Burn wound infection risk real-time early warning method based on multi-modal sensing

Abstract

The present application relates to the technical field of physiological monitoring, in particular to a burn wound infection risk real-time early warning method based on multi-modal sensing, comprising the following steps: obtaining oxygen pressure data to extract mutation area, fusing exudation path and thermal gradient data to identify unstable blocks, combining pH and oxygen pressure characteristics to locate hypoxic segments, screening three types of signal overlapping fragments, dividing early warning levels and outputting infection state labels. In the present application, a multi-period continuous directional analysis chain is constructed in the oxygen pressure change trend identification, the amplitude mutation and trend reversal judgment logic are superimposed, the wound oxygen supply abnormal space fragments are accurately captured, the coupling judgment condition of thermal gradient change and exudation trajectory is introduced in the spatial feature screening, the acid-oxygen covariant trend fragments are identified by fusing the double conditions of pH value downward feature and oxygen pressure fluctuation stability, and the wound infection risk grading ability with high timeliness, strong objectivity and clear structure is achieved, which improves the identification accuracy and clinical early warning efficiency of pathological changes.

Description

Technical Field

[0001] This invention relates to the field of physiological monitoring technology, and in particular to a method for real-time early warning of infection risk in burn wounds based on multimodal sensing. Background Technology

[0002] The field of physiological monitoring technology involves real-time or timed monitoring of human vital signs, physiological states, and pathological changes through sensing methods to achieve health status assessment, disease risk identification, and intervention support. Core aspects include physiological parameter acquisition, biosignal monitoring, pathological change identification, and health risk early warning, covering the entire process from single-sensor data acquisition to multi-parameter integrated analysis. Physiological monitoring plays a crucial role in clinical nursing, rehabilitation management, and disease prevention and control. Its technological system relies on a cluster of sensors of various types, including thermal, electrochemical, and optical sensors, combined with continuous monitoring of target areas to achieve real-time perception of changes in human health status and provide data support for clinical intervention. Traditional methods for identifying the risk of burn wound infection rely on the visual judgment of medical staff regarding changes in the wound's appearance and laboratory bacterial culture results during clinical nursing. These methods involve nurses observing wound color and exudate characteristics to record the wound's condition, and analyzing bacterial samples for the presence of infectious agents such as Pseudomonas aeruginosa and Staphylococcus aureus. The culture process takes 24 to 72 hours. In infection monitoring, traditional methods rely heavily on subjective judgment, such as observing visual indicators like redness and swelling for preliminary assessment, lacking real-time accuracy and objectivity. Regarding wound healing assessment and nursing management, traditional nursing models primarily rely on manual methods for recording wound status, consuming significant time for nursing staff, and hindering standardized and systematic data management. This patent addresses these clinical needs by integrating multimodal sensing technologies and real-time collection of objective parameters such as wound temperature, humidity, pH, and specific biomarker concentrations to support dynamic assessment and early warning of burn wound infection risk.

[0003] Existing infection risk identification methods, which rely primarily on manual observation and supplemented by laboratory culture, have significant shortcomings in the management of rapidly changing burn wounds. Visual judgment depends on experience and is difficult to quantify specific conditions, especially when wound exudate is weak or color changes are not obvious, making it difficult to accurately determine infection tendencies. Laboratory culture takes a long time and cannot meet the real-time early warning needs when wound pathology changes rapidly. In addition, data recording is mainly done manually and lacks a standardized parameter system, resulting in a large workload for nursing staff, inconsistent recording results, difficulty in continuous management and trend judgment, and affecting the timing of recovery and the deployment of intervention strategies. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a real-time early warning method for burn wound infection risk based on multimodal sensing, comprising the following steps: To achieve the above objectives, the present invention adopts the following technical solution: a real-time early warning method for burn wound infection risk based on multimodal sensing, comprising the following steps: S1: Acquire the time series data recorded by the oxygen pressure sensing units deployed in the burn wound area, and compare the change amplitude and direction continuity of the oxygen pressure values ​​between adjacent time periods according to the change direction of the location point in the time periods before and after the location point to generate the location result of the oxygen pressure trend change area. S2: Based on the spatial location range marked in the oxygen pressure trend change area location result, retrieve the exudate path information recorded by the pressure sensing point of the wound area and the thermal gradient data recorded by the thermal detection point, and perform corresponding position overlap analysis with the temperature distribution curve to generate the exudate-thermal synergistic instability block identification result. S3: Based on the spatial region coordinates in the identification results of the infiltration-thermal synergistic instability block, determine whether there is a synchronous feature in the block where the pH value continues to decline and the oxygen pressure value fluctuates within a stable range, and generate the acid-oxygen synergistic hypoxia performance segment results. S4: Based on the coordinates of the monitoring units in the results of the acid-oxygen synergistic hypoxia manifestation segment, superimpose the spatial overlap markers of whether the comparison area includes the location results of the oxygen pressure trend change area and the identification results of the infiltration-thermal synergistic instability block to generate a multi-source early warning intersection tag set.

