Intelligent analysis method for curative effect of sepsis patient based on combination of scraping therapy and turbidity dissolving method

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an intelligent curative effect analysis method for a sepsis patient based on a scraping therapy and turbidity dissolving method, which comprises the following steps: acquiring a tongue picture image, a scraping therapy part image and turbidity dissolving method parameters of the sepsis patient, segmenting a tongue picture area of the tongue picture image to extract tongue picture characteristics of the sepsis patient, and calculating the curative effect of the sepsis patient according to the tongue picture characteristics; establishing a scraping therapy video sequence of the sepsis patient so as to analyze the diffusion speed of the scarf and the local capillary reaction intensity of the sepsis patient; a first channel of the trained double-channel sepsis analysis model is used for analyzing the tongue coating turbid toxin fading rule of the sepsis patient under the turbidimetry parameters; analyzing a scraping therapy part-inflammatory response topological relation of the sepsis patient by utilizing a second channel of the double-channel sepsis analysis model; and analyzing a sepsis response index of the sepsis patient to generate a scraping therapy optimization path and a turbidity dissolving prescription optimization composition of the sepsis patient. According to the invention, the accuracy of the curative effect analysis of the sepsis patient by combining the scraping therapy with the turbidity dissolving method can be improved.

Description

Technical Field

[0001] The present invention relates to an intelligent analysis method for the therapeutic effect on sepsis patients based on scraping combined with turbidity removal method, and belongs to the technical field of artificial intelligence. Background Art

[0002] Sepsis refers to a systemic inflammatory response syndrome caused by infection. Simply put, it is an excessive or uncontrolled immune response of the body to infection (such as bacteria, viruses, fungi, etc.), causing inflammation to spread throughout the body, thereby damaging its own organs and tissues.

[0003] Sepsis usually relies on TCM physicians to subjectively judge the efficacy of treatment based on their rich clinical experience and observation, smell, questioning and palpation (especially observation of the tongue and observation of scraping reactions). Physicians will evaluate whether the treatment is effective based on changes in the patient's tongue image (such as whether the thickness of the tongue coating has been reduced, whether the tongue color has improved), the color, shape and disappearance speed of the sha marks after scraping, and the patient's overall symptom improvement. This method is highly subjective, lacks quantitative standards, and is difficult to accurately track subtle changes. There may be differences in judgment between different physicians, and it is impossible to deeply reveal the complex relationship between tongue image, scraping reaction and sepsis pathophysiology, which limits the objectivity of efficacy evaluation. Summary of the invention

[0004] The present invention provides an intelligent analysis method for the efficacy of scraping combined with turbidity removal method on patients with sepsis, and its main purpose is to improve the accuracy of the analysis of the efficacy of scraping combined with turbidity removal method on patients with sepsis.

[0005] To achieve the above object, the present invention provides an intelligent analysis method for the efficacy of scraping combined with turbidity removal method on patients with sepsis, comprising: Obtaining a tongue image, a scraping site image, and a turbidity removal method parameter of a patient with sepsis, and segmenting the tongue image region of the tongue image to extract tongue image features of the patient with sepsis, wherein the tongue image features include tongue color, tongue coating thickness, and tongue coating crack index; According to the scraping site image, a scraping video sequence of the sepsis patient is established to analyze the diffusion speed of the scraping marks and the intensity of the local capillary reaction of the sepsis patient; Based on the tongue image characteristics, the first channel of the trained dual-channel sepsis analysis model is used to analyze the disappearance rule of the tongue coating turbidity of the sepsis patient under the parameters of the turbidity removal method; Based on the diffusion speed of the scraping marks and the intensity of the local capillary reaction, the second channel of the dual-channel sepsis analysis model is used to analyze the topological relationship between the scraping site and the inflammatory reaction of the sepsis patient; Analyze the sepsis response index of the sepsis patient in combination with the law of the disappearance of tongue coating turbidity toxin and the topological relationship between the scraping site and the inflammatory response to generate the optimized scraping path and the optimized composition of the turbidity-resolving prescription for the sepsis patient.

[0006] Optionally, the segmentation of the tongue image region includes: Perform bilateral filtering on the tongue image to obtain a filtered tongue image; And convert the filtered tongue image into an HSV tongue image; Locate the initial tongue body region of the HSV tongue image; Define the aspect ratio of the length to width of the tongue body in the tongue image, and use the preset YOLOv7 algorithm to segment the initial tongue body region to obtain the tongue image region.

[0007] Optionally, the extraction of the tongue image features of the sepsis patient includes:

[0008] Extract the channel histogram of the corresponding tongue image region of the sepsis patient; Establish a corresponding rule base for the color gamut and syndrome type of the sepsis patient, and output the tongue color card code of the sepsis patient according to the channel histogram; Map the tongue color corresponding to the tongue color card code point; Extract the thickness gradient map of the tongue image region to determine the tongue coating thickness of the tongue image region; Identify the crack edge response of the tongue image region to analyze the crack index of the tongue image region; Combine the tongue color, the tongue coating thickness, and the crack index to determine the tongue image features of the sepsis patient.

[0009] Optionally, the analysis of the crack index of the tongue image region includes: Identify the crack region of the tongue image region, and analyze the main crack length and the branch crack length of the crack region; Based on the crack edge response corresponding to the tongue image region, the main crack length, and the branch crack length, calculate the crack index of the crack region.

