ratio.

[0048] S24: Microenvironmental penalty and weighting and calculation of pressure ulcer comprehensive risk index Traditional early warning systems rely on single-point static thresholds (such as triggering an alarm when pressure exceeds 32 mmHg), which easily leads to frequent false alarms. In actual clinical medicine, pressure ulcers are formed by the cumulative effect of pressure over time, and this cumulative destructive process is exponentially accelerated by high temperature and high humidity environments.

[0049] Therefore, the reconstructed effective feature matrix generated by system call S23 Based on the various net data, a dynamic damage model based on spatiotemporal integration and microenvironment penalty weighting is established.

[0050] First, utilize the real temperature characteristic variables. Scalar of deviation from real-time humidity Dynamically calculate the microenvironment penalty weight factor : ; in, It characterizes the degree to which the current temperature and humidity microenvironment weakens the skin tissue's tolerance (the baseline is 1, and the more severe the environment, the greater the weight). and These are the clinically defined dangerous temperature thresholds for skin microcirculation (e.g., 37.5°C) and dangerous thresholds for humidity increment (e.g., +15%RH). and This is a pathological acceleration factor used to amplify the risk contribution under harsh environments.

[0051] Subsequently, the concept of Pressure-Time Integral (PTI) was introduced to calculate the real-time cumulative comprehensive risk index of pressure ulcers. : ; in, The patient's local tissues since the last time they were turned over (starting point in time) The total cumulative load of dynamic damage suffered to date; These are the real and effective pressure characteristic variables along the historical to the current timeline; The system reference synchronization sampling period (100ms) is determined for phase S1.

[0052] The generated pressure ulcer comprehensive risk index By converting static pressure into dynamic load and incorporating environmental factors as a nonlinear catalyst into the integral equation, the system truly achieves a leap from "passive threshold comparison" to "active trend prediction," and this risk index will directly flow to the downstream action execution layer.

[0053] S25: Adaptive hierarchical alarm signal generation based on risk index Finally, the early warning module receives the pressure ulcer comprehensive risk index output by S24. And by calling the dynamic warning threshold vector preset in the system memory (Corresponding to the cumulative damage thresholds for Level 1, Level 2, and Level 3, respectively) Continuous comparison and interception are performed: when Breakthrough to the first level At that time, the system outputs a primary intervention prompt signal. The yellow warning area on the human-computer interaction terminal is illuminated to alert nursing staff that the microenvironment has begun to deteriorate; when Accelerated breakthrough to the second level due to high-weight and harsh environment At that time, the system outputs a secondary alarm dispatch signal. This triggers the voice broadcast at the ward nurses' station; when Approaching the third level of irreversible tissue damage At that time, the system outputs the highest level emergency red signal. .

[0054] S3: Intelligent matching and closed-loop verification of clinical intervention strategies based on multidimensional risk characteristics. In the S2 phase, the system eliminates cross-sensitivity errors through a personalized dynamic compensation algorithm based on the underlying physical characteristics, and finally calculates and outputs a quantified pressure ulcer comprehensive risk index with a high signal-to-noise ratio. and the corresponding signals at each level The goal of this phase is to overcome the limitations of existing products that "can only issue alarms and rely on nurses' subjective experience to handle cases," and to completely map decoupled high-dimensional physiological data to a clinical medical knowledge base at the data and algorithm level, automatically matching and issuing the optimal intervention plan; at the same time, through continuous sensor data feature capture, to automatically identify the intervention actions of nursing staff and quantify and verify the intervention effect at the pure algorithm level.

[0055] S31: Spatial Mapping and Similarity Matching of Clinical Interventions Based on Multimodal Features When the pressure ulcer comprehensive risk index output by the early warning module Trigger signals at any level At that time, the system immediately activates the auxiliary decision-making engine.

[0056] First, the system calls the reconstructed effective feature matrix output in stage S23. (Includes real and effective stress characteristic variables) True temperature characteristic variables Real-time humidity deviation scalar (and the comprehensive pressure ulcer risk index generated in stage S24) By concatenating the dimensions of the four core parameters mentioned above, a real-time risk state vector representing the current deterioration state of the local organization is constructed. : .

[0057] Subsequently, the system reads a database of standard clinical intervention cases pre-loaded in local storage or the cloud. The database contains a large number of historical clinical data dictionaries calibrated by the dermatology and nursing departments of tertiary hospitals. Each dictionary entry contains a historical feature vector and the corresponding successful intervention plan (e.g., requirements for turning angle, whether sweat needs to be wiped away, whether local airflow drying is required, etc.).

[0058] To eliminate the differences in physical dimensions (dimensionless representation of pressure, temperature, and humidity) and the inherent correlation interference between temperature and humidity, the system employs the Mahalanobis distance algorithm to calculate the real-time risk state vector. With the database Historical feature vectors Intervention matching scalar : ; in, It is the inverse of the multidimensional feature covariance matrix, used to eliminate the influence of the linear correlation between physical quantities such as temperature and humidity on the weight of distance calculation.

[0059] The system traverses the database and filters out intervention plan matching scalars. The smallest (i.e., most similar in medical features) Top-1 historical records are used to extract their associated nursing execution codes and generate an optimal clinical intervention strategy matrix. The matrix contains standardized structured text data, such as specific instructions like "It is recommended to turn over 30 degrees to the right side" and "In areas with high humidity, remove the bedding for ventilation for 5 minutes."

[0060] S32: Issuance of instructions for the integration of tiered alarm and intervention strategies In generating the optimal clinical intervention strategy matrix Then, the system compares it with the corresponding level signals generated in stage S25. Data packets are encapsulated to generate encrypted integrated intervention and scheduling data packets. .

[0061] The system uses its internal wireless communication module (such as Wi-Fi or 4G / 5G module) to transmit the integrated intervention and scheduling data packet. The alerts are pushed to the central monitoring screen at the nurses' station and to the mobile handheld terminals of the responsible nurses. Nursing staff no longer receive only a simple "beep" alarm sound and a red light, but instead receive information including "current risk level, abnormal area heat map (based on..."). The integrated information panel, which includes mapping and quantitative intervention guidance, completely solves the clinical pain point of traditional equipment "lacking precise intervention basis after early warning".

[0062] S33: Adaptive Recognition of Physical Intervention Actions Based on Temporal Pressure Gradient Abrupt Changes like Figure 4 As shown, after the command is issued, the system does not passively wait for manual alarm clearing, but enters an "intervention action sensitive capture state". In order to identify at the software level whether the caregiver has performed physical intervention operations such as turning over or wiping, the system tracks the pressure topology changes fed back by the sensor array in real time.

