Multi-mode-based traditional Chinese medicine external curative effect fruit detection method and system

Abstract

The invention relates to a multi-mode-based traditional Chinese medicine external curative effect fruit detection method and system, and belongs to the technical field of medical detection. The method comprises the following steps: acquiring infrared thermal imaging, laser speckle blood perfusion, skin conductance and body surface temperature signals of a traditional Chinese medicine external application area of a subject, performing time synchronization, noise suppression and spatial registration, and inputting the signals into a multi-modal fusion model to generate a fusion feature vector; calling preset physiological response reference data, comparing the fused feature vector with the first comparison result data and the second comparison result data respectively, generating first comparison result data and second comparison result data, and generating physiological response intensity data according to the first comparison result data and the second comparison result data; and further inputting the intensity data of continuous time periods into a trend prediction model, outputting physiological response trend data in a future time window, and realizing objective, real-time, quantitative and predictable monitoring of local physiological state changes under the action of external application of the traditional Chinese medicine through multi-dimensional signal fusion and double-reference dynamic comparison.

Description

Technical Field

[0001] This invention belongs to the field of medical testing technology, and in particular relates to a method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality. Background Technology

[0002] Traditional Chinese medicine (TCM) external application, when applied to the skin, induces a series of complex physiological changes. Currently, there is a lack of objective and quantitative monitoring and analysis methods for these physiological responses. Existing observation methods rely primarily on the operator's subjective experience and the subject's self-description, which has significant technical limitations: First, the assessment results are highly dependent on the operator's personal experience, and judgments may differ significantly between operators, leading to a lack of consistency and comparability. Second, existing methods cannot continuously and in real-time monitor the dynamic evolution of the body's physiological state during TCM external application; they typically only provide a one-time, delayed summary after the application. Third, traditional observation methods often focus only on single physiological indicators (such as body surface temperature or subjective comfort), making it difficult to comprehensively and holistically depict physiological changes from multiple dimensions, such as blood circulation, metabolic activity, and skin electrical properties. Finally, due to the lack of in-depth mining and analysis of multi-dimensional physiological data, it is impossible to objectively predict the trend of physiological changes. These problems seriously hinder a deeper understanding of the mechanism of action of TCM external application and the scientific and standardized research on its application process. Therefore, there is an urgent need for a technical solution that can objectively, in real time, and in multiple dimensions monitor and analyze changes in the local physiological state of the human body under the influence of external application of traditional Chinese medicine. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for real-time monitoring and analysis of physiological responses to external application of traditional Chinese medicine based on multimodality, so as to solve the technical problems in the prior art, such as strong subjectivity in assessing changes in physiological state caused by external application of traditional Chinese medicine, inability to monitor in real time, lack of comprehensive analysis of multi-dimensional data, and inability to objectively predict the trend of physiological response.

[0004] To achieve the above objectives, a method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality is provided in the first aspect of the present invention, comprising the following steps: The subjects obtained multidimensional physiological signals from the area where the traditional Chinese medicine was applied externally. These multidimensional physiological signals included infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. The multidimensional physiological signals are subjected to time synchronization, noise suppression and spatial registration to obtain preprocessed multidimensional physiological signals. The preprocessed multidimensional physiological signal is input into a multimodal fusion model, and the multimodal fusion model outputs a fusion feature vector representing the local physiological state. According to the preset physiological response baseline data, the physiological response baseline data includes first baseline data and second baseline data, wherein the first baseline data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second baseline data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. The fused feature vector is compared with the first benchmark data to generate first comparison result data, which represents the degree of deviation of the current physiological state from the group baseline. The fused feature vector is compared with the second benchmark data to generate second comparison result data, which represents the magnitude of change in the current physiological state relative to the individual's initial state. Based on the first comparison result data and the second comparison result data, physiological response intensity data is generated; The physiological response intensity data generated within a continuous time period is input into the trend prediction model, and the trend prediction model outputs the physiological response trend data within a future preset time window.