[0010] Optionally, the establishment of the scraping video sequence of the sepsis patient according to the scraping site image includes: Mark the reference points of the scraping site image to calibrate the scraping site image to obtain a calibrated scraping site image; Identify the position of the scraping board in the scraping site image, and generate the scraping path of the scraping site image through the position of the scraping board; Based on the scraping path, establish the scraping video sequence of the sepsis patient.

[0011] Optionally, the analyzing the diffusion speed of the sha marks and the intensity of the local capillary reaction of the sepsis patient includes: Analyzing the optical flow amplitude of the scraping video sequence corresponding to the sepsis patient; Extracting the diffusion contour of the sha marks of the sepsis patient according to the optical flow amplitude; Analyzing the centroid displacement of the sha scar diffusion contour to analyze the sha scar diffusion speed of the sepsis patient; Calculating the erythema index and blood oxygen saturation of the sepsis patient according to the scraping video sequence; The local capillary reaction intensity of the sepsis patient is analyzed by combining the erythema index and blood oxygen saturation.

[0012] Optionally, based on the tongue image features, using the first channel of the trained dual-channel sepsis analysis model to analyze the disappearance rule of tongue coating turbidity of the sepsis patient under the turbidity removal method parameters includes: Splicing the tongue image feature and the turbidity removal method parameter to obtain a splicing vector; Based on the splicing vector, analyzing the attenuation rate of turbidity concentration of the sepsis patient using the first channel of the dual-channel sepsis analysis model; Constructing a turbidity kinetic curve of the turbidity concentration decay rate; The disappearance pattern of turbid toxins in the tongue coating of the sepsis patient is analyzed by the turbid toxins dynamics curve.

[0013] Optionally, based on the diffusion speed of the scraping marks and the intensity of the local capillary reaction, the second channel of the dual-channel sepsis analysis model is used to analyze the scraping site-inflammatory reaction topological relationship of the sepsis patient, including: Establishing a scraping site surface model of the sepsis patient; marking the scraped positions of the septic patient on the scraped position surface model, and encoding the scraped positions to obtain encoded scraped positions; According to the diffusion speed of the scraping marks, the intensity of the local capillary reaction and the coded scraping positions, the second channel is used to analyze the position-inflammatory reaction correlation coefficient of the sepsis patient; Based on the position-inflammatory response correlation coefficient, the scraping position-inflammatory response topological relationship of the sepsis patient is defined.

[0014] Optionally, the combining of the tongue coating turbidity disappearance rule and the scraping site-inflammatory response topological relationship to analyze the sepsis response index of the sepsis patient includes: Based on the disappearance rule of the turbid tongue coating, analyzing the instantaneous change rate of sepsis in the sepsis patient; Perform a spatial integration on the topological relationship between the scraping site and the inflammatory response to obtain an inflammatory topological integral; Define the weights of the instantaneous sepsis change rate and the inflammatory topological integral to obtain the instantaneous sepsis change weight and the inflammatory topological integral weight; Combine the instantaneous sepsis change rate, the inflammatory topological integral, the instantaneous sepsis change weight, and the inflammatory topological integral weight to analyze the sepsis response index of the sepsis patient.

[0015] Optionally, the performing a spatial integration on the topological relationship between the scraping site and the inflammatory response to obtain an inflammatory topological integral includes: Grid the topological relationship between the scraping site and the inflammatory response to obtain a grid topological map of the scraping site - inflammatory response; Define the meridian weight coefficient corresponding to the scraping nodes of the grid topological map of the scraping site - inflammatory response; Calculate the IRM spatial gradient of the grid topological map of the scraping site - inflammatory response; Based on the meridian weight coefficient of the scraping nodes and the IRM spatial gradient, calculate the inflammatory topological integral of the grid topological map of the scraping site - inflammatory response, represents the left boundary of the corresponding scraping area of the grid topological map of the scraping site - inflammatory response in the axis direction, represents the right boundary of the corresponding scraping area of the grid topological map of the scraping site - inflammatory response in the axis direction, represents the starting point in the longitudinal direction of the corresponding scraping area of the grid topological map of the scraping site - inflammatory response, represents the ending point in the longitudinal direction of the corresponding scraping area of the grid topological map of the scraping site - inflammatory response, represents the number of scraping nodes corresponding to the grid topological map of the scraping site - inflammatory response.

[0016] To solve the above problems, the present invention also provides an electronic device, which includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to implement the above-mentioned intelligent analysis method for the curative effect of sepsis patients based on scraping combined with the method of resolving turbidity.

[0017] To solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned intelligent analysis method for the curative effect of sepsis patients based on scraping combined with the method of resolving turbidity.

