A method and system for real-time assisted first aid based on edge devices

Abstract

The application provides an edge device-based real-time first-aid method and system, which comprises the following steps: determining a target switching frequency band of an edge device based on acquired electromagnetic spectrum data and a preset signal interference spectrum, and acquiring multi-source data based on the target switching frequency band; correcting patient sign data based on environment data of the edge device to obtain target sign data, and compensating patient image data based on the environment data to obtain target image data; fusing the target sign data and the target image data based on an edge computing node of the edge device to obtain target fusion data; constructing a patient sign knowledge graph based on the target fusion data; performing diagnostic analysis based on the patient sign knowledge graph to obtain an abnormal diagnosis result, and triggering an alarm instruction based on the abnormal diagnosis result. The application solves the problem of electromagnetic interference of complex environments affecting data transmission of edge devices, and avoids interference of complex environments on data and image acquisition.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, in particular to a real-time auxiliary first-aid method and system based on edge devices. BACKGROUND

[0002] With the development of edge device technology, real-time auxiliary first-aid methods based on edge devices have gradually been applied. Edge devices, with the advantage of being close to data sources, can collect patient vital sign data and on-site environmental information in real time, providing strong support for first-aid work.

[0003] However, the current auxiliary first-aid method has a major problem of environmental interference when triggering an alarm in complex environments such as the wild, aviation, and mobile vehicles. For example, in the wild environment, the electromagnetic signals of communication base stations and power facilities can seriously interfere with data transmission between edge devices and medical devices, causing patient vital sign data to deviate, lose, or be delayed, which makes it difficult for first-aid personnel to obtain accurate medical information in a timely manner, delaying treatment. At the same time, the temperature, humidity, and light conditions in the wild also affect data collection. High temperature, low temperature, high humidity, and low humidity can interfere with device performance, and strong or weak light can affect image collection, thereby interfering with first-aid judgment and ultimately affecting the accuracy and timeliness of first aid. SUMMARY

[0004] The present application provides a real-time auxiliary first-aid method and system based on edge devices to solve the problem of electromagnetic interference affecting edge device data transmission in complex environments, while avoiding interference with data and image collection in complex environments.

[0005] In a first aspect, the present application provides a real-time auxiliary first-aid method based on edge devices, comprising:

[0006] Based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, the target switching frequency band of the edge device is determined, and multi-source data is acquired based on the target switching frequency band; the multi-source data includes patient sign data and patient image data;

[0007] The patient sign data is corrected based on the environmental data of the edge device to obtain target sign data, and the patient image data is compensated based on the environmental data to obtain target image data;

[0008] The target sign data and the target image data are fused based on the edge computing node of the edge device to obtain target fusion data;

[0009] A patient sign knowledge graph is constructed based on the target fusion data;

[0010] Based on the patient sign knowledge graph, a diagnostic analysis is performed to obtain an abnormal diagnosis result, and an alarm instruction is triggered based on the abnormal diagnosis result.

[0011] In a second aspect, the present application further provides an edge device-based real-time first-aid system, applied to the edge device-based real-time first-aid method as described in the first aspect; the edge device-based real-time first-aid system comprises:

[0012] a frequency band switching module, configured to determine a target switching frequency band of the edge device based on the acquired electromagnetic spectrum data and a preset signal interference spectrum, and acquire multi-source data based on the target frequency band; the multi-source data comprises patient sign data and patient image data;

[0013] a correction compensation module, configured to correct the patient sign data based on environmental data of the edge device to obtain target sign data, and compensate the patient image data based on the environmental data to obtain target image data;

[0014] a data fusion module, configured to fuse the target sign data and the target image data based on an edge computing node of the edge device to obtain target fusion data;

[0015] a knowledge graph construction module, configured to construct a patient sign knowledge graph based on the target fusion data;

[0016] an analysis and early warning module, configured to perform diagnostic analysis based on the patient sign knowledge graph to obtain an abnormal diagnosis result, and trigger an alarm instruction based on the abnormal diagnosis result.

[0017] In a third aspect, the present application further provides an electronic device, comprising: a memory, configured to store a computer software program; and a processor, configured to read and execute the computer software program, thereby realizing the edge device-based real-time first-aid method as described in any one of the above aspects.

[0018] In a fourth aspect, the present application further provides a non-transitory computer readable storage medium, wherein the storage medium stores a computer software program, and the computer software program is executed by a processor to realize the edge device-based real-time first-aid method as described in any one of the above aspects.

[0019] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, and the computer program is executed by a processor to realize the edge device-based real-time first-aid method as described in any one of the above aspects.

[0020] The edge device-based real-time auxiliary first-aid method provided by the embodiment of the present application determines the target switching frequency band of the edge device through the preset signal interference map and the acquired electromagnetic spectrum data, realizes real-time adjustment of the frequency band, can effectively avoid field electromagnetic interference, and further avoid data transmission obstruction, guarantee the integrity and accuracy of the data, and solve the problem of edge device data transmission affected by complex environment electromagnetic interference. Further, the patient sign data is corrected in combination with the environment data, the patient image data is compensated, more real patient sign data and clearer patient image data are obtained, the interference of the field environment on data and image acquisition is avoided, the quality and accuracy of the data are improved, more accurate data support is provided for subsequent first-aid diagnosis, and the accuracy and timeliness of the first-aid are further guaranteed. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a flowchart of the edge device-based real-time auxiliary first-aid method provided by the embodiment of the present application;

[0022] Figure 2 is a structural schematic diagram of the edge device-based real-time auxiliary first-aid system provided by the embodiment of the present application;

[0023] Figure 3 is an embodiment diagram of the electronic device provided by the embodiment of the present application;

[0024] Figure 4 is an embodiment diagram of the computer readable storage medium provided by the embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0026] The terms "first", "second", etc. are only used for description purpose, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features.

[0027] The word "for example" is used herein to mean "serving as an example, instance, or illustration." Any embodiment described as "for example" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the application. Details are set forth in the following description for purpose of explanation. It should be appreciated that one of ordinary skill in the art will readily recognize that the application can be practiced with the elements and / or steps in various order, combinations, and / or with additional or different elements and / or steps. In other instances, well known structures and / or processes are not described in detail in order to avoid obscuring the application. Thus, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0028] Referring to Figure 1 , Figure 1 is a flowchart of a real-time assisted first aid method based on an edge device provided by the present application. The execution subject of the real-time assisted first aid method based on an edge device in the embodiment of the present application is a real-time assisted first aid system. Therefore, the real-time assisted first aid method based on an edge device comprises:

[0029] Step 10, based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, the target switching frequency band of the edge device is determined, and the multi-source data is acquired based on the target switching frequency band. The multi-source data includes patient sign data and patient image data.