[0005] Furthermore, before inputting the preprocessed multidimensional physiological signals into the multimodal fusion model, the process also includes: The drug diffusion area is identified based on the infrared thermal imaging signal, and drug diffusion area data is generated. Microcirculation response time-series patterns are extracted based on the laser speckle blood perfusion signal, and microcirculation response pattern data is generated; abnormal conduction areas corresponding to preset meridian paths or acupoint locations are identified based on the skin conductance signal, and meridian response feature data is generated. The drug diffusion area data, the microcirculation response pattern data, and the meridian response feature data are used as auxiliary inputs, and the preprocessed multi-dimensional physiological signals are input into the multimodal fusion model.

[0006] Furthermore, the multimodal fusion model is configured based on preset drug penetration parameters, which include the component information of the traditional Chinese medicine external application preparation, the concentration of the effective ingredient, and the application duration. The multimodal fusion model generates a fusion feature vector based on the correlation between the drug penetration parameters and the preprocessed multidimensional physiological signals, which is used to represent the coupling characteristics between drug action and physiological response.

[0007] Furthermore, the step of comparing the fused feature vector with the first benchmark data to generate the first comparison result data includes: Calculate the Mahalanobis distance between the fused feature vector and the first benchmark data, and use it as the first comparison result data; The step of comparing the fused feature vector with the second benchmark data to generate the second comparison result data includes: The cosine similarity between the fused feature vector and the second benchmark data is calculated and used as the second comparison result data.

[0008] Furthermore, the method for generating physiological response intensity data based on the first comparison result data and the second comparison result data includes: The physiological response intensity data were calculated using a weighted fusion formula. : ; in, The Mahalanobis distance represented by the first comparison result data. The preset maximum allowable Mahalanobis distance threshold, The cosine similarity represented by the second comparison result data. The weighting coefficient has a range of values. Used to balance the relative importance of population baseline deviation and individual state changes.

[0009] Furthermore, the generation process of the physiological response intensity data is further combined with TCM syndrome differentiation rules data, which include: Discrimination rules for hot and cold attributes based on the spatial distribution characteristics of thermal imaging signals; The rules for judging the degree of blood and qi stagnation based on the temporal fluctuation characteristics of blood perfusion signals; The physiological response intensity data is weighted and adjusted based on the matching results between the fused feature vector and the TCM syndrome differentiation rules data.

[0010] Furthermore, the step of comparing the fused feature vector with the first benchmark data to generate the first comparison result data also includes: The fused feature vector is calculated as follows: ; Where Z represents the first comparison result data. This formula calculates the standardized Mahalanobis distance of the current fused feature vector relative to the baseline multivariate distribution of the healthy group. The larger the Z value, the higher the degree of deviation of the current local physiological state from the group's healthy baseline.

[0011] Furthermore, after generating physiological response intensity data, an evaluation step is also included: The physiological response intensity data is compared with a preset intensity-therapeutic effect correlation threshold; If the physiological response intensity data continuously exceeds the high response threshold, a positive therapeutic effect indication signal is generated; If the physiological response intensity data remains below the low response threshold, a parameter adjustment suggestion signal is generated, which includes adjusting the concentration of the traditional Chinese medicine external application formula or the application time.

[0012] Secondly, the present invention also provides a multimodal-based method for detecting the fruit efficacy of external Chinese medicine, used in any of the multimodal-based methods for detecting the fruit efficacy of external Chinese medicine, comprising: The signal acquisition module is used to acquire multi-dimensional physiological signals of the subject's herbal external application area, including infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. The preprocessing module is used to perform time synchronization, noise suppression and spatial registration on the multidimensional physiological signals to obtain preprocessed multidimensional physiological signals. The fusion processing module is used to input the preprocessed multi-dimensional physiological signals into the multimodal fusion model, and the multimodal fusion model outputs a fusion feature vector representing the local physiological state. The benchmark and comparison module is used to base the physiological response benchmark data on preset physiological response benchmark data, which includes first benchmark data and second benchmark data. The first benchmark data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second benchmark data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. The benchmark and comparison module is further configured to compare the fused feature vector with the first benchmark data to generate first comparison result data, and compare the fused feature vector with the second benchmark data to generate second comparison result data; The response intensity generation module is used to generate physiological response intensity data based on the first comparison result data and the second comparison result data; The trend prediction module is used to input physiological response intensity data generated within a continuous time period into the trend prediction model, and the trend prediction model outputs physiological response trend data within a future preset time window. The output module is used to output the physiological response intensity data and the physiological response trend data, so as to objectively record the dynamic changes in the local physiological state of the human body under the action of external application of traditional Chinese medicine.