[0018] Compared with the problems described in the background technology, firstly, this method captures the subtle changes in the turbidity and toxin status and local inflammatory response in the patient's body through tongue image features (tongue color, tongue coating thickness, crack index) and scraping site video analysis (sha mark diffusion speed, capillary reaction intensity). This information is difficult to obtain by traditional methods, especially the analysis of the disappearance rules of tongue coating turbidity and toxin and the topological relationship between scraping site and inflammatory response, which reveals the inherent mechanism of disease evolution and the complex effects of scraping intervention, making the evaluation more in-depth and comprehensive. Secondly, at the treatment level, this method significantly improves intervention measures. The accuracy and individual level of sepsis are complex and changeable, and the treatment needs to be individualized. Through the dual-channel sepsis analysis model, the change law of tongue coating is combined with the spatial relationship of inflammatory response caused by scraping. The calculated sepsis response index can dynamically reflect the patient's immediate response to the current treatment (including scraping and turbidity-clearing prescriptions). The scraping optimization path generated based on this can guide the clinic to select more effective scraping sites, sequences and strengths, and maximize the guidance of inflammation dissipation; and the optimized composition of the turbidity-clearing prescription can flexibly adjust the prescription and drug compatibility according to the real-time turbidity state to ensure that the drug power reaches the diseased area directly. Therefore, the present invention can improve the accuracy of the efficacy analysis of scraping combined with turbidity-clearing method for sepsis patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flow chart of an intelligent analysis method for the efficacy of scraping combined with turbidity removal method on patients with sepsis provided by one embodiment of the present invention; Figure 2 A schematic diagram of tongue image region determination in an intelligent analysis method for the efficacy of scraping combined with turbidity removal method on sepsis patients provided in one embodiment of the present invention; The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0020] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0021] The embodiment of the present application provides an intelligent analysis method for the efficacy of scraping combined with turbidity removal for patients with sepsis. The execution subject of the intelligent analysis method for the efficacy of scraping combined with turbidity removal for patients with sepsis includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided in the embodiment of the present application. In other words, the intelligent analysis method for the efficacy of scraping combined with turbidity removal for patients with sepsis can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0022] Embodiment 1: Referring to Figure 1 as shown, it is a schematic flowchart of an intelligent analysis method for the curative effect of sepsis patients based on scraping combined with the method of resolving turbidity provided by an embodiment of the present invention. In this embodiment, the intelligent analysis method for the curative effect of sepsis patients based on scraping combined with the method of resolving turbidity includes: S1. Obtain the tongue image, scraping site image and turbidity-resolving method parameters of the sepsis patient, and segment the tongue region of the tongue image to extract the tongue features of the sepsis patient. Among them, the tongue features include the color of the tongue body, the thickness of the tongue coating, and the tongue coating crack index.

[0023] It should be explained that the tongue image refers to a standardized photo of the patient's tongue taken by a specific device (such as a digital camera, tongue image acquisition instrument, etc.), the scraping site image refers to a photo or video of the skin surface changes in the scraping operation area, and the turbidity-resolving method parameters refer to the specific implementation details and data related to the traditional Chinese medicine treatment method of "resolving turbidity". For example, the components of traditional Chinese medicine and the dosage of administration.

[0024] The segmentation of the tongue region of the tongue image in the present invention can provide data support for subsequent tongue image analysis.

[0025] Specifically, the segmentation of the tongue region of the tongue image includes: Perform bilateral filtering on the tongue image to obtain a filtered tongue image; And convert the filtered tongue image into an HSV tongue image; Locate the initial tongue body region of the HSV tongue image; Define the aspect ratio of the length to width of the tongue body of the tongue image, and use the preset YOLOv7 algorithm to segment the initial tongue body region to obtain the tongue region.

[0026] Among them, the filtered tongue image refers to a new image obtained by applying the bilateral filtering algorithm to the original tongue image. Bilateral filtering can not only smooth the image (remove noise), but also well protect the edge information of the image. The HSV tongue image refers to an image obtained by converting the image (filtered tongue image) after bilateral filtering processing from the common RGB color space to the HSV color space. The initial tongue body region refers to the position outlined in the HSV tongue image and considered to be the possible location of the tongue. The aspect ratio of the length to width of the tongue body refers to the typical aspect ratio (length / width) of the tongue in a normal or specific state obtained based on medical knowledge or statistical analysis of a large number of samples. For example, a typical aspect ratio may be 1.5:1 or 2:1. The YOLOv7 algorithm refers to an algorithm trained with a large number of tongue images for performing tongue image segmentation tasks. The tongue region refers to the final result obtained by accurately segmenting through the above YOLOv7 algorithm, that is, the accurate pixel region of the tongue part in the image.

[0027] Optionally, the initial tongue body region for positioning the HSV tongue image can be roughly outlined by color thresholding, region growing, simple geometric shape detection, or other image processing techniques.

[0028] Optionally, the aspect ratio of the tongue body of the tongue image is defined to segment the initial tongue body region using the preset YOLOv7 algorithm, specifically referring to Figure 2 the schematic diagram of tongue image region determination for the intelligent analysis method of the curative effect of scraping combined with turbidity-removing method on sepsis patients provided in an embodiment of the present invention: Among them, the Anchor size is optimized by judging whether the length and width of the tongue body are greater than 2.5 to improve the detection accuracy of the tongue body region.

[0029] The tongue image features extracted by the present invention can be used as the data basis for the analysis of the sepsis manifestations of sepsis patients.

[0030] Specifically, the extraction of the tongue image features of the sepsis patient includes:

[0031] Extracting the channel histogram of the corresponding tongue image region of the sepsis patient; Establishing a color gamut-syndrome type correspondence rule base for the sepsis patient to output the tongue color card code of the sepsis patient according to the channel histogram; Mapping the tongue color of the tongue color card code point; Extracting the thickness gradient map of the tongue image region to determine the tongue coating thickness of the tongue image region; Identifying the crack edge response of the tongue image region to analyze the crack index of the tongue image region; Combining the tongue color, the tongue coating thickness, and the crack index to determine the tongue image features of the sepsis patient.