[0005] As a further aspect of the present invention, the oxygen pressure trend abrupt change region location results include direction reversal location points, amplitude abrupt change characteristic areas, and trend reversal segments; the percolation-thermal co-instability block identification results include percolation boundary abrupt change segments, thermal gradient abnormal contraction segments, and percolation-thermal synchronous abnormal blocks; the acid-oxygen co-hypoxia manifestation segment results include pH value continuously declining segments, oxygen pressure fluctuation stable segments, and acid-oxygen trend co-existing segments; and the multi-source early warning intersection label set includes three types of signal overlap segments, signal overlap quantity distribution, and signal overlap spatial range.

[0006] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Acquire the time series data recorded by the oxygen pressure sensing unit deployed in the burn wound area, and based on the difference in oxygen pressure values ​​in adjacent time periods, determine the points where the value changes from rising to falling or from falling to rising, filter the time sequence positions where the change direction switches, and generate a sequence of oxygen pressure change direction conversion points. S102: Based on the oxygen pressure change direction conversion point sequence, extract the amount and direction of oxygen pressure change before and after each location point in time intervals, calculate the difference in the magnitude and direction of the change in the value between the two time intervals, and filter out the conversion point locations where the direction reverses and the magnitude of the change in the value jumps, thereby generating a set of oxygen pressure trend change feature points. S103: Call the spatial location information in the set of oxygen pressure trend change feature points, aggregate the coordinate points of the sensing units to which the feature points belong, identify the continuously distributed or densely concentrated spatial regions, record the distribution range of the points with trend change features, and obtain the location result of the oxygen pressure trend change region.

[0007] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the spatial location range marked in the oxygen pressure trend change area location result, retrieve the seepage path information recorded by the pressure sensing point and the thermal gradient data recorded by the thermal sensing point in the corresponding area, extract the temporal position change trajectory of the seepage boundary in the seepage path within the same time period, and simultaneously extract the corresponding spatial position of the thermal gradient data to generate a set of seepage boundary change trajectory and temperature distribution curves. S202: Call the boundary trajectory and temperature change trend of the spatial location points in the set of seepage boundary change trajectory and temperature distribution curves, compare the two types of data in the same spatial segment, identify the overlapping segments where the boundary changes abruptly and the temperature change shows a contraction trend, filter the spatial range with dual change characteristics, and obtain the identification result of seepage-thermal synergistic instability block.

[0008] As a further aspect of the present invention, the pressure sensing point records the pressure value at its location in real time, monitors the dynamic changes in the pressure of seepage or gas, and infers the liquid seepage path. The seepage path information refers to the path information formed when the fluid migrates in the medium, including the flow direction, boundary position, and velocity change.

[0009] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the spatial region coordinates in the identification result of the heat permeation synergistic instability block, call the pH value data and oxygen pressure record of the corresponding block for the continuous time period, determine whether the pH value change direction remains downward, extract the range of oxygen pressure value change in the corresponding time period and determine whether the fluctuation amplitude remains in the stable range, and generate a set of acid-base and oxygen pressure state trend features. S302: Call the pH value downward persistence and oxygen pressure fluctuation range stability of spatial location points in the acid-base and oxygen pressure state trend feature set, determine whether they are simultaneously true within the same time period, identify continuous time periods and spatial ranges with a common trend of pH value decrease and oxygen pressure stability, and obtain the results of acid-oxygen synergistic hypoxia performance segments.

[0010] As a further aspect of the present invention, the stability of the oxygen pressure fluctuation range refers to whether the fluctuation range of the oxygen pressure value is stable within a certain period of time. The acid-oxygen synergistic hypoxia manifestation zone is a region in which the pH value continuously decreases and the oxygen pressure fluctuates stably at the same time and spatial location.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the coordinates of the monitoring units in the results of the acid-oxygen synergistic hypoxia manifestation segment, the spatial coordinates are superimposed and compared with the location results of the oxygen pressure trend change area and the identification results of the infiltration-thermal synergistic instability block. Monitoring units with completely consistent positions in the three types of results are filtered out, and coordinate segments with spatial overlap characteristics of the three types of signals are extracted to generate a set of overlapping segments of the three sources of signals: S402: Call the coordinates of the monitoring units in the set of overlapping segments of the three-source signals, count the distribution positions and total number in the overall monitoring area, mark the combination layout features in the spatial plane, and perform coordinate numbering and classification based on the marking results to obtain the set of multi-source early warning intersection labels.

[0012] As a further aspect of the present invention, the method further includes step S5: S5: Based on the signal coverage type, quantity and spatial coverage density characteristics in the multimodal signal fusion tag set, according to the joint distribution hierarchy of the three types of characteristic signals, divide the risk level corresponding to the early warning intersection tag set, and mark the early warning level status of the tag set with different degrees of differentiation, and output the wound infection early warning level status tag. The wound infection early warning level status label includes risk level identifier, label differentiation status, and joint distribution hierarchy.