[0063] For nodes in the core area experiencing excessive pressure, the system calculates the rate of change of local pressure temporal gradient between two consecutive synchronization cycles on the time axis. : ; in, The system reference synchronization sampling period is determined in phase S1.

[0064] In a normal bedridden state, even with slight breathing or muscle twitching, the rate of change of local pressure temporal gradient is... The absolute value of this gradient is usually kept at a low level. When caregivers turn the patient over or elevate a specific area, the load on the previously compressed area is instantly unloaded, causing a sharp negative change in the gradient value.

[0065] The system sets an empirical threshold for recognizing intervention actions. When the system detects a high-voltage area When a sudden increase in pressure occurs in an adjacent area, the system algorithmically determines that "a physical intervention event has occurred" and automatically marks an intervention action timestamp in the system log. This purely algorithmic feature recognition step perfectly replaces the reliance on external mechanical switches or manual clicks for confirmation, achieving seamless monitoring of the nursing process.

[0066] S34: Post-intervention microenvironmental status assessment and adaptive reconstruction of physical baseline triggering The intervention action was marked with a timestamp. Afterward, the system waits for a fixed period of microenvironment stabilization (e.g., 5 minutes) to allow the new pressure posture and local microenvironment (temperature and humidity diffusion) to re-establish thermodynamic and mechanical equilibrium.

[0067] Subsequently, the system calls upon the latest round of sensor inputs, re-executes the complete solution logic from S1 to S2, and obtains the latest comprehensive pressure ulcer risk index after physical intervention. And extract the latest real-time humidity deviation scalar. .

[0068] The system calculates a scalar for evaluating the effectiveness of the intervention used to measure the quality of this nursing procedure. : ; in, This is the extreme risk index before the intervention action is triggered (when an alarm is triggered).

[0069] The system makes a logical judgment: when the intervention effectiveness assessment scalar is verified... The humidity deviation is greater than the set effective intervention threshold (e.g., 60%) and the latest calculated real-time humidity deviation scalar. When the temperature has dropped to a safe level close to zero, the system determines that the clinical intervention was successful and the local adverse microenvironment has been reset.

[0070] To avoid new accumulated errors in subsequent monitoring, the system forcibly issues a low-level reset control flow at this point. The control flow returns directly to stage S21, instructing the signal processing unit to clear the historical integration cache and, using the new body position and new steady-state microenvironment data formed after the patient intervention, adaptively extract and solidify the personalized initial physical baseline matrix for the next cycle. .

[0071] S4: Evolution of Personalized Prediction Models Based on Closed-Loop Feedback and Dynamic Updates of the Intervention Knowledge Base Following the previous stage S3, in stages S1 to S3, the system has perfectly achieved a short-term tactical-level closed loop, from underlying heterogeneous data synchronization and physical environment error compensation to the issuance of clinical intervention strategies and action verification. However, in the real-world scenario of long-term care, the physiological functions of long-term bedridden patients, such as thinning of the stratum corneum, loss of subcutaneous fat, and deterioration of local microcirculation, will undergo irreversible physiological degeneration over time. If the system continues to rely on fixed pathological experience parameters, such as those set in stage S24, the system will suffer irreversible damage. and This will gradually create an underestimation of the actual risks over several weeks or months.

[0072] The purpose of this embodiment is to overcome the technical limitations of existing pressure ulcer prevention equipment, which relies on a "fixed model upon leaving the factory" approach. It utilizes the vast amount of historical execution data generated by the S3 phase closed-loop verification to perform reinforcement learning and data mining over a long time series. By quantifying the rate of decline in individual patient tolerance, it adaptively updates the underlying pathological penalty weights and continuously optimizes the clinical knowledge base, thereby achieving a closed-loop system where the nursing system dynamically evolves throughout a patient's life as their physical condition changes.

[0073] S41: Long-term feature aggregation of core risk parameters and quantification of individual tolerance degradation In the background where the system runs continuously, the data mining module initiates background calculations according to fixed long-term time windows, such as every 24 hours as an evaluation cycle. .

[0074] First, the system traces and extracts all core data from the previous period. The data mining module uses the timestamps of each intervention action generated in phase S33. Using time segmentation anchors, extract the real-time risk state vector between two adjacent intervention actions (i.e., the resting period when the patient maintains the same pressure position). (Including actual pressure, actual temperature, humidity deviations, and risk indices). These multidimensional time-series data are then concatenated, compressed, and stored according to time series to generate a long-term state sequence set for long-term evaluation. .

[0075] To accurately quantify the changes in the fragility of local tissues in patients over time, the system is based on a long-period state sequence set. The cumulative slope of the pressure ulcer risk index is extracted under the same average pressure load level. The system calculates the individual tolerance degradation factor. : ; in, Characterizes the rate of decline in the patient's current skin microcirculation (baseline value is 1, the larger the value, the worse the tolerance). This refers to the number of resting periods within this cycle that meet the same basic pressure conditions. For the current evaluation period, the first The average growth gradient of the comprehensive pressure ulcer risk index over a given time period; The gradient of historical risk index growth under the same stress conditions when the patient first uses this system (i.e., the health baseline established on the first day of admission).

[0076] Calculated individual tolerance degradation factor This will serve as a key individual physiological correction parameter, which will be directly transferred to the downstream weight optimization step.

[0077] S42: Adaptive Optimization and Update of Dynamic Microenvironment Penalty Weights In the microenvironment penalty and weighting logic of stage S24, the preset pathological acceleration coefficient (temperature acceleration coefficient) is... With humidity acceleration coefficient This represents the standard level of punishment in a statistical sense for the population. However, as the individual degenerates, the destructive power of the same hot and humid environment is amplified for the patient.

[0078] Therefore, the system receives the individual tolerance degradation factor output in stage S41. The original basic parameters are nonlinearly amplified and optimized. The system executes the following micro-environment weight iterative algorithm: ; ; in, and These are the updated temperature acceleration coefficient and the updated humidity acceleration coefficient, respectively. and This refers to the old speedup factor currently in use by the system; and The temperature and humidity degradation sensitivity hyperparameter preset for the system (usually) (Because the skin of disabled patients becomes thinner, their sensitivity to immersion in moisture increases more drastically). A small bias (1e-5) is used to prevent the logarithmic function from going out of bounds.

[0079] The generated updated temperature acceleration factor With the updated humidity acceleration factor Completed the underlying integral equation The complete overhaul allows the system to issue early warnings for more vulnerable skin, and these two parameters will be sent to S44 for global replacement.

[0080] S43: Reinforcement Learning and Evolution of Clinical Intervention Case Databases Based on Intervention Efficacy In addition to the evolution of the physical monitoring model, this system also evolves itself to support software-based nursing strategies. The standard clinical intervention case database in Phase S31... It should not be static, unchanging data, but rather a process of natural selection based on actual clinical feedback.