[0013] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the method for detecting the efficacy of traditional Chinese medicine external treatments based on multimodality as described in any one of the claims.

[0014] The beneficial technical effects of the present invention are at least as follows: This invention provides a multimodal method and system for detecting the efficacy of external Chinese medicine treatments, which has the following outstanding beneficial effects in the field of evaluating the efficacy of external Chinese medicine treatments: Firstly, it achieves objective quantification and standardization of the assessment process, effectively overcoming the limitations of traditional methods that rely on physicians' subjective experience. This invention constructs a comprehensive and complementary objective data foundation by simultaneously collecting multi-dimensional physiological signals from the external application area, including infrared thermal imaging, laser speckle blood perfusion, skin conductivity, and body surface temperature. Through systematic preprocessing steps such as time synchronization, noise suppression, and spatial registration, and utilizing a multimodal fusion model to generate fusion feature vectors representing local physiological states, it then compares these vectors with a pre-set group baseline and individual initial state to ultimately generate quantified physiological response intensity data and its future trend prediction. This series of technical methods transforms traditional subjective qualitative assessment into objective and repeatable quantitative assessment, significantly improving the consistency and reliability of assessment results and providing a standardized evaluation paradigm for external application of traditional Chinese medicine.

[0015] Secondly, it enables real-time monitoring and dynamic evaluation of treatment response, significantly enhancing the accuracy and timeliness of clinical intervention. By optimizing data processing workflows and model computation efficiency, this invention can complete a full analysis chain within seconds, from multi-dimensional signal acquisition, fusion feature extraction, dual-benchmark comparison to intensity and trend data generation, achieving near real-time dynamic monitoring of the local physiological response to external application of traditional Chinese medicine. Based on physiological response intensity data over continuous time periods, the trend prediction model can output the physiological response trend within a preset future time window, allowing clinicians to anticipate the development of therapeutic effects. This monitoring mode based on real-time quantitative data and predictive information supports physicians in shifting from traditional "post-treatment retrospective evaluation" to "proactive intervention during treatment," providing real-time data support for implementing personalized and precise treatment plan adjustments.

[0016] Thirdly, a multi-level data analysis framework from individual to group levels is constructed, providing a systematic data foundation and methodological support for efficacy optimization and mechanism research. This invention achieves a comprehensive assessment that simultaneously considers both general group patterns and individual specificities by pre-setting physiological response benchmark data containing both group baseline characteristics and individual initial state characteristics, and then performing comparative analysis on each. The physiological response intensity data, trend data, and corresponding comparison results generated by the system can be structured, stored, and linked to form a complete individual efficacy profile, supporting longitudinal tracking and visual analysis of individual treatment progress. Furthermore, through the aggregated analysis of group profiles, response patterns of different external application regimens and different population characteristics can be uncovered, thus providing a solid evidence-based medical basis for the optimization and screening of traditional Chinese medicine external application regimens, the exploration of treatment mechanisms, and the formulation of clinical treatment guidelines. Attached Figure Description

[0017] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the working steps of a multimodal method for detecting the efficacy of external treatments of traditional Chinese medicine, as disclosed in one embodiment of the present invention. Detailed Implementation

[0019] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0020] Example 1 refer to Figure 1 This invention provides an embodiment of a method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality, comprising the following steps: S1. Obtain multi-dimensional physiological signals from the area where the herbal medicine was applied externally to the subject. The multi-dimensional physiological signals include infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. S2. Perform time synchronization, noise suppression and spatial registration processing on the multi-dimensional physiological signals to obtain preprocessed multi-dimensional physiological signals; S3. Input the preprocessed multi-dimensional physiological signal into the multimodal fusion model, and output a fusion feature vector representing the local physiological state from the multimodal fusion model; S4. According to the preset physiological response baseline data, the physiological response baseline data includes first baseline data and second baseline data, wherein the first baseline data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second baseline data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. S5. The fused feature vector is compared with the first benchmark data to generate first comparison result data, which represents the degree of deviation of the current physiological state from the group baseline. S6. Compare the fused feature vector with the second benchmark data to generate second comparison result data, wherein the second comparison result data characterizes the magnitude of change in the current physiological state relative to the individual's initial state; S7. Generate physiological response intensity data based on the first comparison result data and the second comparison result data; S8. Input the physiological response intensity data generated within a continuous time period into the trend prediction model, and output the physiological response trend data within a future preset time window from the trend prediction model.