[0030] Optionally, the real-time assisted first aid system sets a spectrum monitoring module (such as an electromagnetic spectrum sensor) in the edge device. When monitoring the environment, the surrounding electromagnetic environment is scanned at a fixed frequency (such as once per second) to acquire electromagnetic spectrum data, which contains information such as signal intensity at different frequencies. The real-time assisted first aid system pre-stores a signal interference spectrum, which records information such as interference intensity and interference source type at different frequencies. The acquired electromagnetic spectrum data is compared with the signal interference spectrum to determine the frequency band with the least interference as the target switching frequency band for monitoring and transmission of the edge device, as described in steps 101-105. After determining the target switching frequency band, the edge device acquires multi-source data including patient sign data and patient image data in real time according to the target switching frequency band. For example, the heart rate, electrocardiogram, and other sign data of the patient are acquired through an electrocardiogram sensor, and the facial expression, body movement, and other image data of the patient are acquired through a camera.

[0031] Further, the signal interference spectrum stored in the real-time assisted first aid system also needs to collect the characteristics of newly appearing interference signals in real time, and update the preset signal interference spectrum periodically or when the interference characteristics change, so as to perfect the signal interference spectrum interference sources under different environments.

[0032] Further, in an embodiment, in a field first aid scene, a large number of WiFi signal interferences exist in the 2.4GHz frequency band, and fewer interferences exist in the 5.8GHz frequency band, so the 5.8GHz frequency band is set as the target switching frequency band. Then, the data of the smart bracelet (obtaining vital sign data such as heart rate and blood pressure) and the camera (obtaining image data of the patient) worn by the patient are obtained through the communication between the frequency band and the devices.

[0033] Step 20, based on the environmental data of the edge device, the patient vital sign data is corrected to obtain target vital sign data, and the patient image data is compensated based on the environmental data to obtain target image data.

[0034] Optionally, after obtaining the patient vital sign data and the patient image data, the real-time auxiliary first aid system collects environmental data such as environmental temperature data, humidity data, air pressure data, environmental illumination intensity data, and environmental reflectivity data in real time through the environmental sensors built-in or externally connected to the edge device. And because the environmental temperature, humidity, and air pressure will affect the measurement of the patient vital sign data, the patient vital sign data is corrected according to the temperature data, humidity data, and air pressure data in the environmental data, eliminating the interference of environmental factors on the data, obtaining more accurate target vital sign data, as described in steps 2011-2015.

[0035] Further, because the environmental illumination intensity and the environmental reflectivity and other environmental data will affect the collection of the image data, the patient image data is compensated according to the environmental illumination intensity and the environmental reflectivity, and the target image data with clearer image details is obtained, as described in steps 2021-2025.

[0036] Step 30, based on the edge computing node of the edge device, the target vital sign data and the target image data are fused to obtain target fusion data.

[0037] Optionally, the real-time auxiliary first aid system fuses the target vital sign data and the target image data based on the time series fusion based on the edge computing node of the edge device, as described in steps 301-304. Through data fusion, different types of data features are integrated, and the complementarity of multi-source data is fully utilized to provide more rich and comprehensive patient information.

[0038] Step 40, based on the target fusion data, a patient vital sign knowledge graph is constructed.

[0039] Optionally, after the real-time assisted first-aid system receives the target fusion data, the target fusion data is first normalized and preprocessed, and then a patient sign knowledge graph is constructed according to the sign change characteristics in the time dimension and the coupling characteristics between signs in the space dimension, as described in steps 401-404, to present the sign information of the patient and its associated relationship in a structured manner, facilitate quick query and analysis, provide intuitive and comprehensive knowledge support for diagnosis, and help medical personnel quickly understand the patient's condition.

[0040] Step 50, based on the patient sign knowledge graph, performing diagnosis analysis to obtain an abnormal diagnosis result, and triggering an alarm instruction based on the abnormal diagnosis result.

[0041] Optionally, the real-time assisted first-aid system determines the diagnosis target path based on the constructed patient sign knowledge graph and the preset diagnosis knowledge base, and then determines the abnormal diagnosis result in the patient sign knowledge graph in a path search matching manner, as described in steps 501-504, to determine whether the patient has an abnormal condition, such as determining that there may be a heart problem according to the knowledge graph, and then obtaining an abnormal diagnosis result and triggering an alarm instruction. The edge device sends an alarm instruction to the first-aid center, medical personnel, and other related parties through a pre-set communication module (such as a 4G / 5G module), and the alarm instruction contains patient basic information, abnormal diagnosis result, and other contents.

[0042] The embodiment of the present application determines the target switching frequency band of the edge device by the preset signal interference graph and the acquired electromagnetic spectrum data, realizes real-time adjustment of the frequency band, can effectively avoid electromagnetic interference in the wild, and further avoid data transmission blockage, thereby protecting the integrity and accuracy of the data, and solving the problem of electromagnetic interference in complex environments affecting the data transmission of the edge device. Further combining the environmental data to correct the patient sign data and compensate the patient image data, more real patient sign data and clearer patient image data are obtained, the interference of the wild environment on data and image acquisition is avoided, the quality and accuracy of the data are improved, more accurate data support is provided for subsequent first-aid diagnosis, and the accuracy and timeliness of the first-aid are further ensured.

[0043] In an embodiment, steps 101-105 are described as follows:

[0044] Step 101, the acquired electromagnetic spectrum data is divided into multiple modal data, and the multiple modal data are preprocessed respectively to obtain preprocessed multiple modal data.

[0045] Optionally, the real-time assisted first-aid system divides the collected electromagnetic spectrum data into signal strength mode, frequency mode and phase mode according to preset division rules (such as data characteristics) according to the electromagnetic spectrum data collected by the spectrum monitoring module (such as a spectrum analyzer) of the edge device. After obtaining the signal strength mode, frequency mode and phase mode, for the signal strength mode, the signal strength values collected by different devices may differ greatly, and normalization preprocessing is performed for subsequent processing; for the frequency mode, a filtering algorithm is used to remove noise frequency interference in the data; for the phase mode, a smoothing algorithm is used for preprocessing due to possible abnormal jumps in the data. Through modal division and targeted preprocessing, the data of different modes is more standardized and pure, providing a high-quality data basis for subsequent feature extraction and analysis, and reducing the influence of noise and data differences on the results.

[0046] Step 102, feature extraction is performed on the preprocessed signal strength mode, frequency mode and phase mode respectively to obtain signal strength features, frequency features and phase features.