[0013] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the tags in the multi-source early warning intersection tag set, extract the total number of corresponding signal coverage types, the number of signal points contained, and the spatial coverage density value. Generate a fused signal coverage feature table based on the combination of signal types, the distribution of signal quantity, and the density of spatial distribution. S502: Call the fused signal coverage feature table, construct the distribution hierarchy of multimodal features according to the signal type joint occurrence level, signal quantity ratio range and spatial coverage density sorting, define the differentiated risk level according to the structural hierarchy, and generate a risk level table corresponding to the label distribution hierarchy. S503: Based on the risk level results of the label items in the risk level table corresponding to the label distribution hierarchy, the label set with differentiated level division is identified by status distinction, the level of the warning status is marked, the level status label is assigned according to the risk level, and the wound infection warning level status label is output.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by constructing a multi-time-period continuous directional analysis chain in the identification of oxygen pressure change trends, and superimposing judgment logic for amplitude abrupt changes and trend reversals, abnormal spatial segments of wound oxygen supply are accurately captured. In the spatial feature screening, the coupling judgment condition of thermal gradient change and exudate trajectory is introduced to effectively locate the dual instability area of ​​exudation and heat. By integrating the dual conditions of pH value downward characteristics and oxygen pressure fluctuation stability, acid-oxygen co-variation trend segments are identified. Multi-signal intersection areas are extracted to form a label set, and a signal distribution hierarchy model is constructed to classify risk levels. It has the ability to classify wound infection risk with high timeliness, strong objectivity, and clear structure, avoiding the problems of strong subjectivity, slow response and difficulty in data standardization in traditional methods, and improving the accuracy of identification of pathological changes and the efficiency of clinical early warning. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1 This invention provides a method for real-time early warning of burn wound infection risk based on multimodal sensing, comprising the following steps: S1: Obtain the time series data recorded by the oxygen pressure sensing units deployed in the burn wound area. Based on the direction of change of oxygen pressure values ​​between adjacent time periods, extract the location points where the direction changes. Compare the change amplitude and direction continuity between the time periods before and after the location points in turn. Filter the spatial location areas with abrupt change in amplitude and trend reversal characteristics in the trend change and generate the oxygen pressure trend change area location result. S2: Based on the spatial location range marked in the oxygen pressure trend change area location results, retrieve the exudate path information recorded by the pressure sensing point of the wound area and the thermal gradient data recorded by the thermal detection point, extract the exudate boundary change trajectory within the same time period and perform corresponding position overlap analysis with the temperature distribution curve, screen out spatial segments that simultaneously exhibit exudate boundary change and abnormal thermal gradient contraction, and generate the exudate-thermal synergistic instability block identification results. S3: Based on the spatial region coordinates in the identification results of the infiltration-thermal synergistic instability block, call the continuous time period data of pH value and oxygen pressure records in the spatial region, determine whether there is a synchronous feature of continuous decline in pH value and stable fluctuation range of oxygen pressure value in the block, verify whether the two signals show a common trend and maintain a consistent state, and generate the acid-oxygen synergistic hypoxia performance segment results. S4: Based on the coordinates of the monitoring units in the results of the acid-oxygen co-hypoxia performance segment, superimpose and compare the spatial overlap markers of whether the region includes the location results of the oxygen pressure trend change area and the identification results of the infiltration-thermal co-instability block, extract spatial segments with the overlapping distribution characteristics of three types of signals, summarize the location distribution and quantity characteristics in the monitoring range, and generate a multi-source early warning intersection label set. S5: Based on the signal coverage type, quantity and spatial coverage density characteristics in the multimodal signal fusion tag set, according to the joint distribution hierarchy of the three types of characteristic signals, divide the risk level corresponding to the early warning intersection tag set, and mark the early warning level status of the tag set with different degrees of differentiation, and output the wound infection early warning level status tag. The results of the oxygen pressure trend change area location include the direction change location point, amplitude change characteristic area, and trend reversal segment; the results of the instability block identification of the osmotic heat synergy include the exudate boundary change segment, the thermal gradient abnormal contraction segment, and the osmotic heat synchronous abnormal block; the results of the acid-oxygen synergy hypoxia manifestation segment include the pH value continuously declining segment, the oxygen pressure fluctuation stable segment, and the acid-oxygen trend synergy segment; the multi-source early warning intersection label set includes three types of signal overlap segments, the signal overlap quantity distribution, and the signal overlap spatial range; the wound infection early warning level status label includes the risk level identifier, the label differentiation status, and the joint distribution level division; S1: Based on a miniature oxygen sensor (such as a solid-state electrochemical sensor or oxygen concentration probe), it monitors the oxygen pressure in the wound area and provides real-time oxygen concentration data. The sensor has high response speed and stability and is suitable for accurate detection in low-oxygen environments. S2: Based on pressure sensors (such as thin-film pressure sensors or piezoelectric sensors), monitor the fluid permeation in the wound area and record the exudate path and boundary changes; use infrared temperature sensors (such as thermocouple sensors or thermistors) to monitor the temperature distribution in the wound area in real time and record thermal gradient data (temperature gradient information). S3: It adopts a glass electrode type pH sensor, which can monitor the changes in pH of the wound area in real time. The continuous downward trend of pH value is due to metabolic changes in the local hypoxic environment. S4: By analyzing the spatial overlap of oxygen pressure, pH value, pressure and temperature, the overlapping regions of signals are extracted, further refining the spatial location of anomalies in the dataset; S5: Based on fused multimodal sensor data (oxygen, pH, pressure, temperature), a risk assessment model is constructed. Combined with machine learning algorithms (such as cluster analysis or hierarchical analysis), the characteristics of spatial segments are classified, and a wound infection warning level status label is output.