[0021] In this embodiment, the operator first guides the subject to expose the area of ​​the body surface to be treated. An integrated multimodal sensor array is then aligned with this area for synchronous acquisition, obtaining raw data including infrared thermal images, laser speckle perfusion maps, skin conductivity distribution maps, and point temperature signals. These signals are transmitted to the central processing unit (CPU) via wired or wireless means. Subsequently, the CPU initiates a time synchronization algorithm to align the timestamps of all signals with the system clock as a reference; it uses wavelet transform and Kalman filtering to suppress image noise and temporal noise, respectively; and based on the marker registration method, it unifies the spatial coordinates of the infrared image and the blood flow image to the same coordinate system, completing preprocessing. Next, the preprocessed signals are concatenated into a multi-channel tensor and input into a pre-trained multimodal fusion model. This model can employ a structure combining convolutional neural networks and self-attention mechanisms, ultimately outputting a high-dimensional fusion feature vector to comprehensively represent the current physiological state of the area. The system accesses two types of baseline data stored in its local database: one is the population baseline feature vector obtained from statistical analysis of data from a large number of healthy individuals before intervention; the other is the feature vector of the subject at the start of this external application (serving as the individual's initial baseline). The system calculates the deviation (e.g., Mahalanobis distance) and similarity change (e.g., cosine similarity) between the current fused feature vector and these two baselines, generating first and second comparison results. Then, according to a preset fusion algorithm (e.g., weighted summation), the two results are combined into a single physiological response intensity value. Finally, the system inputs this value, along with the response intensity sequence generated over a past period, into a time series prediction model (e.g., LSTM or Transformer), outputting predicted data on the physiological response trend for the next few minutes or hours, completing one real-time detection cycle.

[0022] Furthermore, before inputting the preprocessed multidimensional physiological signals into the multimodal fusion model, the process also includes: The drug diffusion area is identified based on the infrared thermal imaging signal, and drug diffusion area data is generated. Microcirculation response time-series patterns are extracted based on the laser speckle blood perfusion signal, and microcirculation response pattern data is generated; abnormal conduction areas corresponding to preset meridian paths or acupoint locations are identified based on the skin conductance signal, and meridian response feature data is generated. The drug diffusion area data, the microcirculation response pattern data, and the meridian response feature data are used as auxiliary inputs, and the preprocessed multi-dimensional physiological signals are input into the multimodal fusion model.

[0023] In this embodiment, before inputting the preprocessed multidimensional physiological signals into the multimodal fusion model, the system first extracts specific features from each modality. Specifically, based on infrared thermal imaging signals, a U-Net segmentation network is used to identify regions with significantly elevated and diffused temperatures, generating data on the diffusion region of medicinal effects (represented in binary mask form); based on laser speckle blood perfusion signals, the energy proportion of the 0.01–0.1 Hz frequency band is extracted using Fourier transform as microcirculation response mode data; based on skin conductivity signals, the measured conductivity values ​​are spatially matched with pre-stored standard paths of the twelve meridians. If the conductivity value of a certain acupoint exceeds two standard deviations from the population mean, it is marked as an anomaly, generating meridian response feature data. These three types of feature data are concatenated with the original preprocessed signals and input into the multimodal fusion model to enhance the model's ability to identify the specific physiological responses of external application of traditional Chinese medicine.

[0024] Furthermore, the multimodal fusion model is configured based on preset drug penetration parameters, which include the component information of the traditional Chinese medicine external application preparation, the concentration of the effective ingredient, and the application duration. The multimodal fusion model generates a fusion feature vector based on the correlation between the drug penetration parameters and the preprocessed multidimensional physiological signals, which is used to represent the coupling characteristics between drug action and physiological response.