[0047] Optionally, the real-time assisted first-aid system calculates the mean and variance of the signal strength mode for the preprocessed signal strength mode to obtain signal strength features, and the calculation formula is as follows: mean , represents the number of signal strength samples, represents the preprocessed signal strength value), and variance . The mean and variance reflect the overall level and fluctuation of the signal strength. For the preprocessed frequency mode, the frequency distribution features (i.e. frequency features) are obtained by counting the number of frequencies in different frequency intervals. For example, the frequency range is divided into intervals, and the number of frequencies in each interval is counted, such as there are 20 frequency samples in the interval, and there are 15 frequency samples in the interval, and so on to describe the frequency distribution. For the preprocessed phase mode, the difference between adjacent phase values is calculated, and the variation of these differences is counted to obtain the phase change features (i.e. phase features). For example, the difference sequence is reflects the trend of phase change. Different modal data contains different information forms, and through specific feature extraction methods, the key information of each modal data can be extracted to provide effective features for subsequent multi-modal fusion and frequency band decision.

[0048] Step 103, feature extraction based on signal interference atlas library to obtain interference information features.

[0049] Optionally, the real-time emergency assistance system extracts key information from the signal interference map stored in the edge device through image recognition and data analysis techniques. For the interference frequency range characteristics, an edge detection algorithm such as the Canny algorithm is used to identify the interference frequency band boundaries in the map to determine the start and end frequencies of the interference frequency band. For the interference intensity distribution characteristics, the map is converted into an intensity matrix based on the mapping relationship between the map gray value or color coding and the intensity, and the element distribution in the matrix is analyzed to obtain the interference intensity distribution characteristics. By correlating these information with the electromagnetic spectrum characteristics, such as comparing the interference frequency band with the signal frequency distribution in the electromagnetic spectrum, the potential interference situation can be determined.

[0050] Step 104, the signal intensity characteristics, frequency characteristics, phase characteristics and interference information characteristics are fused in series to obtain a comprehensive feature vector.

[0051] Optionally, the real-time emergency assistance system uses feature-level fusion to fuse the signal intensity characteristics such as mean and variance of the signal intensity mode, frequency distribution characteristics of the frequency mode, phase change characteristics of the phase mode, and interference information characteristics such as interference frequency range and interference intensity distribution of the interference map into a comprehensive feature vector in sequence. The signal intensity feature vector is , the frequency feature vector is , , the phase feature vector is , and the interference information feature vector is , , for example, the interference frequency range and intensity related parameters. The comprehensive feature vector forms a long vector containing multiple information. The feature-level fusion integrates features from different sources and types, making the comprehensive feature vector contain comprehensive electromagnetic spectrum and interference related information, providing more complete data support for subsequent frequency band decision.

[0052] Step 105, analyze and filter the comprehensive feature vector to determine the target frequency band.

[0053] Optionally, the real-time emergency assistance system sets filtering decision rules based on the fused comprehensive feature vector, such as setting interference intensity threshold , signal quality threshold , etc. If the interference intensity of a certain frequency band in the interference information characteristics exceeds , the frequency band is not considered as the target frequency band; if the mean value in the signal intensity characteristics is lower than , the frequency band is also not considered as the target frequency band. Traverse all the comprehensive feature vectors corresponding to the frequency bands, and filter out the frequency bands that meet the requirements as the target frequency band according to these filtering decision rules. For example, in a certain scenario, Set to medium intensity interference, Set to 0.4 (normalized signal intensity mean), after screening, it is found that the 5.8GHz frequency band meets the condition that the interference intensity is lower than the medium intensity and the signal intensity mean is greater than 0.4, so it is determined as the target frequency band for edge device transmission frequency band switching. Through the analysis and screening of the comprehensive feature vector according to the explicit rules, the most suitable target frequency band can be determined from numerous frequency bands scientifically and reasonably, the accuracy and reliability of the frequency band switching are improved, and the stability of the edge device communication is ensured.

[0054] The embodiment of the application determines the target switching frequency band through multi-modal information fusion, fully utilizes various modal information of electromagnetic spectrum and signal interference spectrum information, and avoids the one-sidedness of frequency band selection according to only single information. From data preprocessing to feature extraction, fusion and decision, the whole process is closely linked, and key information is gradually mined and integrated, so that the frequency band switching decision is more scientific and accurate, and the communication quality and stability of the edge device in the complex electromagnetic environment can be effectively improved.

[0055] In an embodiment, steps 2011-2015 are described as follows:

[0056] Step 2011, the temperature data, humidity data and air pressure data are normalized respectively to obtain temperature normalized data, humidity normalized data and air pressure normalized data.

[0057] Optionally, the real-time auxiliary first-aid system acquires temperature data, humidity data and air pressure data from the sensor Humidity data And air pressure data . In order to eliminate the influence of different environmental data dimensions and value range differences on subsequent calculation, normalization processing is performed. For temperature data, assuming that its value range is between to , according to the formula , it is mapped to the interval . The humidity data and the air pressure data are normalized by and respectively, so as to obtain temperature normalized data , humidity normalized data and air pressure normalized data .

[0058] Further, in an embodiment, if the collected temperature range is . According to the normalization formula , if the current collected temperature , , , the temperature normalized data . Similarly, for humidity data, if the collection range is , humidity of the certain collection , by , humidity normalized data . The air pressure data is also processed in a similar way. If the collected air pressure range is , air pressure of the certain collection , then the air pressure normalized data .

[0059] Step 2012, based on the temperature normalized data, humidity normalized data and air pressure normalized data, the influence of different sign data in the patient sign data on the environmental impact factor matrix is constructed.

[0060] Optionally, the real-time first aid system analyzes the influence degree of the three environmental factors of temperature, humidity and air pressure on different sign data (such as heart rate, blood pressure, respiratory rate, etc.) according to a large amount of clinical research and experimental data. Taking heart rate as an example, a large number of heart rate data samples are collected under different combinations of temperature, humidity and air pressure, and the influence factor of temperature on heart rate , the influence factor of humidity on heart rate , and the influence factor of air pressure on heart rate are calculated by statistical methods (such as regression analysis). Similarly, for other sign data (assuming there are n kinds of sign data), the corresponding influence factors are calculated respectively, and a 3xn environmental impact factor matrix A is constructed, where represents the influence factor of the environmental factor (T =1 corresponds to temperature, T =2 corresponds to humidity, and T =3 corresponds to air pressure) on the sign data. For example, through research, it is found that the heart rate increases by 0.5 times / minute (assuming value) on average when the temperature increases by 1℃. After data statistics and calculation, T . The influence degree of different environmental factors on different sign data is different, and the influence relationship is complex. The construction of the environmental impact factor matrix can quantitatively systematize the influence relationship and provide a basis for subsequent accurate calculation of the comprehensive influence of the environment on the sign data.