[0023] Please see Figure 2 The specific steps of S1 are as follows: S101: Acquire the time series data recorded by the oxygen pressure sensing unit deployed in the burn wound area, and based on the difference in oxygen pressure values ​​in adjacent time periods, determine the points where the value changes from rising to falling or from falling to rising, filter the time sequence positions where the change direction switches, and generate a sequence of oxygen pressure change direction conversion points. In clinical treatment of burn wounds, multiple oxygen pressure sensing units are deployed to monitor the oxygenation status of local tissues. Sensors are evenly spaced and fixed to the wound surface, continuously recording time-series data of local tissue oxygen pressure through an adhesive sensing module. The acquired data is sequentially ordered along a time axis. The program reads the oxygen pressure change trend of each sensor at different time points, constructing a one-dimensional sequence along the time axis. Oxygen pressure values ​​between adjacent time points are continuously compared, determining whether the trend is upward or downward. For example, in a time series, if the sensor's oxygen pressure value rises from 41 to 43 between the 5th and 6th second, it is considered an upward trend. If the value changes to 40 in the next second, it is considered a shift from upward to downward. This process is repeated continuously, recording the time points where the trend changes from upward to downward or vice versa. In a practical example, one sensor records 120 data points per minute. After processing, approximately 15 time points with trend reversals can be identified, representing significant changes in the direction of the oxygen pressure curve, generating a sequence of oxygen pressure change direction reversal points.

[0024] S102: Based on the sequence of oxygen pressure change direction conversion points, extract the amount and direction of oxygen pressure change before and after each location point in time intervals, calculate the difference in the magnitude and direction of the change in the value between the two time intervals, and filter out the conversion point locations where there is a jump in the magnitude of the change in the value while the direction reverses, generating a set of oxygen pressure trend change feature points. The oxygen pressure change characteristics before and after a given time point are analyzed to determine if there are any sudden trend anomalies. In the operation, for each directional change point, a period of time before and after it is selected as the analysis window. The 5-second time intervals before and after the change point are used as references to extract the change amplitude of the oxygen pressure value within these two time intervals and determine whether the directional change is opposite. For example, if the oxygen pressure rises from 35 to 45 in the 5 seconds before the change point, and then drops sharply from 45 to 32 in the next 5 seconds, such cases with opposite directions and large differences in amplitude are marked as abrupt changes. For each change point, the context data segment needs to be extracted and compared. Points that not only have directional reversal characteristics but also show obvious fluctuations at the numerical level are gradually filtered out. In the actual operation, a sensor detects 8 directional change points within one minute, but only 3 of them have a change amplitude exceeding a certain threshold and opposite directions in the time intervals before and after them. These 3 points are selected and included in the oxygen pressure trend abrupt change feature point set.

[0025] S103: Call the spatial location information in the set of oxygen pressure trend change feature points, aggregate the coordinate points of the sensing units to which the feature points belong, identify the continuously distributed or densely concentrated spatial regions, record the distribution range of the points with trend change features, and obtain the location result of the oxygen pressure trend change region. Each point corresponds to the location coordinates of a sensor. By extracting the coordinate information, a spatial distribution map of trend mutations can be constructed within the entire burn wound area. During operation, multiple mutation feature points belonging to the same area need to be clustered to determine whether there is a relatively dense distribution pattern. In practice, the entire wound can be divided into several small areas, each covering a portion of the sensing units. For example, the wound can be divided into multiple 4cm×4cm grid areas, and the number of mutation points contained in each area can be counted. If multiple mutation points appear in a certain area and the spatial distance between the points is relatively close, then the area is marked as a trend mutation area. This process can be automatically identified by clustering algorithms or by manually analyzing the grid area to make a judgment on a case-by-case basis. If, in a single detection, all four sensors in a certain area detect mutation points, but a similar distribution does not appear in part of the area, then the coordinates of that area are marked as mutation cluster areas to obtain the location results of oxygen pressure trend mutation areas.

[0026] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the spatial location range marked in the oxygen pressure trend change area location results, retrieve the seepage path information recorded by the pressure sensing point and the thermal gradient data recorded by the thermal sensing point in the corresponding area, extract the temporal position change trajectory of the seepage boundary in the seepage path within the same time period, and simultaneously extract the corresponding spatial location of the thermal gradient data to generate a set of seepage boundary change trajectory and temperature distribution curves. It is necessary to access the data records of pressure sensing points and thermal detection points already deployed within the corresponding area of ​​the wound. The sampling data from these two different types of sensors within the same time period are processed synchronously to extract the exudate path information recorded in the pressure sensing points. In practice, the exudate path manifests as a pressure change area expanding outward over time. By continuously identifying changes in the boundary pressure threshold, the trajectory of the exudate boundary is generated. If the boundary of a certain area extends from its initial position to the position of an adjacent sensing point within 10 seconds, it is considered a boundary shift. This type of temporal boundary position data is continuously recorded as a trajectory line. The thermal gradient data collected by the corresponding thermal detection points also needs to be read synchronously. The thermal gradient information reflects the spatial gradient difference in wound temperature distribution. The extraction method converts the temperature data of the thermal sensing points into spatial temperature curves of isochronous slices, forming a set of exudate boundary position trajectories and a set of temperature distribution curves within the same time range. Through the synchronous extraction and numbering management of these two types of data, a unified spatial and temporal dimension structure is ensured for subsequent comparisons, generating a set of exudate boundary change trajectories and temperature distribution curves.