[0032] Among them, the channel histogram refers to a chart where the horizontal axis represents pixel values (such as 0 - 255), and the vertical axis represents the number of pixels with that pixel value. The color gamut - syndrome type correspondence rule base refers to a rule base used to map objective color data to the classifications of traditional Chinese medicine diagnosis. Exemplarily, such as H ∈ [0, 15] → pale tongue → yang deficiency syndrome, H ∈ [15, 30] → red tongue → excess heat syndrome. The tongue texture color card encoding refers to the encoding that describes the color of the tongue texture (such as light red, dark red, purple, crimson, etc.). For example, #FFB6C1 → "light purple". The tongue texture color refers to the specific, understandable, and visually representable tongue texture color converted according to its definition or encoding standard. The thickness gradient map refers to a map that reflects the relative thickness differences of the tongue coating at different positions. The tongue coating thickness refers to the quantification of the overall thickness of the tongue coating. The crack edge response refers to the response intensity to the specific texture pattern of cracks in the tongue image. The crack index refers to a measure of the severity of cracks in the tongue image area.

[0033] Optionally, the establishment of the color gamut - syndrome type correspondence rule base for sepsis patients is constructed by systematically sorting out the association rules between different tongue texture colors (such as pale, red, crimson, purple, blue, etc.) and different syndrome types (such as qi deficiency, blood stasis, heat toxin exuberance, yin injury, yang deficiency, etc.) according to traditional Chinese medicine theories (such as "Huangdi Neijing", "Treatise on Febrile and Miscellaneous Diseases", etc.).

[0034] Furthermore, the analysis of the crack index of the tongue image area includes: Identifying the crack area in the tongue image area and analyzing the main crack length and branch crack length of the crack area; Based on the crack edge response corresponding to the tongue image area, the main crack length, and the branch crack length, calculate the crack index of the crack area using the following formula; , Where, represents the crack index of the crack area, represents the total pixel area of the crack area, represents the total pixel area of the tongue image area corresponding to the crack area, represents the area ratio weight, represents the pixel point 's crack edge response, represents the total number of pixels in the crack area, represents the edge response weight, represents the main crack length, represents the branch crack length, represents the crack complexity weight.

[0035] Among them, the crack area refers to a set of pixels that are identified in the tongue image area and show crack-like (crack-like texture) features. The main crack length refers to the length of the longest main crack line in the identified crack area. The branch crack length refers to the total length of the smaller crack lines branching from the main crack or other larger cracks. The total pixel area refers to the total area of the pixels contained in the identified crack area. The area proportion weight refers to the weight coefficient assigned to the factor "the proportion of the crack area to the tongue area" when calculating the crack index. It indicates the relative importance of the area proportion in the comprehensive evaluation of the severity of the crack. The edge response weight refers to the weight coefficient assigned to the "clarity of the crack edge" when calculating the crack index. It indicates the relative importance of the edge feature in the comprehensive evaluation of the severity of the crack. The crack complexity weight refers to the weight coefficient assigned to the factor "complexity of the crack" when calculating the crack index. It indicates the relative importance of the structural complexity of the crack in the comprehensive evaluation.

[0036] S2. Establishing a scraping video sequence of the sepsis patient based on the scraping site image to analyze the diffusion speed of the scraping marks and the local capillary reaction intensity of the sepsis patient.

[0037] It should be explained that the scraping video sequence refers to a dynamic video clip reorganized or generated based on a series of scraping site images (which may be continuous frames or photos taken from multiple angles) taken during the scraping process. It can more comprehensively and continuously capture and record the implementation process of the scraping intervention measure in sepsis patients and their directly visible physiological reactions (such as the formation of sha marks).

[0038] In detail, the method of establishing the scraping video sequence of the sepsis patient according to the scraping site image includes: Marking the reference points of the scraping part image to calibrate the scraping part image to obtain a calibrated scraping part image; Identifying the scraping board position of the scraping area image, and generating the scraping path of the scraping area image based on the scraping board position; Based on the scraping path, a scraping video sequence of the sepsis patient is established.

[0039] Among them, the reference point refers to a point with a stable spatial position or anatomical significance. It can be a specific bone protrusion (such as a spinal spinous process, bony landmark), a fixed mark on the skin (such as a sticker), or other immovable feature points in the image. The calibrated scraping site image refers to the scraping site image after geometric correction. The scraping board position refers to the physical position where the scraping tool (scraping board) is located in each scraping site image. The scraping path refers to the trajectory of the position of the scraping board changing over time during the scraping process. The scraping video sequence refers to a dynamic video showing the scraping operation process.

[0040] Optionally, the calibration of the scraping site image to obtain the calibrated scraping site image can be achieved by image processing techniques (such as perspective transformation, affine transformation, etc.) to eliminate or correct these geometric distortions.

[0041] The present invention analyzes the scar diffusion speed and local capillary reaction intensity of the sepsis patient to analyze the impact of scraping on the sepsis patient.

[0042] Specifically, the analysis of the scar diffusion speed and local capillary reaction intensity of the sepsis patient includes: Analyzing the optical flow amplitude of the corresponding scraping video sequence of the sepsis patient; Extracting the scar diffusion contour of the sepsis patient according to the optical flow amplitude; Analyzing the centroid displacement of the scar diffusion contour to analyze the scar diffusion speed of the sepsis patient; Calculating the erythema index and blood oxygen saturation of the sepsis patient according to the scraping video sequence; Combining the erythema index and blood oxygen saturation to analyze the local capillary reaction intensity of the sepsis patient.