[0061] Step 2013, for each sign data in the patient sign data, the environmental comprehensive influence value is determined based on the environmental impact factor matrix.

[0062] Optionally, the real-time first aid system combines the previously obtained normalized environmental data with the environmental impact factor matrix to calculate the environmental comprehensive influence value of each sign data in the patient sign data ( , such as representing heart rate, representing blood pressure, representing respiratory rate). and environmental impact factor matrix elements in the matrix , the environmental comprehensive impact value of the patient's vital sign data is calculated according to the formula , wherein, represents the normalized environmental data of the patient's vital sign data (T = 1 corresponds to temperature normalization data, T = 2 corresponds to humidity normalization data, and T = 3 corresponds to air pressure normalization data). Further, taking heart rate as an example, it is known that are the temperature, humidity, and air pressure impact factors on heart rate, respectively, are the normalized temperature, humidity, and air pressure data, and the environmental comprehensive impact value of the heart rate is

[0063] The environmental comprehensive impact value comprehensively reflects the overall impact of the current environmental factors on the vital sign data.

[0064] Step 2014, the category and environmental comprehensive impact value of the patient's vital sign data are matched with the preset correction function library to obtain the target correction function corresponding to each vital sign data.

[0065] Optionally, the real-time assisted first aid system pre-stores the preset correction function library in the storage module of the edge device, wherein the correction function library sets corresponding correction functions for different vital sign data categories and different environmental comprehensive impact value ranges. For example, for the heart rate vital sign data category, according to a large amount of clinical data and medical knowledge, when the environmental comprehensive impact value is within a certain range, the heart rate is affected by the environment and presents a certain specific nonlinear change rule, and accordingly the corresponding correction function is designed. When the category and environmental comprehensive impact value of the patient's vital sign data are obtained, they are matched with the conditions in the correction function library.

[0066] Further, for the heart rate vital sign data, when the environmental comprehensive impact value is in the interval , the matched target correction function is ; for the blood pressure vital sign data, if the environmental comprehensive impact value is in the interval , the matched target correction function is The target correction function corresponding to the current patient's vital sign data category and the calculated environmental comprehensive impact value is searched from the library.

[0067] Step 2015, the patient's vital sign data is corrected based on the target correction function to obtain the target vital sign data.

[0068] ​​​​​Optionally, the real-time assisted first-aid system corrects the patient sign data according to the determined target correction function, for example, the original heart rate data in the patient sign data Second minutes, by obtaining the environmental comprehensive influence value corresponding to the heart rate , the matched target correction function , according to the formula , the corrected target sign data (heart rate) , and the other sign data are corrected in the same way, but it should be noted that this only represents an example way and is not limited to this specific formula. According to the actual needs, the corresponding adjustment can be made. Finally, the original sign data is corrected by using the target correction function, and the nonlinear influence of the environmental factors is fully considered. Compared with the simple processing method, the real sign situation can be restored more accurately.

[0069] The embodiment of the application fully considers the complex influence of environmental factors on different sign data, realizes individualized and nonlinear correction by using mathematical functions, can adapt to various clinical scenes, and effectively improves the accuracy and reliability of the patient sign data, providing a better data basis for medical diagnosis and first-aid decision.

[0070] In an embodiment, steps 2021-2025 are described as follows:

[0071] Step 2021, difference analysis is performed on the spectral data of the environmental light and the spectral data of the patient image data to obtain the spectral difference degree.

[0072] Optionally, the real-time assisted first-aid system uses a hyperspectral imaging device to respectively collect the spectral data of the environmental light and the spectral data of the patient image data , wherein represents the wavelength. The spectral difference degree is measured by calculating the difference between the two , the formula is , which is the absolute value of the difference between the spectral intensity of the environmental light and the spectral intensity of the patient image at each wavelength , and the sum is obtained. For example, in a spectral data containing 100 wavelength sampling points, the absolute value of the difference between and is calculated and accumulated at each wavelength point. If at a certain wavelength , , the contribution of the difference at this point is , and finally the contributions of all wavelength points are accumulated to obtain the spectral difference degree .

[0073] Step 2022, based on the spectral difference, the spectral data of the patient image data is adjusted by spectral matching, and a spectral matching adjustment image is obtained.

[0074] Optionally, the real-time assisted first aid system adjusts the patient image spectral data by linear interpolation and polynomial fitting method according to the obtained spectral difference. For example, the formula is used to adjust the patient image spectral data, wherein is the adjusted spectral data, is the adjustment coefficient, which can be determined by optimization algorithm (such as gradient descent method, etc.) to make the adjusted spectral as close as possible to the ambient light spectrum . For example, if , at a certain wavelength , , adjust the spectral data corresponding to all wavelengths in this way to obtain the spectral matching adjustment image of the adjusted spectral data. The spectral data of the patient image is more matched with the ambient light spectrum, and the color deviation and other problems caused by the spectral difference of the image are corrected, and the color accuracy of the image is improved.

[0075] Step 2023, convert the spectral matching adjustment image from the original image space to the CIE-LUV nonlinear space to obtain a nonlinear transformation image.

[0076] Optionally, the real-time assisted first aid system converts the spectral matching adjustment image from the original image space (such as the common RGB space) to the CIE-LUV nonlinear space according to the obtained spectral matching adjustment image using an image conversion algorithm. The CIE-LUV space is a color space that conforms to the visual characteristics of the human eye, in which L represents brightness, and U and V represent chroma. For example, for a pixel point (R, G, B) = (200, 100, 50) in the RGB space, through a series of conversion formulas (based on the conversion relationship of CIE standard), it is converted into the coordinates (L, U, V) in the CIE-LUV space. After conversion, a nonlinear transformation image is obtained, which can be more conveniently adjusted based on illumination and reflectivity in the new color space.

[0077] Step 2024, based on the ambient light intensity, the L channel of the nonlinear transformation image is enhanced, and based on the ambient light reflectivity, the U channel and the V channel of the nonlinear transformation image are respectively adjusted in color saturation and hue, and an enhanced nonlinear transformation image is obtained.

[0078] Optionally, the real-time assisted first aid system first obtains the ambient light intensity from the ambient light sensor, and simultaneously obtains the ambient reflectivity from the reflectivity measuring instrument, and then adjusts the Channel (luminance channel), according to ambient light intensity is processed. Lower, using an enhanced luminance nonlinear function , wherein and are parameters that can be adjusted according to actual scenes and experience. For example, when , the original channel value , then . For channel and channel (chrominance channel), according to ambient light reflectivity adjustment, using the formula , wherein and are adjustment coefficients. For example , , then , . Through these calculations, the enhanced nonlinear transformed image is obtained. Since the ambient light intensity and reflectivity will affect the visual effect of the image, directly affecting the perception of the human eye to the brightness and color of the image. Therefore, according to the two environmental data, the image channel is enhanced, which can more truly restore the visual performance of the image in the actual environment.