[0081] The system extracts all intervention effectiveness assessment scalars from the historical records of stage S34. And the corresponding optimal clinical intervention strategy matrix The system establishes a reward and punishment mechanism similar to reinforcement learning: For each historical feature vector retrieved from the database Calculate the policy confidence reward scalar of the policy scheme and its associated strategy. : ; in, The updated policy credibility weight; This is the previous copyright repetition of the strategy; The learning rate (e.g., 0.1) is used to smooth the transition between historical weights and new feedback. Based on the expected threshold of basic effectiveness (e.g., 60%).

[0082] If a certain optimal clinical intervention strategy matrix After actual implementation, its intervention effectiveness assessment scalar If the performance consistently falls short of expectations, the strategy confidence reward scalar will be affected. The weight will decrease accordingly. When the scalar falls below the elimination threshold, the system will move it to the degraded storage area; conversely, if certain intervention strategies combined with new posture combinations achieve excellent mitigation effects, its weight will increase significantly.

[0083] The system recompiles the database based on the latest weights and updates of all the above strategies, and outputs an evolved clinical intervention case database. This database will serve as a higher-quality knowledge base to guide clinical care decisions in the next cycle.

[0084] S44: Global Model Closed-Loop Distribution and System Lifecycle Reset As the final closing operation of the "specific implementation method", the system control main control program updates the temperature acceleration coefficient generated in stage S42. With the updated humidity acceleration factor This overwrites the existing steady-state hyperparameters in the flash memory register of the early warning module; simultaneously, it rewrites the evolved clinical intervention case database generated in stage S43. Covers legacy mapping dictionaries, whether in the cloud or locally.

[0085] Subsequently, the system releases the memory space of the S41 stage cache and resets the current clock pointer to the starting point of the next evaluation cycle.

[0086] To fully demonstrate the non-obviousness and substantial technical contribution of the above-described specific embodiments of the present invention (S1 to S4 stages) compared to the prior art, this section will introduce a clear comparative benchmark, conduct an extremely objective comparative analysis of physical performance and algorithm logic, and thereby demonstrate how the present invention has successfully overcome the long-standing technical bias in the field.

[0087] In this case, the publicly available academic paper "Research on Radio Frequency Photonic Sensors for Pressure Ulcer Prevention" (authors: Chen Kaixin et al., published in Optical Communication Technology) is set as the closest prior art (comparative example) in this analysis.

[0088] The technical architecture and physical performance parameters of the comparison are as follows: This comparative example presents a radio frequency photonic sensor for pressure ulcer prevention. It modulates a radio frequency signal onto an optical carrier and utilizes the notch filter frequency variation of the interference spectrum formed by reflection from a fiber Bragg grating (FBG) array to demodulate pressure and temperature. Its core physical specifications are: within a static pressure range of 0–42 kPa, the sensor achieves a pressure sensitivity of 0.64 MHz / kPa and an excellent pressure accuracy of 16 Pa; within a temperature range of 35–40°C, the temperature sensitivity reaches 1.07 MHz / °C and the temperature accuracy reaches 0.0093°C. In a practical application scenario, the encapsulated FBG was fixed to a hard wooden board, and an insulated sheet was placed on top for human body testing.

[0089] This comparison represents a typical conventional technical approach in the current field of pressure ulcer prevention, which pursues the ultimate single-point / dual-parameter static measurement accuracy. Essentially, it is an isolated open-loop physical measurement system that does not integrate a humidity sensing dimension and assumes that the hardware is within a relatively ideal static physical boundary. It directly uses the raw physical quantities collected for medical assessment, lacks an underlying decoupling mechanism for complex multi-physical field cross-coupling interference, and does not include any dynamic closed-loop evolution logic.

[0090] When the complete embodiments S1 to S4 of the present invention are objectively compared with the comparative examples in a real, complex clinical scenario of "frequent turning over of long-term bedridden patients accompanied by profuse local sweating (coexistence of drastic temperature and humidity changes and dynamic shear forces)," the comparative examples reveal the following fatal physical and logical shortcomings: Logical blind spots in dealing with sudden changes in high humidity microenvironment When a patient experiences sustained local pressure and profuse sweating, the humidity of the skin's microenvironment increases dramatically from 40%RH to over 90%RH in a short period. After absorbing a large amount of moisture, the fabric's physical specific heat capacity and thermal conductivity undergo a nonlinear and dramatic change, accelerating the conduction of the previously constant surface heat radiation to the deeper layers.

[0091] The weakness of the comparative amplifier: Due to the lack of a sensing and constraint mechanism for the environmental variable of humidity, the FBG grating experiences a drastic center wavelength drift due to changes in thermal conductivity. The comparative amplifier's demodulation system mechanically interprets this physical phenomenon as "abnormal local temperature rise," outputting distorted temperature data and directly causing false alarms.

[0092] Advantages of this embodiment: In stage S22, this invention introduces a multidimensional dynamic compensation coefficient fitting mechanism, utilizing the extracted real-time humidity deviation scalar. As an environmental variable constraint, the pseudo-temperature drift characteristic scalar was accurately calculated. ( In stage S23, the property is stripped in reverse to generate pure, true temperature feature variables. Compared with real effective stress characteristic variables This embodiment eliminates the temperature measurement illusion caused by changes in thermal conductivity at the algorithm level, resulting in a more effective reconstructed feature matrix in the encapsulated output. It maintains an extremely high signal-to-noise ratio even in extremely humid environments.

[0093] Physical structure failure in response to dynamic deformation shear forces When the patient undergoes irregular dynamic deformations such as turning to the side or struggling, the pad will be subjected to a complex mixture of horizontal shear forces and vertical compressive forces.

[0094] The comparative example's weakness: The comparative example rigidly fixes the FBG to a wooden board or a single-layer medium. If it is simply stacked with a flexible capacitive humidity sensor according to the conventional thinking in this field, the upper rigid electrode will directly squeeze the lower optical fiber under shear force, resulting in severe micro-bending loss. This causes a huge "pseudo-pressure" jump in the optical notch frequency, and the static accuracy of the comparative example, which is as high as 16 Pa, collapses instantly under dynamic shear, and the error can be amplified by hundreds of times.

[0095] The advantage of this embodiment is that the present invention innovatively constructs a 1mm thick material containing a specific relative permittivity in stage S11. With mechanical buffer modulus The soft silicone insulating layer. This physical component is not a simple stacking, but rather acts as a mechanical low-pass filter, absorbing the localized uneven compression of the underlying optical fiber by lateral shear force, while simultaneously mitigating parasitic capacitance deviations between the electrodes. By reducing the noise level to below microfarads, structural noise crosstalk in dynamic environments is eliminated at its physical source.