[0043] Among them, the optical flow amplitude refers to the speed and distance of pixel points moving between adjacent frames. The scar diffusion contour refers to the boundary line of the ecchymosis (scar) area formed after scraping spreading or changing outward over time. The centroid displacement refers to the amount of position change of the centroid point of the scar diffusion contour between different time points in the video sequence. The scar diffusion speed refers to the rate at which the scar area spreads outward. The erythema index refers to an index quantifying the degree of skin surface flushing. The blood oxygen saturation refers to the proportion of oxyhemoglobin in total hemoglobin in the blood. The local capillary reaction intensity refers to the degree of reaction of the patient's local skin capillaries after scraping stimulation.

[0044] Optionally, the optical flow amplitude of the scraping video sequence corresponding to the sepsis patient can be calculated directly based on the brightness change of pixels between image frames using algorithms such as Lucas-Kanade, Horn-Schunck, and Farneback.

[0045] Optionally, the scar diffusion speed of the sepsis patient can be analyzed by calculating the ratio of the centroid displacement to the corresponding time interval.

[0046] S3. Based on the tongue image features, use the first channel of the trained dual-channel sepsis analysis model to analyze the law of the disappearance of tongue coating turbidity toxin in the sepsis patient under the Huazhuo method parameters.

[0047] Based on the tongue image features, the present invention uses the first channel of the trained dual-channel sepsis analysis model to analyze the law of the disappearance of tongue coating turbidity toxin in the sepsis patient under the Huazhuo method parameters, and deeply analyzes the law of the disappearance of tongue coating turbidity toxin in the sepsis patient over time, providing data support for evaluating the efficacy of the Huazhuo method and optimizing the treatment plan.

[0048] Specifically, the method of using the first channel of the trained dual-channel sepsis analysis model to analyze the law of the disappearance of tongue coating turbidity toxin in the sepsis patient under the Huazhuo method parameters includes: Concatenate the tongue image features and the Huazhuo method parameters to obtain a concatenated vector; Based on the concatenated vector, use the first channel of the dual-channel sepsis analysis model to analyze the attenuation rate of the turbidity toxin concentration in the sepsis patient; Construct a turbidity toxin kinetic curve of the attenuation rate of the turbidity toxin concentration; Through the turbidity toxin kinetic curve, analyze the law of the disappearance of tongue coating turbidity toxin in the sepsis patient.

[0049] Among them, the splicing vector refers to connecting data from two different sources (here, tongue image features and turbidity-resolving method parameters) in the feature dimension to form a longer and single feature vector. The dual-channel sepsis analysis model refers to a machine learning model containing two independent processing paths (channels), which is specifically designed to analyze sepsis-related data and is trained based on a large amount of tongue image feature data and scraping data. In the first channel of the dual-channel sepsis analysis model, it is the channel that is specifically responsible for processing tongue image data and extracting relevant features. The turbidity-toxicity concentration decay rate refers to the speed at which the "turbidity-toxicity" degree (concentration) reflected by tongue image features decreases per unit time. The turbidity-toxicity kinetics curve refers to a curve plotted with time as the abscissa and turbidity-toxicity concentration (or some quantitative representation thereof) as the ordinate. The law of turbidity-toxicity disappearance on the tongue coating refers to the typical pattern of the turbidity-toxicity state on the tongue coating disappearing over time revealed by analyzing the turbidity-toxicity kinetics curve. For example, a rapid decrease may indicate a significant effect of the turbidity-resolving method; a slow decrease may indicate a severe condition or poor response to treatment.

[0050] Optionally, the splicing of the tongue image features and the turbidity-resolving method parameters to obtain the splicing vector can be performed through a feature fusion function.

[0051] S4. Based on the scar diffusion speed and the local capillary reaction intensity, use the second channel of the dual-channel sepsis analysis model to analyze the topological relationship between the scraping site and the inflammatory response of the sepsis patient.

[0052] Based on the scar diffusion speed and the local capillary reaction intensity, the present invention uses the second channel of the dual-channel sepsis analysis model to analyze the topological relationship between the scraping site and the inflammatory response of the sepsis patient, which can construct and visualize a network structure describing how different scraping site stimulations topologically affect the inflammatory response, so as to deeply understand the mechanism of action and site specificity of scraping intervention in the treatment of sepsis.

[0053] Specifically, the analysis of the topological relationship between the scraping site and the inflammatory response of the sepsis patient based on the scar diffusion speed and the local capillary reaction intensity using the second channel of the dual-channel sepsis analysis model includes: Establish a surface model of the scraping site of the sepsis patient; Mark the scraped positions of the sepsis patient on the surface model of the scraping site and encode the scraped positions to obtain encoded scraped positions; According to the scar diffusion speed, the local capillary reaction intensity, and the encoded scraped positions, use the second channel to analyze the position-inflammatory response correlation coefficient of the sepsis patient; Define the topological relationship between the scraping site and the inflammatory response of the sepsis patient based on the location-inflammation response correlation coefficient.