[0027] S202: Call the boundary trajectory and temperature change trend of spatial location points in the set of seepage boundary change trajectory and temperature distribution curves, compare the two types of data side by side in the same spatial segment, identify the overlapping segment where the boundary changes abruptly and the temperature change shows a contraction trend, filter the spatial range with dual change characteristics, and obtain the identification results of seepage-thermal synergistic instability block. For each spatial location, it is necessary to perform a parallel comparison of the boundary change trend and temperature change trend within the same time period. In the operation, each seepage boundary trajectory segment within the spatial segment is matched one-to-one with the temperature change curve within its coverage area. Segments with obvious jumps in seepage boundary are identified as analysis targets. If the boundary of a certain area suddenly extends outward from 10mm to 18mm within two seconds, it is considered a jump phenomenon. At the same time, the temperature curve of that area shows a contraction in temperature value distribution, that is, the high temperature area shrinks or the central temperature decreases. If the original temperature peak area shrinks from a coverage diameter of 20mm to 14mm, or the central temperature drops from 42℃ to 36℃, then the spatial segment is judged to have both boundary jump and temperature contraction characteristics. Such spatial segments are screened and classified as seepage-thermal co-instability blocks. This processing method classifies and summarizes the comparison results, extracts the spatial areas that meet the dual change criteria, and numbers and records them to provide structured input data for spatial modeling and recovery response, and obtains the identification results of seepage-thermal co-instability blocks.

[0028] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the spatial coordinates of the instability block identification results, call the pH data and oxygen pressure records of the corresponding block for continuous time periods, determine whether the pH change direction remains downward, calculate the pH continuous decreasing trend characteristic value, extract the range of oxygen pressure value changes in the corresponding time period and determine whether the fluctuation amplitude remains within the stable range, and generate a set of acid-base and oxygen pressure state trend characteristics. The characteristic value of the continuous decrease in pH is expressed by the formula: ; in, Values ​​representing a continuous decreasing trend in pH Represents the total number of pH sampling points. Representing the The pH value at that time. This represents the pH value at the previous moment. represents the dimensionless offset value, and F represents the range of oxygen pressure value changes within the corresponding time period; Formula calculation logic: The degree of regional water acidification is reflected by judging the direction and magnitude of pH value differences over a continuous period. In a given time series, the pH value difference between each pair of adjacent time points is calculated item by item, and the magnitude of change is obtained using the absolute value. At the same time, the direction of the difference is used to retain the directionality of the downward trend, and a minimum value is introduced. To avoid the denominator being undefined when the difference is zero, normalization is used to ensure numerical stability. The sum is standardized using the square root function to ensure that the number of samples in different time periods has a consistent impact on the calculation results. The overall process integrates two types of indicators: difference intensity and trend direction, so that a stable downward trend can still be identified even when pH data fluctuates. The pH value of a continuous downward trend indicates whether the pH value of groundwater generally shows a continuous downward trend over a period of time. It combines the magnitude and direction of pH value changes to quantitatively describe the downward trend. If the value is less than zero, it means that the pH value tends to become acidic in a significant and continuous manner during that period. How to obtain each parameter and the actual value setting: Data is collected via on-site online sensors every 10 minutes. 6.85 (Example) A high-precision pH monitoring electrode is used, with a fixed time interval (e.g., 10 minutes) for each sampling cycle to collect real-time values. Previous sampling period data 6.89 (Example) and One-to-one correspondence, forming a continuous sequence of changes. Calculate the number of sample pairs within the interval 5 If the monitoring duration is 60 minutes, with 10-minute intervals, then there will be a total of 6 points and 5 sets of data differences. Determined based on equipment accuracy 0.01 Set the instrument's minimum resolution to 0.01 units of pH. Explanation of parameter numerical and non-numerical processing methods: Non-numerical pH trend data (such as "decreasing" or "increasing") have been converted into numerical expressions in the formula through structural inlining: using ; It achieves sign normalization for +1 (increase) and -1 (decrease), which is applicable to subsequent numerical summation; If a continuous decrease occurs within a certain interval (e.g., from 6.89→6.85→6.80→6.77→6.72→6.70), the formula will obtain a negative consistent trend judgment; Threshold setting instructions: The criteria for determining whether something is a "continuous decline" are set as follows: ; ; The threshold here is 0, and the logic is set based on the sign. Complete the calculation by substituting actual data: The pH data for a certain monitoring segment is set as follows (unit: dimensionless): Table 1: pH change sequence at monitoring points 1 6.85 6.89 -0.04 2 6.80 6.85 -0.05 3 6.77 6.80 -0.03 4 6.72 6.77 -0.05 5 6.70 6.72 -0.02 As shown in Table 1, this data segment exhibits a clear downward trend, calculated as follows: ; ; The corresponding weighted terms are ; Following this logic, the calculations for the 5 sets of data are as follows: : #imgpt24#; The summation result is: #imgpt25#; After normalization: #imgpt26#; The results showed that the characteristic value of the continuous downward trend of pH was -0.0687, which is lower than 0, indicating that there is a continuous downward trend in the current monitoring block, indicating that the judgment is passed. The pH trend continues to decline. The fluctuation amplitude of oxygen pressure is analyzed to determine whether it is in an abnormal synergistic trend. The advantages of the formula are: by introducing an absolute value and directionality fusion structure, combined with the denominator normalization mechanism of minimal perturbation, the formula can stably extract trend direction information in continuous data with large measurement noise or frequent fluctuations. The fusion weighted product and square root normalization design can be adapted to different sampling densities, significantly improving the sensitivity and robustness of trend recognition.