[0096] Heterogeneous spatiotemporal misalignment and lack of long-term evolution capability The limitations of the comparative scale: It only has an optical acquisition clock (1000Hz level), which cannot handle the fusion of multi-source heterogeneous data. Forcibly incorporating humidity electrical signals will inevitably lead to severe spatiotemporal misalignment, causing downstream early warning algorithms relying on multi-parameter product integration to fail. Furthermore, the comparative scale's evaluation criteria are factory-fixed and cannot cope with the physiological tolerance degradation experienced by disabled patients over several months.

[0097] The advantage of this embodiment is that, in stages S12 and S13, it targets the original set of high-frequency optical features. (sampling rate) timestamp ) and the set of original characteristics of low-frequency electricity (sampling rate) timestamp Based on a global absolute time reference source Using interpolation and extrapolation algorithms at the target frequency With synchronization period constant Below, virtual reconstructed values ​​with strictly aligned timestamps were generated. and Finally assembled in High-dimensional synchronization feature matrix with absolute time alignment ,Depend on Composition. And in phase S4, through a long-period state sequence set. With the extracted individual tolerance degradation factors The updated temperature acceleration coefficient has been achieved. With the updated humidity acceleration factor The adaptive iteration is a long-term lifecycle closed loop that is completely absent in proportional systems.

[0098] In the conventional field of medical wearable / flexible care pads, those skilled in the art are generally constrained by design biases such as "low power consumption, low computing power, and limited heat dissipation." They believe that to simultaneously perform interpolation reconstruction calculations on optical and electrical signals at a raw rate of 1000Hz on an embedded microcontroller (MCU), and then calculate the inverse covariance matrix (… ) intervention program matching scalar And continuously iterate and update the strategy confidence reward scalar. With the evolution of clinical intervention case database This will inevitably lead to MCU computing power overload, memory overflow, or even system crash.

[0099] This embodiment first utilizes a specific and The physical silicone layer filters out most of the high-frequency mechanical noise, giving the underlying physical signal a high degree of smoothness. Based on this smoothness, this embodiment creatively uses the global reference synchronization frequency in stage S13. Anchored at 10Hz ( This specific downsampling frequency point perfectly satisfies the Nyquist sampling theorem for the physiological changes (minute-level hysteresis) of pressure ulcers in humans, while also drastically reducing the computational load of industrial-grade algorithms by 99%, thus enabling the extremely complex dynamic early warning threshold vector... Calculation and pressure ulcer comprehensive risk index This makes it possible to run long-term axis integration smoothly on low-power MCUs. Those skilled in the art, constrained by computational power anxiety, could not foresee that "physical layer low-pass filtering + 10Hz specific frequency dimensionality reduction and spatiotemporal alignment" could perfectly support such advanced decoupling algorithms. This solution overcomes the computational power bias.

[0100] This embodiment is based on environmental fluctuation characteristic values ​​in stage S21. Synchronization Time Window A rigorous physical convergence criterion mechanism (based on 3000 sample points) is employed. The system does not blindly operate decoupled at all times; rather, it must satisfy certain conditions. Less than the physical convergence threshold Only after meeting the boundary conditions can the mean within the window be extracted. (etc.), solidify the personalized initial physical baseline matrix ,Include Furthermore, in stage S3, this invention does not merely issue an alarm, but rather calculates the target's travel distance. The closed-loop drive of a micro telescopic motor is used to construct the local support gap volume. From the semicircular cross-sectional area function With volumetric loss The calculation was performed using the airflow conduction drag coefficient. With sealing coefficient Initial volume of the ventilation chamber Instantaneous extraction negative pressure is generated based on this. At atmospheric pressure With the cooperation of the authorities, moisture was forcibly extracted. Subsequently, the effectiveness of the intervention was assessed using a scalar model. With the latest humidity scale After the microenvironment is reset, a forced reset and recalculation of the new baseline are performed. By physically intervening to forcibly create the steady-state boundary conditions required by the algorithm, the problem of high-precision decoupling algorithms being unable to run in flexible environments is solved.

[0101] To further provide conclusive quantitative evidence, the following is a comparison of test data between the specific embodiment system of this invention and a comparative example (a baseline system containing only FBG radio frequency photon demodulation and without underlying compensation logic) in a medical clinical environment:

[0102] The above experimental data irrefutably demonstrate that while the comparative model can achieve high-precision physical parameters under ideal static conditions, its physical performance completely collapses due to cross-coupling interference once it deviates from its ideal laboratory boundaries and enters the real and harsh clinical nursing scenario of pressure ulcers. However, the specific embodiment of this invention achieves this through "bottom-layer physical isolation shielding (…)" (etc.) + high-dimensional timestamp alignment compensation ( (etc.) + Core multiphysics nonlinear decoupling inversion ( (etc.) + Closed-loop negative pressure physical reset microenvironment ( (etc.) + Long-term strategy confidence learning ( The full-stack technical architecture of "etc." transforms the originally mutually restrictive measurement errors into mutually calibrated reference benchmarks. It not only perfectly preserves the high resolution of radio frequency photonic sensors, but also achieves unparalleled robustness in harsh environments and long-term individual evolution capabilities. The design logic and final effect of this technical solution have far exceeded the expectations of ordinary people in the field who can achieve it by using conventional methods, and have extremely outstanding substantive features and significant progress.

[0103] Example 2 The system in this embodiment consists of the following four main modules: Heterogeneous multimodal sensor signal underlying physical acquisition and spatiotemporal synchronization fusion module This module serves as the physical data input and spatiotemporal reference anchoring port for the entire system. Its core function is to block signal crosstalk at the hardware physical layer and to perform virtual reconstruction of heterogeneous clock domain data. This module comprises the following three sub-units: 1. Sensor Network Underlying Dual Physical Isolation Building Block This unit is a purely hardware-level physical constraint structure. For the physical wiring at the junction of the main body pad and the detachable U-shaped core module, the system incorporates a variable thickness of a soft silicone insulating layer between the optical network and the capacitor network. (Constant is 1mm). This unit relies on the specific relative permittivity of silicone. By implementing electromagnetic shielding logic, the parasitic capacitance deviation caused by the electrode spacing between heterogeneous sensing layers is significantly reduced. At the same time, relying on its specific mechanical buffer modulus It executes mechanical low-pass filtering logic to absorb the horizontal shear force generated by the dynamic deformation of the patient and block the optical micro-bending loss caused by the extrusion of the rigid electrode plate.