[0054] Among them, the surface model of the scraping site refers to a three-dimensional or two-dimensional geometric model established according to the anatomical structure of the patient's body surface, especially the area where scraping is performed. The scraped position refers to the specific point or area on the patient's body where scraping has actually been performed. The encoded scraped position refers to the digital representation of the scraped position marked on the surface model. The location-inflammation response correlation coefficient refers to the strength of the association between a specific scraping position (and the local response it causes) and the patient's local inflammatory response. The topological relationship between the scraping site and the inflammatory response refers to a structured and networked relationship representation established on the surface model of the scraping site based on the calculated "location-inflammation response correlation coefficient".

[0055] Optionally, the surface model of the scraping site of the sepsis patient can be established based on an accurate model reconstructed from medical images (such as CT, MRI).

[0056] Optionally, based on the location-inflammation response correlation coefficient, defining the topological relationship between the scraping site and the inflammatory response of the sepsis patient can be constructed by using graph theory and network analysis, with the encoded scraped position as nodes, the node associations as edges, and the location-inflammation response correlation coefficient as the edge weights.

[0057] S5. Combine the law of the subsidence of tongue coating turbidity toxin and the topological relationship between the scraping site and the inflammatory response to analyze the sepsis response index of the sepsis patient, so as to generate the optimized scraping path and the optimized composition of the turbidity-removing prescription for the sepsis patient.

[0058] The present invention combines the law of the subsidence of tongue coating turbidity toxin and the topological relationship between the scraping site and the inflammatory response to analyze the sepsis response index of the sepsis patient, which can determine the elimination of sepsis in the sepsis patient, thereby improving the reliability of later optimization.

[0059] Specifically, combining the law of the subsidence of tongue coating turbidity toxin and the topological relationship between the scraping site and the inflammatory response to analyze the sepsis response index of the sepsis patient includes: Analyze the instantaneous change rate of sepsis in the sepsis patient based on the law of the subsidence of tongue coating turbidity toxin; Perform spatial integration on the topological relationship between the scraping site and the inflammatory response to obtain the inflammatory topological integral; Define the weights of the instantaneous change rate of sepsis and the inflammatory topological integral to obtain the weight of the instantaneous change of sepsis and the weight of the inflammatory topological integral; Analyze the sepsis response index of the sepsis patient in combination with the sepsis instantaneous change rate, the inflammatory topological integral, the sepsis instantaneous change weight, and the inflammatory topological integral weight.

[0060] Among them, the sepsis instantaneous change rate refers to the change rate of the overall state of the sepsis patient at a specific time point relative to the previous moment. The inflammatory topological integral refers to the overall intensity that quantifies the correlation between all current scraping sites and the inflammatory response. The sepsis instantaneous change weight refers to the relative importance of the sepsis instantaneous change rate in the sepsis response analysis. The inflammatory topological integral weight refers to the relative importance of the inflammatory topological integral in the sepsis response analysis. The sepsis response index refers to the degree of response of the overall sepsis state of the sepsis patient at a specific time point or time period, based on the instantaneous change rate of the tongue coating turbidity toxin and the topological integral performance of the inflammatory response under scraping intervention.

[0061] Furthermore, performing a spatial integration on the scraping site - inflammatory response topological relationship to obtain the inflammatory topological integral includes: Grid the scraping site - inflammatory response topological relationship to obtain a grid scraping site - inflammatory response topological map; Define the meridian weight coefficient corresponding to the scraping nodes of the grid scraping site - inflammatory response topological map; Calculate the IRM spatial gradient of the grid scraping site - inflammatory response topological map; Based on the meridian weight coefficient of the scraping nodes and the IRM spatial gradient, use the following formula to calculate the inflammatory topological integral of the grid scraping site - inflammatory response topological map: , Among them, represents the inflammatory topological integral of the grid scraping site - inflammatory response topological map, represents the node inflammatory response value of the th scraping node of the grid scraping site - inflammatory response topological map, represents the grid cell area of the th scraping node of the grid scraping site - inflammatory response topological map, represents the meridian weight coefficient of the th scraping node of the grid scraping site - inflammatory response topological map, represents the IRM spatial gradient of the th scraping node of the grid scraping site - inflammatory response topological map in the axis direction, represents the IRM spatial gradient of the th scraping node of the grid scraping site - inflammatory response topological map in the axis direction, Indicates the grid scraping site-inflammatory response topology map corresponding to the scraping area The left boundary of the axis direction, Indicates the grid scraping site-inflammatory response topology map corresponding to the scraping area The right boundary of the axis direction, Indicates the grid scraping site-inflammatory response topology map corresponding to the vertical starting point of the scraping area, Indicates the vertical end point of the scraping area corresponding to the grid scraping site-inflammatory response topology map. Indicates the number of scraping nodes corresponding to the grid scraping site-inflammatory response topology map.

[0062] Among them, the grid scraping site-inflammatory response topology map refers to a graphical representation after spatial gridding processing based on the original "scraping site-inflammatory response topology relationship", the meridian weight coefficient refers to the importance of the meridian or specific site where the node is located in traditional Chinese medicine theory, the IRM spatial gradient refers to the rate of change of the inflammatory response value in the grid scraping site-inflammatory response topology map space, the node inflammatory response value refers to the degree of inflammatory change caused by the scraping operation at the node, and the grid unit area refers to the area of the grid unit corresponding to the i-th scraping node.