[0079] Step 2025, convert the nonlinear transformed image back to the original image space to obtain target image data.

[0080] Optionally, the real-time auxiliary first-aid system converts the nonlinear transformed image from the CIE-LUV space back to the original image space (such as the RGB space). For example, according to the coordinates in the CIE-LUV space , , ), through the corresponding conversion formula, it is converted to , , ), to obtain target image data , completing the compensation of the patient image data. Converting the image back to the original image space facilitates the use in the original image display and processing system, and at the same time, the image compensated after the previous steps has been optimized in color, brightness, etc. It can provide clearer and more accurate visual information for medical diagnosis, etc.

[0081] The embodiment of the present application fully considers the relationship between the ambient light and the image spectrum and the influence of environmental factors on the visual effect of the image. Through scientific and reasonable spectral and color space processing means, the image color deviation can be effectively corrected, the brightness and color performance of the image can be improved, and the image can be closer to the visual presentation under the real scene.

[0082] In an embodiment, steps 301-304 are described as follows:

[0083] Step 301, time series analysis is performed on the target sign data to obtain time series analysis results.

[0084] Optionally, the real-time first aid system usually collects the target sign data such as heart rate, blood pressure, etc. according to the time sequence of the collected sequence data. Therefore, when performing time series analysis on the target sign data, taking heart rate data as an example, it is assumed that the edge device collects the heart rate of the patient every 1 minute, and obtains the heart rate sequence An autoregressive integrated moving average (ARIMA) model is used for modeling. First, the parameters of the model need to be determined (the order of autoregression), (the order of difference), (the order of moving average). The value of can be determined by observing the stationarity of the data and using unit root tests (such as ADF test). For and , different values can be tried, and the values of information criteria such as AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion) can be calculated to select the combination of and that minimizes the criterion value. Assuming that is determined through calculation, the ARIMA model obtained is , and the prediction model is , such as , where represents a functional relationship determined according to the model parameters. Through this model, the change in heart rate data in the future period (such as 5 minutes) can be predicted. Similarly, the same analysis is performed on other sign data, and the time series analysis results are finally obtained. By deeply mining the time series characteristics of the sign data, the change trend of the sign data is accurately predicted, which provides strong support for comprehensively understanding the physiological state of the patient and improves the ability to dynamically monitor the health status of the patient.

[0085] Step 302, the acquisition time of the target image data is marked, and the change trend analysis is performed based on the time marking result to obtain a time feature trend vector.

[0086] Optionally, when the camera in the edge device of the real-time first aid system captures the patient image, the acquisition time of each image is also recorded. For a series of continuously captured images, the optical flow method is used to calculate the pixel motion vector between adjacent images. The optical flow method is based on the assumption that the brightness of pixels in an image remains unchanged within a short period of time, and the motion of pixels is calculated by solving the optical flow equation. Assuming that the pixel motion vector between two adjacent frames of images and For a pixel point , its motion vectors in the direction and the direction are and respectively, which are calculated by the optical flow algorithm. By statistically analyzing these motion vector information, such as calculating the average motion vector, the variance of the motion vector, and other statistical quantities, and combining the position changes of the facial feature points (such as eyes, mouth, etc.) in the image, the change information of the image is quantified into a time-dependent feature vector , that is, a time feature trend vector is obtained, such as the feature vector may contain average optical flow velocity, mean displacement of facial feature points, etc. to reflect the change trend of facial expression or physiological state.

[0087] Step 303, based on the time sequence analysis result and the time feature trend vector, a space-time correlation model is constructed according to the order and mutual influence.

[0088] Optionally, the real-time first aid system combines the obtained time sequence analysis result (such as the predicted data of heart rate and the time feature trend vector ). Considering their order and mutual influence in time. For example, when the vital sign data (such as heart rate) shows abnormal fluctuations, observe whether there are corresponding change characteristics (such as the increase of the tension of the patient's facial expression) in the subsequent image data. By collecting a large amount of historical data, including vital sign data sequences, corresponding image data sequences, and their time mark information, etc., a machine learning algorithm (such as neural network, random forest, etc.) is used to train the model. Taking the neural network as an example, the time sequence features of the vital sign data and the time feature vectors of the image data are input into the input layer, and after the calculation and processing of multiple layers of neurons, a numerical value or function relationship representing the correlation degree of the two is output, the parameters of the space-time correlation model are determined, and the correlation function is obtained. It can comprehensively consider the change rule and mutual relationship of vital sign data and image data in space-time.

[0089] Step 304, based on the space-time correlation model, the target vital sign data and the target image data are fused to obtain a fusion intermediate result, and the fusion intermediate result is smoothed to obtain the target fusion data.

[0090] Optionally, the real-time first aid system combines the current target vital sign data (such as the actually measured heart rate and the target image data (such as the current collected facial image and its corresponding time-dependent feature vector ), and combines the constructed space-time correlation model for fusion calculation. The predicted vital sign data Temporally related feature vectors of images Substitute into the correlation function , obtain intermediate fusion results Right now Because the data may contain noise or outliers, which can affect the accuracy of the fusion results, intermediate fusion results should be... Smoothing is performed. Median filtering is used for... The sequence (arranged chronologically) is given a window of a certain size (e.g., window size is 1). For data within a window, the median value of the data within that window is used as the new value at that position. For example, for a sequence... When the window size At that time, for Pick 3) The median as the new (2) Value, and so on, to obtain the final target fusion data. The fusion calculation fully integrates the information of the vital signs data and image data, and the smoothing process effectively removes noise and outliers, improving the quality of the target fusion data so that it can more accurately reflect the patient's comprehensive status in different times and spaces, providing a reliable basis for subsequent analysis and diagnosis based on the fusion data.

[0091] This invention fully considers the spatiotemporal characteristics of vital sign data and image data. Through time series analysis and image time stamping and trend analysis, it uncovers the temporal features of both types of data. The spatiotemporal correlation model further integrates their spatiotemporal relationships. The resulting target fusion data better reflects the patient's overall condition in different times and spaces, making it suitable for scenarios requiring dynamic monitoring of patients' physiological and physical changes, and providing more comprehensive and accurate data support for medical diagnosis and analysis.

[0092] In one embodiment, steps 401-404 are described as follows:

[0093] Step 401: Normalize the target fusion data to obtain preprocessed target scale data.