[0104] 2. Independent parallel acquisition unit for heterogeneous signals This unit contains two independent demodulation circuits. The radio frequency photonic demodulation sub-circuit operates at a natural frequency of 1000Hz, marked by a high-frequency timestamp within an internal clock. Below, the high-frequency optical original feature set under the dual effects of compressive strain and temperature is continuously extracted. The capacitor demodulator circuit is limited by charge and discharge hysteresis, operating at a low-frequency timestamp of 50Hz. Extracting the original set of low-frequency electrical features These two sets are naturally asynchronous on the timeline and will be transmitted as raw input data to the next unit.

[0105] 3. Heterogeneous clock domain timestamp dynamic alignment compensation unit This unit is the core of hardware-software co-registration, responsible for eliminating system-level time misalignment caused by asynchronous sampling. The unit has a built-in global absolute time reference source. And set the system's reference synchronization frequency. With synchronization period constant This unit receives the original set of high-frequency optical features. With low-frequency electrical primitive feature set By using interpolation and extrapolation logic, timestamps are aligned to a specific target. Virtual reconstructed values ​​are generated. Finally, the decoded original stress variables are... Original temperature variable and the original humidity variable The assembly is aligned to output a high-dimensional synchronous feature matrix with uniform dimensions and absolute temporal consistency. This matrix will be injected into Module 2 as an absolutely pure and unique baseline data stream.

[0106] Baseline dynamic drift inverse compensation and risk calculation module based on multiphysics coupling model (corresponding to S2) This module aims to solve the problem of underlying temperature measurement distortion caused by sudden changes in the fabric's humid and thermal microenvironment, completely abandoning static medical thresholds. It is the core algorithm for achieving "active de-biasing early warning" in pressure ulcer prevention. This module is further subdivided into the following five collaborative processing units: 1. Personalized Initial Physical Baseline Matrix Adaptive Extraction Unit This unit continuously receives high-dimensional synchronization feature matrices from the first module. And maintain a synchronization time window of 300 seconds in memory. During the window sliding process, the unit calculates the environmental fluctuation characteristics of temperature and humidity within the window. Only when environmental fluctuation characteristics are continuously monitored. Less than the system's preset physical convergence threshold Only then does the unit determine that the thermodynamics of the underlying microenvironment has converged, extract the window mean, and solidify the output of a personalized initial physical baseline matrix. (It includes the initial humidity baseline) (e.g., absolute zero parameters).

[0107] 2. Thermal conductivity nonlinear modeling and pseudo-temperature drift factor extraction unit This unit is responsible for "identifying hardware distortion". It receives a high-dimensional synchronization feature matrix. With personalized initial physical baseline matrix Extract real-time raw humidity variables Compared with the initial humidity baseline in steady state The real-time humidity deviation scalar is obtained by differential calculation. Subsequently, the unit invokes a nonlinear parabolic fitting equation, using the aforementioned deviation scalar as the independent variable, to calculate and extract the pseudo-temperature drift characteristic scalar caused by changes in the fabric's water absorption and thermal conductivity. .

[0108] 3. Multiphysics coupling compensation relative effective eigenvalue reconstruction unit This unit is responsible for "reverse closed-loop debiasing". It receives the original pressure variable. Original temperature variable and the aforementioned newly generated pseudo-temperature drift characteristic scalar By subtracting the pseudo-temperature drift stripping error in reverse, the pure, true temperature characteristic variable is output. Furthermore, by combining the cross-sensitivity coefficient of the grating, the true and effective pressure characteristic variables, stripped of the influence of thermal expansion, were reconstructed. Finally, these pure quanta are encapsulated and packaged to output a reconstructed effective feature matrix. .

[0109] 4. Microenvironment penalty weighting and pressure ulcer comprehensive risk index calculation unit This unit performs dynamic damage integration. It calls for the reconstruction of the effective feature matrix. The parameters in the calculation are used to determine the microenvironment penalty weight factor based on the real-time deterioration level. Combining this penalty factor with real, effective stress characteristic variables... Integrating over time, the system continuously outputs a dynamically accumulating comprehensive pressure ulcer risk index. .

[0110] 5. Adaptive hierarchical alarm signal generation unit This unit will calculate the comprehensive pressure ulcer risk index in real time. With the dynamic warning threshold vector stored in the register A three-level comparison is performed. Once the corresponding critical point is exceeded, signals at each level are generated and output sequentially according to their severity. (i.e., primary intervention warning signals) Secondary alarm dispatch signals Highest level emergency red signal ), and then transfer to the intervention module.

[0111] III. Intelligent matching of clinical intervention strategies and closed-loop verification of intervention effects (corresponding to S3) This module breaks away from the industry's old practice of a single "beep alarm," directly mapping quantified multidimensional physiological risks into specific clinically executable nursing actions, and achieving unmanned monitoring of the effects of manual intervention at the software level. It comprises four units: 1. Spatial mapping and similarity matching unit for clinical intervention Upon triggering the alarm, the unit synchronously receives the reconstructed valid feature matrix. Comprehensive Risk Index of Pressure Ulcers The two are concatenated to form a real-time risk state vector with reduced dimensionality. Subsequently, the unit was used in the standard clinical intervention case database. The Mahalanobis distance traversal operation is performed to calculate the minimum intervention scheme matching scalar. This allows us to identify and output the optimal clinical intervention strategy matrix that best reflects the current deterioration state. (Contains text-based instructions for turning over or ventilation).

[0112] 2. Integrated Intervention and Scheduling Data Packet Encapsulation and Distribution Unit This unit will combine signals from each level. Severity grading and optimal clinical intervention strategy matrix The specific instructions are aggregated and encapsulated into standardized integrated intervention and scheduling data packets. The data is pushed to the ward nurse station and handheld terminal via wireless communication link to complete the closed loop of precise scheduling.

[0113] 3. Adaptive recognition unit for sudden changes in time-series pressure gradients After the command is issued, the unit enters a sniffing state to track the real and effective pressure characteristic variables transmitted from the underlying layer in real time. Calculate the rate of change of local pressure temporal gradient within adjacent time stamps. Once a large negative abrupt change in the gradient is detected (indicating that the nurse has lifted and turned the patient), the system automatically timestamps the intervention in the event log. This enables algorithm-level seamless confirmation of nursing actions.

[0114] 4. Closed-loop assessment of intervention effectiveness and baseline reconstruction triggering unit After recording the intervention timestamp, the unit waits for the environment to stabilize and then re-acquires the post-intervention risk values. By calculating the rate of change before and after the intervention, a scalar for assessing the intervention's effectiveness is generated. If the verification meets the standards and the latest real-time humidity deviation scalar value is available at this time... Upon fallback, the unit will force a low-level hardware reset interrupt, returning control to the "Personalized Initial Physical Baseline Matrix Adaptive Extraction Unit" in Module 2, instructing it to re-extract the personalized initial physical baseline matrix according to the new attitude. This completes the tactical-level system closed loop.