[0063] Finally, the present invention generates the scraping optimization path and the optimized composition of the anti-turbidity prescription for the sepsis patient, which together constitute an individualized and dynamic comprehensive intervention strategy for sepsis patients. Among them, the scraping optimization path refers to the scraping operation scheme that is determined after comprehensive optimization based on the patient's specific condition, constitution, dynamic changes in the current sepsis response index, the law of tongue coating turbidity and toxins disappearance, and the scraping site-inflammatory response topological relationship analysis results, which can most effectively guide the dissipation of inflammation, promote the discharge of turbidity and toxins, and improve the overall state. For example, for sepsis patients, the sepsis response index analysis shows that the inflammatory topological integral (ITI) is higher in the lower jiao area (such as the foot Shaoyin kidney meridian and the foot Taiyin spleen meridian circulation area), and the tongue coating is thick and greasy at the root and yellow in color. The scraping topological relationship analysis suggests that scraping the inner side of the lower limbs (corresponding to the spleen and kidney meridians) can effectively reduce ITI. Therefore, the scraping optimization path focuses on scraping the inner side of the calf (the spleen meridian of the foot Taiyin, such as near Sanyinjiao and Yinlingquan) and the inner side of the heel (the kidney meridian of the foot Shaoyin, such as near Taixi). The optimized composition of the turbidity-eliminating prescription refers to the traditional Chinese medicine prescription determined after comprehensive optimization that can most effectively remove turbid toxins from the body, regulate qi and blood, and balance yin and yang.

[0064] First, this method captures the subtle changes in the patient's body turbidity and toxins and local inflammatory response through tongue image features (tongue color, tongue coating thickness, crack index) and scraping site video analysis (sha mark diffusion speed, capillary reaction intensity). This information is difficult to obtain with traditional methods, especially the analysis of the disappearance pattern of tongue coating turbidity and toxins and the topological relationship between scraping site and inflammatory response, which reveals the inherent mechanism of disease evolution and the complex effects of scraping intervention, making the evaluation more in-depth and comprehensive. Secondly, at the treatment level, this method significantly improves the accuracy and individuality of intervention measures. At the level of physical treatment, sepsis is complex and changeable, and treatment needs to be individualized. Through the dual-channel sepsis analysis model, the change pattern of tongue coating is combined with the spatial relationship of inflammatory response caused by scraping, and the calculated sepsis response index can dynamically reflect the patient's immediate response to the current treatment (including scraping and turbidity-clearing prescriptions). The scraping optimization path generated based on this can guide the clinic to select more effective scraping sites, sequences and strengths, and maximize the guidance of inflammation dissipation; and the optimized composition of the turbidity-clearing prescription can flexibly adjust the prescription and drug compatibility according to the real-time turbidity state to ensure that the drug power reaches the diseased area directly. Therefore, the present invention can improve the accuracy of the efficacy analysis of scraping combined with turbidity-clearing method for sepsis patients.

Claims

1. An intelligent analysis method for the curative effect of scraping combined with the method of resolving turbidity on sepsis patients, characterized in that, The method includes: Obtain the tongue image, scraping site image and turbidity-removing method parameters of the sepsis patient, segment the tongue region of the tongue image to extract the tongue image features of the sepsis patient, where the tongue image features include tongue color, tongue coating thickness and tongue coating crack index; According to the scraping site image, establish the scraping video sequence of the sepsis patient to analyze the scar diffusion speed and local capillary reaction intensity of the sepsis patient; Based on the tongue image features, use the first channel of the trained dual-channel sepsis analysis model to analyze the rule of turbidity toxin regression of the tongue coating of the sepsis patient under the turbidity-removing method parameters; Based on the scar diffusion speed and the local capillary reaction intensity, use the second channel of the dual-channel sepsis analysis model to analyze the topological relationship between the scraping site and the inflammatory reaction of the sepsis patient; Combine the rule of turbidity toxin regression of the tongue coating and the topological relationship between the scraping site and the inflammatory reaction to analyze the sepsis response index of the sepsis patient, so as to generate the optimized scraping path and the optimized composition of the turbidity-removing prescription of the sepsis patient.

2. The intelligent analysis method for the curative effect of scraping combined with turbidity-removing method on sepsis patients according to claim 1, characterized in that, The segmenting the tongue region of the tongue image includes: Perform bilateral filtering on the tongue image to obtain a filtered tongue image; And convert the filtered tongue image into an HSV tongue image; Locate the initial tongue body region of the HSV tongue image; Define the aspect ratio of the tongue body of the tongue image, and use the preset YOLOv7 algorithm to segment the initial tongue body region to obtain the tongue region.

3. The intelligent analysis method for the curative effect of scraping combined with turbidity-removing method on sepsis patients according to claim 2, wherein The extracting the tongue image features of the sepsis patient includes: Extract the channel histogram of the corresponding tongue region of the sepsis patient; Establish a color gamut-syndrome type correspondence rule base for the sepsis patient to output the tongue color card code of the sepsis patient according to the channel histogram; Map the tongue color of the tongue color card coding point; Extract the thickness gradient map of the tongue region to determine the tongue coating thickness of the tongue region; Identify the crack edge response of the tongue region to analyze the crack index of the tongue region; Combine the tongue color, the tongue coating thickness and the crack index to determine the tongue image features of the sepsis patient.