[0029] S302: Call the pH value downward persistence and oxygen pressure fluctuation range stability of spatial location points in the acid-base and oxygen pressure state trend feature set, determine whether they are simultaneously true within the same time period, identify continuous time periods and spatial ranges with the common trend of pH value decrease and oxygen pressure stability, and obtain the results of acid-oxygen synergistic hypoxia performance segments. By re-analyzing the spatial point information in the set of acid-base and oxygen pressure state trend characteristics, it is necessary to determine whether the downward trend of pH value is continuous and whether the oxygen pressure fluctuation range is stable, and to confirm whether the two state characteristics exist simultaneously within the same time period. During execution, acid-base trend data and oxygen pressure fluctuation are read according to the sensor point number, and the two types of data are aligned and compared along the time axis. A scenario is set where the pH value at a certain point continuously decreases from t=0 to t=20 seconds, and the oxygen pressure fluctuation remains within a set stable range from t=5 to t=20 seconds. It is necessary to determine whether the two characteristics overlap. Whether the overlapping time period meets the continuity requirement is determined. If both data points maintain the corresponding state within t=5 to t=20 seconds, the interval can be identified as the time period in which the acid-oxygen synergy characteristic is established, and the corresponding spatial location is marked. Such continuous spatial point segments that simultaneously exhibit pH decrease and oxygen pressure stability are extracted as acid-oxygen synergy hypoxia performance segments. During the processing, each segment that meets the conditions will be recorded as a unit structure, including start and end time, sensor number, spatial coordinates, and trend feature description. The results of acid-oxygen synergy hypoxia performance segments are obtained through point-by-point discrimination and multi-block classification.

[0030] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the coordinates of monitoring units in the results of the acid-oxygen synergistic hypoxia manifestation segment, the spatial coordinates of the monitoring units with the results of the oxygen pressure trend change area location and the results of the infiltration-heat synergistic instability block identification are superimposed and compared with those of the results of the oxygen pressure trend change area location and the results of the infiltration-heat synergistic instability block identification are selected. The monitoring units with completely consistent positions in the three types of results are screened, and the coordinate segments with the spatial overlap characteristics of the three types of signals are extracted to generate a set of overlapping segments of the three sources of signals. The coordinate data of the monitoring units need to be compared item by item with the coordinate data recorded in the location results of the oxygen pressure trend change area and the identification results of the seepage and heat instability block. During the execution, a spatial index table is established for each of the three types of coordinate results, and a complete matching judgment is made for the coordinates of each monitoring unit. During the processing, the accurate coordinate value is used as the reference to compare whether there are completely identical coordinate segments in the three sets. If a certain monitoring point appears in the three types of results of acid-oxygen synergy, oxygen pressure change, and seepage and heat instability, it is identified as a point with spatial overlap characteristics of the three types of signals. This point is extracted as an effective overlap point and added to the result set. In actual operation, the coordinates of the monitoring units in the acid-oxygen-deficiency set are traversed, and each item is checked to see if it also exists in the two sets. If there is a coordinate (32mm, 44mm) with the number A17 in the acid-oxygen segment, if this point also exists in the oxygen pressure trend change and seepage and heat instability results, then A17 is the overlapping segment of the three-source signals. The coordinate segments that meet this spatial overlap condition are uniformly organized and archived to generate a set of overlapping segments of the three-source signals.

[0031] S402: Call the coordinates of the monitoring units in the set of overlapping segments of three-source signals, count the distribution and total number in the overall monitoring area, mark the combination layout features in the spatial plane, number and classify the coordinates according to the marking results, and obtain the set of multi-source early warning intersection labels; It is necessary to statistically analyze the spatial distribution and total number of coordinate points in the entire wound monitoring area. The processing method divides the entire monitoring area into a standardized grid structure, set according to a 10mm×10mm grid. The number of three-source overlapping points contained in each grid is counted one by one, and spatial annotation is performed on a two-dimensional plane map. The layout position of each three-source overlapping segment is marked using different colors or symbols. It is extracted whether it exhibits clustering, dispersion, or linear distribution characteristics in the overall structure. At the same time, a unique number is assigned to each overlapping segment in the plane coordinate system. The numbers are set according to the row and column number rules and named T1, T2, etc. Then, the overlapping points are classified according to the number. Each label contains the number, the coordinate position, the grid number, and information on the three associated signal types to support multi-dimensional processing of spatial distribution analysis and severity classification, and obtain a set of multi-source early warning intersection labels.