[0115] Personalized prediction model evolution and dynamic update module for intervention knowledge base (corresponding to S4) This module is the core backend evolution cluster that endows the system with the ability of "lifelong learning" and "adaptive adjustment as physical condition declines," completely overcoming the technical barrier of models being rigid from the outset. It includes four long-term processing units: 1. Long-term feature aggregation and individual tolerance degradation quantification unit At the end of the preset 24-hour evaluation period, this unit starts the background process. It is based on the timestamps of each intervention action. Segment the data stream and extract the real-time risk status vector during the rest period. Push and concatenate to generate a long-period state sequence set. By comparing the risk escalation slopes of historical and current stress levels, the system calculates and outputs an individual tolerance degradation factor characterizing microcirculatory decline. .

[0116] 2. Dynamic microenvironment penalty weight adaptive optimization unit This unit receives individual tolerance degradation factors. The original system's steady-state parameters are amplified by a penalty calculation. Guided by the logarithmic decay function, the system calculates and outputs the updated temperature acceleration coefficient. With the updated humidity acceleration factor These two parameters fundamentally tighten the system's tolerance for alarms related to this vulnerable individual.

[0117] 3. Clinical intervention case database reinforcement learning evolution unit For nursing guidance at the software level, this unit extracts the matrix of all optimal clinical intervention strategies within the cycle. and its corresponding intervention effectiveness assessment scalar For strategies that perform poorly, the system reduces their policy confidence and reward scalar. These strategies are then downgraded and eliminated; those with excellent results are given increased weight, and ultimately a high-confidence evolutionary clinical intervention case database is recompiled. .

[0118] 4. Global model closed-loop distribution and lifecycle reset unit As the main control scheduling and cleanup unit, it will generate the newly updated temperature acceleration coefficient. With the updated humidity acceleration factor Overwrite to the early warning flash memory in Module 2; evolve the clinical intervention case database. Overwrite the mapping dictionary in Module 3. After thoroughly completing the multi-dimensional parameter and knowledge base overwrite, clear the long-period state sequence set. The cache is cleared, freeing up memory, allowing the entire multi-parameter fusion early warning system to enter the patient's next long care cycle with a more acute physiological perception and a more intelligent nursing database.

[0119] Example 3 This embodiment provides an intelligent pressure ulcer prevention electronic device. This device, as the physical carrier of an integrated signal processing unit (host), establishes a physical communication connection with the front-end modular care pad body and its internal multi-parameter flexible sensor network. The electronic device's physical architecture includes at least: a processor, a memory, a communication bus, and a multimodal sensor communication interface. Further explanation is provided in conjunction with Embodiments 1 and 2, as follows... Figure 5 The structure shown. Figure 5 A schematic diagram of the structure of a computer device provided in an embodiment of this application. The computer device includes: Processor, memory, communication bus, and computer programs stored in memory that can run on the processor.

[0120] The processor can call the computer program in memory, and when executing the program, implement the pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling provided in the above embodiments. The method includes: S1, physically decoupling the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet through a flexible isolation layer, and collecting high-frequency optical raw feature sets in the heterogeneous clock domain in parallel. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable S2. When the monitored microenvironment is determined to have reached a steady state, extract the personalized initial physical baseline matrix. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

[0121] Furthermore, computer equipment also includes: The Communications Interface (CI) is used for communication between the memory and the processor.

[0122] The memory may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive.

[0123] If the memory, processor, and communication interface are implemented independently, they can be interconnected via a bus to communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0124] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0125] A processing module is used to perform calculations and processing on data. In this application, the processing module supports the system of Embodiment 2 and the method of Embodiment 1. In one implementation, the processing module is one or more processors, such as an application processor (AP), an application-specific integrated circuit (ASIC), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural network processing unit (NPU). Different processing units can be independent devices or integrated into one or more processors. The controller can generate operation control signals based on the instruction opcode and timing signals to control instruction fetching and execution. The high-speed cache memory in the processor can store recently used or repeatedly used instructions or data, reducing processor waiting time.

[0126] The controller can serve as a central nervous system and command center. Based on the instruction opcode and timing signals, the controller generates operation control signals to control instruction fetching and execution. The processor may also include memory for storing instructions and data. In some embodiments, the processor's memory is a cache memory. This memory can store instructions or data that the processor has just used or that is used repeatedly. If the processor needs to reuse the instruction or data, it can directly retrieve it from the memory. This avoids repeated accesses, reduces processor waiting time, and thus improves system efficiency.

[0127] The aforementioned processor, memory, and multimodal sensor communication interface are physically connected and exchange data via a communication bus. Specifically, the multimodal sensor communication interface includes independent radio frequency optical interferometry receivers and ADC analog-to-digital converter receivers, respectively used to receive the high-frequency optical raw feature set from the front-end at a sampling rate of 1000Hz. Low-frequency electrical primitive feature set with a sampling rate of 50Hz The communication bus adopts a high-bandwidth industrial-grade bus standard to ensure that heterogeneous data is transmitted to the processor without additional channel blocking delay.

[0128] As a non-volatile computer-readable storage medium, memory internally stores computer program code and instruction sets that can be called by the processor. When the processor is physically powered on and calls the program, it strictly executes the following hardware-level operations: Clock-aligned execution: The processor calls its internal global absolute time reference source. The arithmetic logic unit (ALU) is driven to execute interpolation and extrapolation algorithms to force synchronization of non-synchronous optical / electrical signals received via the bus at a hardware clock cycle of 10Hz, and to generate and cache a high-dimensional synchronization feature matrix in the register. This matrix contains the original pressure variables with absolutely aligned timestamps. Original temperature variable and original humidity variables .

[0129] Hardware distortion reverse stripping execution: The processor extracts a personalized initial physical baseline matrix based on the physical convergence threshold set in memory. Subsequently, the processor invokes the floating-point unit (FPU) to calculate the real-time humidity deviation scalar value. By performing nonlinear polynomial fitting, a pseudo-temperature drift characteristic scalar representing the effect of changes in thermal conductivity on the underlying hardware is calculated. The processor performs inverse subtraction logic in the accumulator to remove the aforementioned spurious temperature drift and cross-sensitivity errors, outputting a high signal-to-noise ratio reconstructed effective feature matrix. (Includes real and effective stress characteristic variables) With real temperature characteristic variables ).