[0094] Optionally, the real-time assisted emergency rescue system uses target fusion data. Since this data includes various vital signs such as body temperature, blood pressure, and heart rate, which have different dimensions and value ranges, the target fusion data is normalized to eliminate the impact of these dimensional differences on subsequent analysis, resulting in preprocessed target-scale data.

[0095] Step 402: Based on the spatiotemporal convolutional neural network, feature extraction is performed on the target scale data to obtain the characteristics of physical feature changes in the time dimension and the characteristics of physical feature coupling in the spatial dimension.

[0096] Optionally, the real-time first aid system pre-constructs the spatio-temporal convolutional neural network structure. In the time dimension, different size convolution kernels are set, such as convolution kernels with sizes of 3, 5, and 7. Taking the convolution kernel with a size of 3 as an example, it slides on the time series data, covering 3 consecutive time steps each time. Assuming that there is a heart rate time series at present, the first sliding of the convolution kernel covers The change features of the heart rate in the 3 time steps are extracted by convolution operation (such as weighted summation operation). In the spatial dimension (different sign dimensions), convolution kernels with sizes of 2 and 3 are set. For example, for the heart rate and blood pressure data, the convolution kernel with a size of 2 operates between the two sign data to extract the coupling features between them, that is, to analyze the correlation between the heart rate and the blood pressure. Through the convolution operation, the sign change features in the time dimension and the sign coupling features in the spatial dimension are obtained respectively. The features of the sign data in the time and spatial dimensions can be comprehensively and deeply extracted, and the complex sign change patterns have good adaptability.

[0097] In step 403, the sign change features and the sign coupling features are spliced and fused to obtain a sign feature vector.

[0098] Optionally, the real-time first aid system splices and fuses the sign change features extracted in the time dimension and the sign coupling features extracted in the spatial dimension. For example, multiple feature vectors are extracted in the time dimension by different size convolution kernels Multiple feature vectors are also extracted in the spatial dimension. These feature vectors are connected in order to form a more rich sign feature vector The comprehensive feature vector contains various feature information of the sign data in the time and spatial dimensions, and can more comprehensively describe the sign situation of the patient.

[0099] In step 404, a patient sign knowledge graph is constructed with the sign feature vector as a node and the correlation between the signs as an edge.

[0100] Optionally, the real-time first-aid system constructs the patient sign knowledge graph by taking the spliced and fused sign feature vectors as node attributes. The correlation between the signs is the edge, for example, for the temperature and pulse signs, they are correlated in physiology. In the knowledge graph, the temperature node and the pulse node are created, and the corresponding sign feature vectors are respectively assigned to the two nodes as attributes. Then, the edge is established according to the correlation between them, such as a directed edge from the temperature node to the pulse node, indicating the correlation between the two. According to the information in the sign feature vector, such as the correlation strength of the change of the temperature and the pulse, the edge weight and other attributes are assigned. If the correlation between the two is strong, the weight of the edge is large; otherwise, it is small. Through the construction of all sign nodes and edges, the patient sign knowledge graph is formed. The constructed patient sign knowledge graph can accurately reflect the characteristics of the sign data and the correlation between the signs, and provide clear and effective knowledge structure for the diagnosis analysis based on the knowledge graph.

[0101] The embodiment of the present application fully utilizes the advantages of deep learning in feature extraction, effectively captures the time and space dimension information of the sign data through the spatio-temporal convolutional neural network, and mines the potential features of the data. The constructed patient sign knowledge graph can accurately reflect the correlation between the signs and has good adaptability to complex sign change patterns, thereby improving the accuracy and reliability of the knowledge graph, providing strong support for the analysis and decision-making based on the knowledge graph in the medical field, and helping to improve the effect of medical diagnosis and treatment.

[0102] In an embodiment, steps 501-504 are described as follows:

[0103] Step 501, determining the diagnosis target path based on the preset diagnosis knowledge base and the patient sign knowledge graph.

[0104] Optionally, the real-time first-aid system stores the preset diagnosis knowledge base in a storage device or a server connected with the edge device, and the preset diagnosis knowledge base contains medical diagnosis knowledge and common disease diagnosis process. After the patient sign knowledge graph is determined, the diagnosis target path is determined according to the structure and semantic relationship of the patient sign knowledge graph combined with the preset diagnosis knowledge base, that is, the medical knowledge in the diagnosis knowledge base is analyzed and mapped and matched with the sign nodes and relationships in the knowledge graph. Taking the diagnosis of respiratory system diseases as an example, according to the correlation between the signs such as “cough” and “dyspnea” and diseases such as “pneumonia” and “asthma” recorded in the diagnosis knowledge base, if there is a “cough” node in the patient sign knowledge graph, a diagnosis target path such as “cough→abnormal sputum appearance→pneumonia diagnosis” can be planned from the node according to the diagnosis logic of the respiratory system diseases in the diagnosis knowledge base.

[0105] Step 502, taking the current sign data node of the patient sign knowledge graph as the initial point, searching for paths matching the diagnosis target path to obtain multiple matching paths.

[0106] Optionally, the real-time first-aid assisting system takes the current sign data node in the patient sign knowledge graph as the initial point, and uses a graph search algorithm (such as a depth-first search algorithm or a breadth-first search algorithm) to search for paths matching the diagnosis target path in the knowledge graph. In the search process, information such as the relationship between nodes and the weight of the edge is considered. For example, when using the depth-first search algorithm, starting from the initial sign node, constantly searching along the edge, and preferentially exploring those path branches that are consistent with the direction of the diagnosis target path and have a larger edge weight (indicating a closer relationship). Assuming that the current sign node is "fever", in the search process, encountering the "abnormal white blood cell count" node connected to "fever", and the weight of the connecting edge is higher, it is preferred to continue searching along this edge to determine whether it can gradually match other nodes in the diagnosis target path, thereby obtaining multiple possible matching paths.

[0107] Further, taking the depth-first search as an example, starting from the "cough" node, searching along the edge connected to "cough" and having a higher weight to the "respiratory tract inflammation" node, and then continuing to search for nodes connected to the "respiratory tract inflammation" node such as the "lung shadow" node, and constantly searching deeper to obtain multiple matching paths such as "cough→respiratory tract inflammation→lung shadow→pneumonia diagnosis".

[0108] Step 503, determining the path confidence for each path in the multiple matching paths based on the reliability of the nodes on the path and the weight of the edge.

[0109] Optionally, the real-time first-aid assisting system considers the reliability of the nodes on the path and the weight of the edge to determine the path confidence for each path searched. The reliability of the node is evaluated according to factors such as the accuracy of the data source and the precision of the collection device. For example, if a certain sign data is collected by a high-precision medical device and the data source is authoritative and reliable, the reliability of the node is relatively high. The weight of the edge reflects the closeness and correlation of the relationship between nodes, and the higher the weight, the closer the relationship. The path confidence can be calculated using a method based on evidence theory, assuming that there are nodes on the path, the node reliabilities are , and the edge weights are , the path confidence can be calculated by the formula .