[0032] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the tags in the multi-source early warning intersection tag set, extract the total number of corresponding signal coverage types, the number of signal points contained, and the spatial coverage density value. Based on the combination of signal types, the distribution of signal quantity, and the density of spatial distribution, generate a fused signal coverage feature table. It is necessary to extract the signal type associated with each tag, the number of signal points it contains, and the spatial coverage density within the monitoring area. Each tag is read to identify the three signal types it covers: abrupt changes in oxygen pressure trends, synergistic instability of percolation and heat, and synergistic hypoxia of acid and oxygen. The number of signal points contained in each tag is counted. For example, a tag number T3 contains 3 oxygen pressure points, 2 percolation points, and 4 acid and oxygen points, totaling 9 signal points. The spatial coverage density is calculated based on its area within the monitoring area, i.e., the number of signal points per unit area within the coverage area of ​​that point. For example, 9 points are contained within a 20mm × 20mm area. The same statistical operation is performed on all tags to form a set of structured data, indexed by the tag number, recording the signal combination type, signal quantity, and density value. Based on the combination method of signal types (e.g., whether all three types are present, or whether one type is missing), the high and low distribution of signal quantity, and the concentration or dispersion characteristics of density, combination features are extracted. These parameters are used as indicators to form a fusion signal coverage feature scale.

[0033] S502: Call the fused signal coverage feature scale, construct the distribution hierarchy of multimodal features based on the joint occurrence level of signal type, the proportion range of signal quantity, and the spatial coverage density sorting, define the differentiated risk level according to the structural hierarchy, and generate a risk level table corresponding to the label distribution hierarchy. The process involves hierarchically classifying tags and establishing a multimodal feature distribution hierarchy based on the combined occurrence of signal types, the proportion of signal quantity, and the spatial density. During the process, a joint signal hierarchy is defined: tags possessing three signal types are classified as the first level, tags possessing any two types as the second level, and tags with only one signal type as the third level. Based on the proportion of signal quantity recorded in the tags, tags with signal quantities above the overall average are classified as high-density, while those below the average are classified as low-density. This is then combined with the specific values ​​of spatial coverage density to construct a system of coverage intensity from high to low. The risk level is comprehensively assessed using these three dimensions: hierarchy, quantity, and density, defining high, medium, and low risk levels. Each tag is assigned to a corresponding risk level based on its performance indicators, forming a clearly defined tag distribution hierarchy and risk level matching relationship. This serves as an important classification basis for judging the regional infection development trend, generating a risk level table corresponding to the tag distribution hierarchy.

[0034] S503: Based on the risk level results of the label items in the risk level table corresponding to the label distribution hierarchy, the status of the label set with differentiated level division is distinguished and identified, the level of the warning status is marked, the level status label is assigned according to the risk level, and the wound infection warning level status label is output. Based on the risk level results of each label item, the members in the label set are classified and labeled according to the level differences. During the execution process, labels with high risk levels are marked as Level I warning status, medium level as Level II warning status, and low level as Level III warning status. The labels are directly visualized through color coding, layer overlay, or label annotation. A risk level field is added to each number in the label database, and the warning status code is recorded. This data is directly used as the data source for the diagnosis and treatment interface to intuitively identify and dynamically track the risk level of lesion areas. This allows different risk segments to be treated and managed differently in subsequent intervention strategies, and outputs wound infection warning level status labels.

[0035] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for real-time early warning of burn wound infection risk based on multimodal sensing, characterized in that, Includes the following steps: S1: Acquire the time series data recorded by the oxygen pressure sensing units deployed in the burn wound area, and compare the change amplitude and direction continuity of the oxygen pressure values ​​between adjacent time periods according to the change direction of the location point in the time periods before and after the location point to generate the location result of the oxygen pressure trend change area. S2: Based on the spatial location range marked in the oxygen pressure trend change area location result, retrieve the exudate path information recorded by the pressure sensing point of the wound area and the thermal gradient data recorded by the thermal detection point, and perform corresponding position overlap analysis with the temperature distribution curve to generate the exudate-thermal synergistic instability block identification result. S3: Based on the spatial region coordinates in the identification results of the infiltration-thermal synergistic instability block, determine whether there is a synchronous feature in the block where the pH value continues to decline and the oxygen pressure value fluctuates within a stable range, and generate the acid-oxygen synergistic hypoxia performance segment results. S4: Based on the coordinates of the monitoring units in the results of the acid-oxygen synergistic hypoxia manifestation segment, superimpose the spatial overlap markers of whether the comparison area includes the location results of the oxygen pressure trend change area and the identification results of the infiltration-thermal synergistic instability block to generate a multi-source early warning intersection label set. The specific steps of S1 are as follows: S101: Acquire the time series data recorded by the oxygen pressure sensing unit deployed in the burn wound area, and based on the difference in oxygen pressure values ​​in adjacent time periods, determine the points where the value changes from rising to falling or from falling to rising, filter the time sequence positions where the change direction switches, and generate a sequence of oxygen pressure change direction conversion points. S102: Based on the oxygen pressure change direction conversion point sequence, extract the amount and direction of oxygen pressure change before and after each location point in time intervals, calculate the difference in the magnitude and direction of the change in the value between the two time intervals, and filter out the conversion point locations where the direction reverses and the magnitude of the change in the value jumps, thereby generating a set of oxygen pressure trend change feature points. S103: Call the spatial location information in the set of oxygen pressure trend change feature points, aggregate the coordinate points of the sensing units to which the feature points belong, identify the continuously distributed or densely concentrated spatial regions, record the distribution range of the points with trend change features, and obtain the location result of the oxygen pressure trend change region. The specific steps of S2 are as follows: S201: Based on the spatial location range marked in the oxygen pressure trend change area location result, retrieve the seepage path information recorded by the pressure sensing point and the thermal gradient data recorded by the thermal sensing point in the corresponding area, extract the temporal position change trajectory of the seepage boundary in the seepage path within the same time period, and simultaneously extract the corresponding spatial position of the thermal gradient data to generate a set of seepage boundary change trajectory and temperature distribution curves. S202: Call the boundary trajectory and temperature change trend of the spatial location points in the set of seepage boundary change trajectory and temperature distribution curve, compare the two types of data in the same spatial segment, identify the overlapping segment where the boundary jumps and the temperature change shows a contraction trend, filter the spatial range with dual change characteristics, and obtain the identification result of the seepage heat co-instability block; The specific steps for S3 are as follows: S301: Based on the spatial region coordinates in the identification result of the heat permeation synergistic instability block, call the pH value data and oxygen pressure record of the corresponding block for the continuous time period, determine whether the pH value change direction remains downward, extract the range of oxygen pressure value change in the corresponding time period and determine whether the fluctuation amplitude remains in the stable range, and generate a set of acid-base and oxygen pressure state trend features. S302: Call the pH value downward persistence and oxygen pressure fluctuation range stability of spatial location points in the acid-base and oxygen pressure state trend feature set, determine whether they are simultaneously true within the same time period, identify continuous time periods and spatial ranges with a common trend of pH value decrease and oxygen pressure stability, and obtain the results of acid-oxygen synergistic hypoxia performance segments.

2. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 1, characterized in that, The oxygen pressure trend abrupt change region location results include direction reversal location points, amplitude abrupt change characteristic areas, and trend reversal segments. The infiltration-thermal co-instability block identification results include infiltration boundary abrupt change segments, thermal gradient abnormal contraction segments, and infiltration-thermal synchronous abnormal blocks. The acid-oxygen co-hypoxia manifestation segment results include pH value continuously declining segments, oxygen pressure fluctuation stable segments, and acid-oxygen trend co-existing segments. The multi-source early warning intersection label set includes three types of signal overlap segments, signal overlap quantity distribution, and signal overlap spatial range.

3. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 1, characterized in that, The pressure sensing point records the pressure value at its location in real time, monitors the dynamic changes in the pressure of seepage or gas, and infers the liquid seepage path. The seepage path information refers to the path information formed when the fluid migrates in the medium, including the flow direction, boundary position, and velocity change.

4. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 1, characterized in that, The stability of the oxygen pressure fluctuation range refers to whether the fluctuation range of the oxygen pressure value remains stable within a certain period of time. The acid-oxygen synergistic hypoxia manifestation zone is a region in which the pH value continuously decreases and the oxygen pressure fluctuates stably at the same time and spatial location.

5. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the coordinates of the monitoring units in the results of the acid-oxygen synergistic hypoxia manifestation segment, the spatial coordinates are superimposed and compared with the location results of the oxygen pressure trend change area and the identification results of the infiltration-thermal synergistic instability block. Monitoring units with completely consistent positions in the three types of results are filtered out, and coordinate segments with spatial overlap characteristics of the three types of signals are extracted to generate a set of overlapping segments of the three sources of signals: S402: Call the coordinates of the monitoring units in the set of overlapping segments of the three-source signals, count the distribution positions and total number in the overall monitoring area, mark the combination layout features in the spatial plane, and perform coordinate numbering and classification based on the marking results to obtain the set of multi-source early warning intersection labels.

6. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 1, characterized in that, The method further includes step S5: S5: Based on the signal coverage type, quantity and spatial coverage density characteristics in the multimodal signal fusion tag set, according to the joint distribution hierarchy of the three types of characteristic signals, divide the risk level corresponding to the early warning intersection tag set, and mark the early warning level status of the tag set with different degrees of differentiation, and output the wound infection early warning level status tag. The wound infection early warning level status label includes risk level identifier, label differentiation status, and joint distribution hierarchy.

7. The method for real-time early warning of burn wound infection risk based on multimodal sensing according to claim 6, characterized in that, The specific steps of S5 are as follows: S501: Based on the tags in the multi-source early warning intersection tag set, extract the total number of corresponding signal coverage types, the number of signal points contained, and the spatial coverage density value. Generate a fused signal coverage feature table based on the combination of signal types, the distribution of signal quantity, and the density of spatial distribution. S502: Call the fused signal coverage feature table, construct the distribution hierarchy of multimodal features according to the signal type joint occurrence level, signal quantity ratio range and spatial coverage density sorting, define the differentiated risk level according to the structural hierarchy, and generate a risk level table corresponding to the label distribution hierarchy. S503: Based on the risk level results of the label items in the risk level table corresponding to the label distribution hierarchy, the label set with differentiated level division is identified by status distinction, the level of the warning status is marked, the level status label is assigned according to the risk level, and the wound infection warning level status label is output.

Images