[0130] Risk calculation and peripheral driver execution: The processor uses the aforementioned pure real variables to calculate the comprehensive risk index of pressure ulcers. And matched the optimal clinical intervention strategy matrix Subsequently, a comprehensive intervention and scheduling data packet is generated by encapsulating the data through a network control chip. This drives external communication modules (such as wireless radio frequency transmitters) to send physical intervention commands to downstream care terminals. Simultaneously, the processor calculates individual tolerance degradation factors based on long-term time-series data. The updated temperature acceleration coefficient With the updated humidity acceleration factor Write back to local memory to complete the hardware-level replacement of underlying parameters.

[0131] In another implementation, the processing module can achieve certain functions through the logical relationships of hardware circuits. These logical relationships can be fixed or reconfigurable. For example, the processor can be a hardware circuit implemented using an Application-Specific Integrated Circuit (ASIC) or a Programmable Logic Device (PLD), such as a Field-Programmable Gate Array (FPGA). In reconfigurable hardware circuits, the processor loads a configuration document to configure the hardware circuit. Furthermore, for the complex quadratic programming and semi-positive definite matrix decomposition involved in the trajectory optimization of this application, hardware circuits designed for artificial intelligence or high-performance matrix operations can also be used, such as Neural Processing Units (NPUs), Tensor Processing Units (TPUs), and Deep Learning Processing Units (DPUs). The aforementioned processor can call and execute instructions from memory to quickly construct a safe corridor and output the globally optimal trajectory.

[0132] A storage module can be used to store computer executable program code and data, wherein the executable program code includes instructions. For example, the storage module can store the code of the method provided in Embodiment 1 of this application. The storage module may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device or flash drive.

[0133] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0134] Display devices are used to display images, videos, etc. Display devices may include display panels, which may employ liquid crystal displays (LCDs), organic light-emitting diodes (OLEDs), active-matrix organic light-emitting diodes (AMOLEDs), flexible light-emitting diodes (FLEDs), MiniLEDs, MicroLEDs, Micro-OLEDs, quantum dot light-emitting diodes (QLEDs), etc.

[0135] Alternatively, in a specific implementation, if the memory, processor, and communication interface are integrated on a single chip, then the memory, processor, and communication interface can communicate with each other through an internal interface.

[0136] On the other hand, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements the above-mentioned method for pressure ulcer prevention early warning and closed-loop intervention based on multi-physics decoupling. The method includes: S1, physically decoupling the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet via a flexible isolation layer, and acquiring in parallel a set of high-frequency optical raw features in a heterogeneous clock domain. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable S2. When the monitored microenvironment is determined to have reached a steady state, extract the personalized initial physical baseline matrix. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

[0137] This embodiment provides a non-volatile computer-readable storage medium (such as a solid-state drive, read-only memory, or flash memory) on which a computer instruction set is physically burned or stored. When this computer instruction set is read and executed by the processor of the electronic device described in Embodiment 3, all method steps of stages S1 to S4 of this invention can be fully implemented. This embodiment also provides a computer program product, including a computer program, which is encapsulated in the aforementioned integrated intervention and scheduling data packet. Alternatively, in a system upgrade package, when running on the controller of the intelligent pressure ulcer prevention system, the system hardware is equipped to perform baseline dynamic drift reverse compensation, closed-loop verification of intervention effects, and evolution of the clinical intervention case database. Dynamically updated physical capabilities.

[0138] On another front, this application also provides a computer program product, which includes a computer program that can be stored on a computer-readable storage medium. The computer program can execute computer instructions. When the computer program is executed by a processor, the computer can perform the pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling provided by the above methods. This method includes: S1, physically decoupling the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet via a flexible isolation layer, and simultaneously acquiring a set of high-frequency optical raw features in the heterogeneous clock domain. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable S2. When the monitored microenvironment is determined to have reached a steady state, extract the personalized initial physical baseline matrix. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

[0139] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0140] For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit a program for use in or in conjunction with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. Furthermore, a computer-readable medium can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0141] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0143] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0144] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0145] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0146] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for early warning and closed-loop intervention of pressure ulcers based on multi-physics field decoupling, characterized in that, Includes the following steps: S1. The radio frequency photonic sensing subnet and the flexible capacitive sensing subnet are physically decoupled via a flexible isolation layer, and the high-frequency optical raw feature set in the heterogeneous clock domain is acquired in parallel. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable ; S2. When the monitored microenvironment is determined to have reached a steady state, extract the personalized initial physical baseline matrix. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. .

2. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 1, characterized in that, It also includes step S3, as follows: S3, Based on the comprehensive pressure ulcer risk index Match the reconstructed feature variables with the optimal clinical intervention strategy matrix. It outputs signals from each level. Integrated intervention and scheduling data packets Tracking real and effective stress characteristic variables Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. And trigger the re-extraction of the personalized initial physical baseline matrix. The steps.

3. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 1, characterized in that, It also includes step S4, as follows: S4. Based on the timestamp of the intervention action Extracting long-period state sequence sets Quantifying individual tolerance degradation factors ; Based on individual tolerance degradation factors Output updated temperature acceleration coefficient With the updated humidity acceleration factor This involves overwriting the system's underlying parameters and updating the clinical intervention case database based on feedback on intervention effectiveness. .

4. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 1, characterized in that, In step S1, the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet are physically decoupled via a flexible isolation layer, including the following specific steps: Based on the pre-set relative permittivity within the flexible isolation layer Electromagnetic shielding is performed to reduce the parasitic capacitance deviation between the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet. ; Based on the pre-set mechanical buffer modulus within the flexible isolation layer Perform mechanical low-pass filtering to absorb horizontal shear forces generated by dynamic deformation and block optical micro-bending loss; Based on the established global absolute time reference source Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. ,include: Discrete measurements from the heterogeneous clock domain are extracted, and the original high-frequency optical feature set is derived based on an interpolation-extrapolation mapping model. With low-frequency electrical primitive feature set Align timestamp with target Perform virtual feature registration calculation and output virtual feature reconstruction values; Dimensionality reduction and assembly are performed on the virtual feature reconstruction values ​​to generate a high-dimensional synchronous feature matrix. .

5. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 1, characterized in that, In step S2, the personalized initial physical baseline matrix is ​​extracted. And reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Output pressure ulcer comprehensive risk index And generate signals at each level. The specific steps include the following: Obtaining the high-dimensional synchronization feature matrix In the synchronization time window The data sequence within the dataset is used to calculate environmental fluctuation characteristic values ​​based on an environmental fluctuation convergence model. ; When environmental fluctuation characteristics Less than the physical convergence threshold At that time, extract the personalized initial physical baseline matrix. To establish an absolute zero reference for error inversion; Based on a nonlinear thermal conductivity fitting model, the real-time humidity deviation scalar Perform mapping processing to extract pseudo-temperature drift feature scalars ; Based on cross-sensitive reverse compensation logic, utilizing pseudo-temperature drift feature scalar With preset cross sensitivity coefficient For the original temperature variable Compared with the original pressure variable Perform reverse error subtraction to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables To eliminate physical cross-interference caused by drastic changes in the microenvironment; Based on the dynamic damage integral model, the true temperature characteristic variables are... Scalar of deviation from real-time humidity Transformed into microenvironment penalty weight factor And combined with real and effective stress characteristic variables Perform weighted time-series integration to output the pressure ulcer comprehensive risk index. ; Based on the comprehensive risk index of pressure ulcers With dynamic early warning threshold vector Based on the critical alignment results, signals at each level are generated. This will trigger auxiliary intervention actions at the corresponding risk level.

6. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 2, characterized in that, In step S3, the optimal clinical intervention strategy matrix is ​​matched. And track the real and effective stress characteristic variables. Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. The specific steps include the following: Pressure ulcer comprehensive risk index With reconstructing the effective feature matrix Dimensionality reduction and stitching are performed to construct a real-time risk state vector representing the local tissue deterioration state. ; Based on the Mahalanobis distance mapping model, the real-time risk state vector With standard clinical intervention case database The similarity of each historical feature vector is calculated, and the scalar of the intervention plan matching degree is output. To eliminate the interference of linear correlation between multidimensional physical variables; Based on intervention protocol matching scalar The sorting results are extracted and processed to output the optimal clinical intervention strategy matrix. ; Based on time-series differential logic, the true effective pressure characteristic variables are analyzed. Gradient calculations are performed to extract the rate of change of local pressure temporal gradient between adjacent synchronization cycles. ; When the local pressure time gradient change rate Meets the preset threshold for intervention action recognition When conditions are met, perform a log marking operation and output the timestamp of the intervention action. To identify physical intervention events at the software layer; Output intervention action timestamp Subsequently, based on the comprehensive pressure ulcer risk index The rate of change before and after the intervention was calculated to generate a scalar for assessing the effectiveness of the intervention. ; When the scale for evaluating the effectiveness of intervention Meeting the preset effective intervention threshold conditions and real-time humidity deviation scalar When the safe fallback condition is met, execute the underlying reset control flow to trigger the re-extraction of the personalized initial physical baseline matrix. The steps.

7. The pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling according to claim 3, characterized in that, In step S4, the individual tolerance degradation factor is quantified. And update the database of evolving clinical intervention cases. The specific steps include the following: Based on the time stamp of the intervention action Perform segmentation and concatenation processing on the time series data to generate a long-period state sequence set. ; Based on long-period state sequence sets The cumulative slope of the risk index during the resting period is extracted, and a degradation quantification model is used to map the cumulative slope of the risk index to output the individual tolerance degradation factor. To quantitatively characterize the decline state of individual skin microcirculation; Based on the logarithmic optimization model, using individual tolerance degradation factors The system's preset base acceleration coefficient is nonlinearly amplified to output the updated temperature acceleration coefficient. With the updated humidity acceleration factor To adaptively tighten the warning tolerance of the system; Based on a reinforcement learning reward model, and utilizing a scalar for evaluating intervention effectiveness. Optimal clinical intervention strategy matrix Associated policy confidence reward scalar Perform weight update processing; Scalar of reward based on policy confidence The dynamic ranking results output evolves into a clinical intervention case database. This allows for the self-iteration of the clinical intervention strategy library.

8. A pressure ulcer prevention early warning and closed-loop intervention device based on multi-physics field decoupling, characterized in that, include: The heterogeneous multimodal sensing signal underlying physical acquisition and spatiotemporal synchronization fusion module is used to physically decouple the radio frequency photonic sensing subnet and the flexible capacitive sensing subnet via a flexible isolation layer, and to acquire in parallel the high-frequency optical raw feature set in the heterogeneous clock domain. With low-frequency electrical primitive feature set Based on the established global absolute time reference source; Synchronization frequency with reference For the original set of high-frequency optical features With low-frequency electrical primitive feature set Perform dynamic timestamp alignment compensation to generate a high-dimensional synchronization feature matrix. High-dimensional synchronization feature matrix Including raw pressure variables Original temperature variable Compared with the original humidity variable ; A baseline dynamic drift reverse compensation and risk calculation module based on a multiphysics coupling model is used to extract a personalized initial physical baseline matrix when the monitored microenvironment is determined to have reached a steady state. Personalized initial physical baseline matrix Including initial humidity baseline ; Calculate the original humidity variable Compared with the initial humidity baseline Real-time humidity deviation scalar Based on real-time humidity deviation scalar Extracting pseudo-temperature drift feature scalar Based on the pseudo-temperature drift characteristic scalar For the original temperature variable Compared with the original pressure variable Cross-sensitivity inverse compensation is performed to reconstruct the true temperature characteristic variables. Compared with real effective stress characteristic variables Based on real and effective pressure characteristic variables True temperature characteristic variables and real-time humidity deviation scalar Perform kinetic damage integration to output a comprehensive pressure ulcer risk index. And generate signals at each level. . The intelligent matching and closed-loop verification module for clinical intervention strategies is used to verify the effectiveness of interventions based on the comprehensive pressure ulcer risk index. Match the reconstructed feature variables with the optimal clinical intervention strategy matrix. It outputs signals from each level. Integrated intervention and scheduling data packets The intelligent matching and closed-loop verification module for clinical intervention strategies is also used to track real and effective stress characteristic variables. Local pressure time gradient change rate Mutations mark the timestamps of intervention actions. And trigger the re-extraction of the personalized initial physical baseline matrix. Steps; The module for dynamic updating of the personalized prediction model evolution and intervention knowledge base is used to update the knowledge base based on the timestamps of intervention actions. Extracting long-period state sequence sets Quantifying individual tolerance degradation factors The module for dynamic updating of the individualized prediction model evolution and intervention knowledge base is also used to analyze individual tolerance degradation factors. Output updated temperature acceleration coefficient With the updated humidity acceleration factor This involves overwriting the system's underlying parameters and updating the clinical intervention case database based on feedback on intervention effectiveness. .

9. A computer device comprising at least one processor coupled to at least one memory storing at least one computer program or instruction, characterized in that, The computer program or instructions are loaded and executed by the processor to implement the steps of the pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program or instructions, which, when executed by a processor, implement the steps of the pressure ulcer prevention early warning and closed-loop intervention method based on multi-physics field decoupling as described in any one of claims 1 to 7.

Images