4. The intelligent analysis method for the curative effect of scraping combined with turbidity-removing method on sepsis patients according to claim 3, characterized in that, The analyzing the crack index of the tongue region includes: Identify the crack region of the tongue region and analyze the main crack length and branch crack length of the crack region; Based on the crack edge response corresponding to the tongue region, the main crack length and the branch crack length, calculate the crack index of the crack region.

5. The intelligent analysis method for the curative effect of scraping combined with turbidity-resolving method on sepsis patients according to claim 4, characterized in that, The establishing the scraping video sequence of the sepsis patient according to the scraping site image includes: Mark the reference points of the scraping site image to calibrate the scraping site image to obtain a calibrated scraping site image; Identify the position of the scraping board in the scraping site image, and generate the scraping path of the scraping site image through the position of the scraping board; Based on the scraping path, establish the scraping video sequence of the sepsis patient.

6. The intelligent analysis method for the curative effect of scraping combined with turbidity-resolving method on sepsis patients according to claim 5, wherein, The analyzing the scar diffusion speed and local capillary reaction intensity of the sepsis patient includes: Analyze the optical flow amplitude of the corresponding scraping video sequence of the sepsis patient; Extract the scar diffusion contour of the sepsis patient according to the optical flow amplitude; Analyze the centroid displacement of the scar diffusion contour to analyze the scar diffusion speed of the sepsis patient; Calculate the erythema index and blood oxygen saturation of the sepsis patient according to the scraping video sequence; Analyze the local capillary reaction intensity of the sepsis patient by combining the erythema index and blood oxygen saturation.

7. The intelligent analysis method for the curative effect of scraping combined with turbidity-removing method on sepsis patients according to claim 6, characterized in that, Based on the tongue image features, analyze the law of the subsidence of the turbid toxin on the tongue coating of the sepsis patient under the Huazhuo method parameters by using the first channel of the trained dual-channel sepsis analysis model, including: Concatenate the tongue image features and the Huazhuo method parameters to obtain a concatenated vector; Based on the concatenated vector, analyze the attenuation rate of the turbid toxin concentration of the sepsis patient by using the first channel of the dual-channel sepsis analysis model; Construct a turbid toxin kinetic curve of the turbid toxin concentration attenuation rate; Analyze the law of the subsidence of the turbid toxin on the tongue coating of the sepsis patient through the turbid toxin kinetic curve.

8. The intelligent analysis method for the curative effect of scraping combined with turbidity-resolving method on sepsis patients according to claim 7, wherein, Based on the scar diffusion speed and the local capillary reaction intensity, analyze the topological relationship between the scraping site and the inflammatory reaction of the sepsis patient by using the second channel of the dual-channel sepsis analysis model, including: Establish a surface model of the scraping site of the sepsis patient; Mark the scraped positions of the sepsis patient on the surface model of the scraping site and encode the scraped positions to obtain encoded scraped positions; According to the scar diffusion speed, the local capillary reaction intensity, and the encoded scraped positions, analyze the position-inflammation reaction correlation coefficient of the sepsis patient by using the second channel; Define the topological relationship between the scraping site and the inflammatory reaction of the sepsis patient based on the position-inflammation reaction correlation coefficient.

9. The intelligent analysis method for the curative effect of scraping combined with turbidity-resolving method on sepsis patients according to claim 8, characterized in that, Combining the law of the subsidence of the turbid toxin on the tongue coating and the topological relationship between the scraping site and the inflammatory reaction, analyze the sepsis response index of the sepsis patient, including: Analyze the instantaneous change rate of sepsis of the sepsis patient based on the law of the subsidence of the turbid toxin on the tongue coating; Perform a spatial integration on the topological relationship between the scraping site and the inflammatory reaction to obtain an inflammatory topological integral; Define the weights of the instantaneous change rate of sepsis and the inflammatory topological integral to obtain the instantaneous change weight of sepsis and the inflammatory topological integral weight; Analyze the sepsis response index of the sepsis patient by combining the instantaneous change rate of sepsis, the inflammatory topological integral, the instantaneous change weight of sepsis, and the inflammatory topological integral weight.

10. The intelligent analysis method for the curative effect of scraping combined with turbidity-removing method on sepsis patients as described in claim 9, characterized in that, The performing a spatial integration on the topological relationship between the scraping site and the inflammatory reaction to obtain an inflammatory topological integral includes: Grid the topological relationship between the scraping site and the inflammatory reaction to obtain a grid topological map of the scraping site and the inflammatory reaction; Define the meridian weight coefficient corresponding to the scraping nodes of the grid topological map of the scraping site and the inflammatory reaction; Calculate the IRM spatial gradient of the grid topological map of the scraping site and the inflammatory reaction; Calculate the inflammatory topology integral of the grid scraping site - inflammatory response topological map based on the meridian weight coefficient of the scraping nodes and the IRM spatial gradient. Indicates the left boundary of the corresponding scraping area of the grid scraping site - inflammatory response topological map in the axis direction. Indicates the right boundary of the corresponding scraping area of the grid scraping site - inflammatory response topological map in the axis direction. Indicates the starting point in the longitudinal direction of the corresponding scraping area of the grid scraping site - inflammatory response topological map. Indicates the ending point in the longitudinal direction of the corresponding scraping area of the grid scraping site - inflammatory response topological map. Indicates the number of scraping nodes corresponding to the grid scraping site - inflammatory response topological map.

Images