[0110] Step 504, screening and excluding the path confidence to obtain a candidate diagnosis path, and determining an abnormal diagnosis result based on the candidate diagnosis path.

[0111] Optionally, the real-time first aid system pre-sets a credibility threshold, such as 0.6, and filters out the paths with a path credibility lower than the threshold. The remaining high-credibility paths are further analyzed and compared, and the path with the highest credibility is selected as the candidate diagnosis path. For example, after filtering, there are two high-credibility paths, one with a credibility of 0.6 and pointing to the "pneumonia" diagnosis, and the other with a credibility of 0.7 and pointing to the "bronchitis" diagnosis. In this case, the "bronchitis" corresponding to the path with a credibility of 0.7 is selected as the candidate diagnosis path. If there are multiple high-credibility paths pointing to different disease diagnoses, the most reasonable abnormal diagnosis result is finally determined by combining other related sign data (such as whether there is a wheezing symptom) and patient history (such as whether there is a history of chronic respiratory disease). Through filtering and comprehensive judgment, the most reliable diagnosis path and result can be determined from numerous matching paths, improving the accuracy and reliability of diagnosis.

[0112] According to the patient sign knowledge graph, the embodiments of the present application determine the diagnosis target path, search the matching path, evaluate the path credibility, and filter and determine the diagnosis result, fully utilizing the structure and semantic information of the knowledge graph for diagnosis reasoning. The logic and systematicness of the diagnosis are improved, making the diagnosis process interpretable and traceable, which helps doctors understand the diagnosis basis and reasoning process.

[0113] Further, the real-time first aid system based on an edge device provided by the present application is described below. The real-time first aid system based on an edge device described below can be correspondingly referred to the real-time first aid method based on an edge device described above.

[0114] Optionally, referring to Figure 2 , Figure 2 is a structural schematic diagram of the real-time first aid system based on an edge device provided by the present application. The real-time first aid system based on an edge device comprises:

[0115] The frequency band switching module 210 is configured to determine a target switching frequency band of the edge device based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, and acquire multi-source data based on the target frequency band. The multi-source data comprises patient sign data and patient image data.

[0116] The correction and compensation module 220 is configured to correct the patient sign data based on the environmental data of the edge device to obtain target sign data, and compensate the patient image data based on the environmental data to obtain target image data.

[0117] The data fusion module 230 is configured to fuse the target sign data and the target image data based on the edge computing node of the edge device to obtain target fusion data.

[0118] Knowledge graph construction module 240 is used to construct a patient vital signs knowledge graph based on target fusion data;

[0119] The analysis and early warning module 250 is used to perform diagnostic analysis based on the patient's vital signs knowledge graph, obtain abnormal diagnostic results, and trigger alarm commands based on the abnormal diagnostic results.

[0120] This invention determines the target switching frequency band for edge devices by using a preset signal interference map and acquired electromagnetic spectrum data, enabling real-time frequency band adjustment. This effectively avoids electromagnetic interference in the field, thus preventing data transmission obstruction and ensuring data integrity and accuracy. It solves the problem of electromagnetic interference in complex environments affecting data transmission from edge devices. Furthermore, by combining environmental data to correct patient vital sign data and compensate for patient image data, more realistic patient vital sign data and clearer patient image data are obtained. This avoids interference from the field environment on data and image acquisition, improving data quality and accuracy. This provides more precise data support for subsequent emergency diagnosis, thereby ensuring the accuracy and timeliness of emergency treatment.

[0121] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3 As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps:

[0122] Based on the acquired electromagnetic spectrum data and the preset signal interference map, the target switching frequency band of the edge device is determined, and multi-source data is acquired based on the target switching frequency band; the multi-source data includes patient vital sign data and patient image data;

[0123] The patient's vital signs data are corrected based on the environmental data of the edge device to obtain the target vital signs data, and the patient's image data is compensated based on the environmental data to obtain the target image data.

[0124] Edge computing nodes based on edge devices fuse target vital sign data and target image data to obtain target fused data;

[0125] Constructing a patient vital signs knowledge graph based on target fusion data;

[0126] Diagnostic analysis is performed based on the patient's vital signs knowledge graph to obtain abnormal diagnostic results, and alarm commands are triggered based on the abnormal diagnostic results.

[0127] Please see Figure 4 , Figure 4An embodiment of a computer readable storage medium provided for an embodiment of the present application is shown in the figure. Figure 4 As shown in the figure, the embodiment provides a computer readable storage medium 400, which stores a computer program 311, and the computer program 311 is executed by a processor to implement the following steps:

[0128] Based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, a target switching frequency band of the edge device is determined, and multi-source data is acquired based on the target switching frequency band; the multi-source data includes patient sign data and patient image data;

[0129] The patient sign data is corrected based on the environmental data of the edge device to obtain target sign data, and the patient image data is compensated based on the environmental data to obtain target image data;

[0130] The target sign data and the target image data are fused based on an edge computing node of the edge device to obtain target fusion data;

[0131] A patient sign knowledge graph is constructed based on the target fusion data;

[0132] Diagnosis analysis is performed based on the patient sign knowledge graph to obtain an abnormal diagnosis result, and an alarm instruction is triggered based on the abnormal diagnosis result.

[0133] On the other hand, the present application also provides a computer program product, which includes a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer can execute the real-time auxiliary first-aid method based on the edge device provided by the above-mentioned methods, and the method includes:

[0134] Based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, a target switching frequency band of the edge device is determined, and multi-source data is acquired based on the target switching frequency band; the multi-source data includes patient sign data and patient image data;

[0135] The patient sign data is corrected based on the environmental data of the edge device to obtain target sign data, and the patient image data is compensated based on the environmental data to obtain target image data;

[0136] The target sign data and the target image data are fused based on an edge computing node of the edge device to obtain target fusion data;

[0137] A patient sign knowledge graph is constructed based on the target fusion data;

[0138] Diagnosis analysis is performed based on the patient sign knowledge graph to obtain an abnormal diagnosis result, and an alarm instruction is triggered based on the abnormal diagnosis result.

[0139] The system embodiments described above are merely illustrative, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0140] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of the embodiments or some parts of the embodiments.

[0141] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for edge device-based real-time assistance in first aid, characterized in that, The method comprises the following steps: Based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, the target switching frequency band of the edge device is determined, and multi-source data is acquired based on the target switching frequency band; the multi-source data includes patient sign data and patient image data; Based on the environmental data of the edge device, the patient sign data is corrected to obtain target sign data, and based on the environmental data, the patient image data is compensated to obtain target image data; Based on the edge computing node of the edge device, the target sign data and the target image data are fused to obtain target fusion data; Based on the target fusion data, a patient sign knowledge graph is constructed; Based on the patient sign knowledge graph, diagnostic analysis is performed to obtain an abnormal diagnosis result, and an alarm instruction is triggered based on the abnormal diagnosis result; The method comprises the following steps: The acquired electromagnetic spectrum data is divided into multiple modal data, and the multiple modal data are preprocessed respectively to obtain preprocessed multiple modal data; the multiple modal data include signal intensity modal, frequency modal and phase modal; The signal intensity modal, the frequency modal and the phase modal are respectively subjected to feature extraction to obtain signal intensity features, frequency features and phase features; Based on the signal interference spectrum, feature extraction is performed to obtain interference information features; The signal intensity features, the frequency features, the phase features and the interference information features are serially fused to obtain a comprehensive feature vector; The comprehensive feature vector is analyzed and screened to determine the target switching frequency band.

2. The edge device based real-time assisted first aid method according to claim 1, c h a r a c t e r i z e d b y, The environmental data includes temperature data, humidity data and air pressure data; The method comprises the following steps: The temperature data, the humidity data and the air pressure data are respectively subjected to normalization processing to obtain temperature normalized data, humidity normalized data and air pressure normalized data; Based on the temperature normalized data, the humidity normalized data and the air pressure normalized data, an environmental influence factor matrix is constructed according to the influence of different sign data in the patient sign data; For each sign data in the patient sign data, an environmental comprehensive influence value is determined based on the environmental influence factor matrix; The class of the patient sign data and the environmental comprehensive influence value are matched with a preset correction function library to obtain a target correction function corresponding to each sign data; Based on the target correction function, the patient sign data is corrected to obtain the target sign data.

3. The edge device based real-time assisted first aid method of claim 1, wherein, The environmental data includes environmental light intensity and environmental reflectivity; The method comprises the following steps: The spectral data of the environmental light and the spectral data of the patient image data are subjected to difference analysis to obtain a spectral difference degree; Based on the spectral difference degree, the spectral data of the patient image data are subjected to spectral matching adjustment to obtain a spectral matching adjustment image; Converting the spectrum matching adjustment image from the original image space to a CIE-LUV nonlinear space to obtain a nonlinear transformation image; Performing brightness enhancement on an L channel of the nonlinear transformation image based on the ambient light intensity, and performing color saturation and hue adjustment on U and V channels of the nonlinear transformation image based on ambient light reflectivity, to obtain an enhanced nonlinear transformation image; Converting the nonlinear transformation image back to the original image space to obtain the target image data.

4. The edge device based real-time assisted first aid method of claim 1, wherein, The edge computing node based on the edge device fuses the target sign data and the target image data to obtain target fusion data, including: Performing time series analysis on the target sign data to obtain a time series analysis result; Performing acquisition time labeling on the target image data, and performing change trend analysis based on the time labeling result to obtain a time feature trend vector; Based on the time series analysis result and the time feature trend vector, a space-time correlation model is constructed; Based on the space-time correlation model, the target sign data and the target image data are fused to obtain a fusion intermediate result, and the fusion intermediate result is smoothed to obtain the target fusion data.

5. The edge device based real-time assisted first aid method of claim 1, wherein, The patient sign knowledge graph is constructed based on the target fusion data, including: The target fusion data is preprocessed to obtain preprocessed target scale data; Based on a space-time convolutional neural network, the target scale data is feature-extracted to obtain sign change features in the time dimension and sign inter-coupling features in the space dimension, respectively; The sign change features and the sign inter-coupling features are spliced and fused to obtain a sign feature vector; The patient sign knowledge graph is constructed with the sign feature vector as a node and the correlation between signs as an edge.

6. The edge device based real-time assisted first aid method of claim 1, wherein, Based on the patient sign knowledge graph, a diagnosis analysis is performed to obtain an abnormal diagnosis result, including: Based on a preset diagnosis knowledge base and the patient sign knowledge graph, a diagnosis target path is determined; The current sign data node of the patient sign knowledge graph is taken as an initial point to search for paths matching the diagnosis target path to obtain multiple matching paths; For each path in the multiple matching paths, the reliability of the nodes and the weight of the edges on the path are determined to determine the path confidence; The path confidence is screened out to obtain a candidate diagnosis path, and the abnormal diagnosis result is determined based on the candidate diagnosis path.

7. An edge device based real-time assisted first aid system, characterized in that, The application is applied to the real-time auxiliary first-aid method based on the edge device as claimed in any one of claims 1 to 6; the real-time auxiliary first-aid system based on the edge device includes: A frequency band switching module is configured to determine a target switching frequency band of an edge device based on acquired electromagnetic spectrum data and a preset signal interference spectrum, and acquire multi-source data based on the target frequency band; the multi-source data includes patient sign data and patient image data; The target switching frequency band of the edge device is determined based on the acquired electromagnetic spectrum data and the preset signal interference spectrum, including: The acquired electromagnetic spectrum data is divided into multiple modal data, and the multiple modal data are preprocessed respectively to obtain preprocessed multiple modal data; the multiple modal data include signal intensity modal, frequency modal and phase modal; The signal intensity modal, the frequency modal and the phase modal are respectively subjected to feature extraction to obtain signal intensity features, frequency features and phase features; Based on the signal interference spectrum, interference information features are extracted; The signal intensity features, the frequency features, the phase features and the interference information features are serially fused to obtain a comprehensive feature vector; The comprehensive feature vector is analyzed and screened to determine the target switching frequency band; The correction compensation module is used for correcting the patient sign data based on the environmental data of the edge device to obtain target sign data, and compensating the patient image data based on the environmental data to obtain target image data; The data fusion module is used for fusing the target sign data and the target image data based on the edge computing node of the edge device to obtain target fusion data; The knowledge graph construction module is used for constructing a patient sign knowledge graph based on the target fusion data; The analysis and early warning module is used for diagnosing based on the patient sign knowledge graph to obtain an abnormal diagnosis result, and triggering an alarm instruction based on the abnormal diagnosis result.

8. An electronic device, comprising: The memory is used for storing a computer software program; The processor is used for reading and executing the computer software program, and when the processor executes the computer software program, the real-time auxiliary first-aid method based on the edge device in any one of claims 1 to 6 is realized.

9. A non-transitory computer readable storage medium having stored therein a computer software program, characterized in that, When the computer software program is executed by the processor, the real-time auxiliary first-aid method based on the edge device in any one of claims 1 to 6 is realized.

Images