[0025] In this embodiment, the modal fusion model is dynamically configured based on the drug penetration parameters of the traditional Chinese medicine external application preparation before deployment. These drug penetration parameters include the type, concentration (mg / mL), and expected application duration (min) of the main active ingredients (such as menthol and borneol) in the preparation. The system incorporates a pharmacokinetic submodule to estimate the drug penetration depth at different time points based on Fick's diffusion law, and adjusts the attention weights of each modal channel in the fusion model accordingly. For example, when the concentration of volatile components is high, the model automatically increases the weights of thermal imaging and skin conductance channels to more sensitively capture the physiological changes they cause. Thus, the fused feature vector not only reflects the physiological state but also implicitly encodes the coupling characteristics between drug efficacy and physiological response.

[0026] In some embodiments, an interface for configuring the external application scheme is provided. Before the detection begins, the operator needs to enter or select information about the traditional Chinese medicine preparation to be applied from the database, including but not limited to the main ingredients (such as peppermint, safflower, borneol, etc.), the known or estimated concentration of the effective ingredients, and the planned application duration. This information constitutes a set of "drug penetration parameters." This set of parameters is encoded into a vector and input into the "parameter adaptation layer" of the multimodal fusion model. This adaptation layer is typically composed of a fully connected network, and its output dynamically adjusts the convolutional kernel weights of the subsequent feature extraction network or the query vector of the attention mechanism. For example, when the input parameter is "high-concentration blood-activating and stasis-removing drugs," the model will correspondingly increase the attention weight of high-frequency pulsating components in the blood perfusion signal; when the parameter is "long-term application of warming and dispersing cold drugs," the model will focus on analyzing the slow diffusion pattern of the temperature field in the thermal imaging signal. In this way, the fusion feature vector output by the model can more accurately reflect the coupling relationship between "specific drug effects" and "local physiological responses," making subsequent evaluations more consistent with specific treatment scenarios.

[0027] Furthermore, the step of comparing the fused feature vector with the first benchmark data to generate the first comparison result data includes: Calculate the Mahalanobis distance between the fused feature vector and the first benchmark data, and use it as the first comparison result data; The step of comparing the fused feature vector with the second benchmark data to generate the second comparison result data includes: The cosine similarity between the fused feature vector and the second benchmark data is calculated and used as the second comparison result data.

[0028] In this embodiment, the first comparison result data is processed by calculating a fused feature vector. Compared with the first baseline data (denoted as the population mean vector) Covariance Matrix The Mahalanobis distance between them is obtained by the formula: ; This distance measures the degree to which the current state deviates from the distribution of a healthy population in a multidimensional physiological space. The second comparison result data is obtained through calculation... Compared with the second benchmark data The cosine similarity of (individual initial fusion vectors) is obtained as follows: ; This value reflects the relative change in an individual's own state, and its range is [-1, 1]. The closer it is to 1, the smaller the change.

[0029] Furthermore, the step of generating physiological response intensity data based on the first comparison result data and the second comparison result data includes: calculating the physiological response intensity data using a weighted fusion formula. : ; in, The Mahalanobis distance represented by the first comparison result data. The preset maximum allowable Mahalanobis distance threshold, The cosine similarity represented by the second comparison result data. The weighting coefficient has a range of values. Used to balance the relative importance of population baseline deviation and individual state changes.

[0030] Furthermore, the generation process of the physiological response intensity data is further combined with TCM syndrome differentiation rules data, which include: Discrimination rules for hot and cold attributes based on the spatial distribution characteristics of thermal imaging signals; The rules for judging the degree of blood and qi stagnation based on the temporal fluctuation characteristics of blood perfusion signals; The physiological response intensity data is weighted and adjusted based on the matching results between the fused feature vector and the TCM syndrome differentiation rules data.

[0031] Furthermore, the step of comparing the fused feature vector with the first benchmark data to generate the first comparison result data also includes: The fused feature vector is calculated as follows: ; Where Z represents the first comparison result data. This formula calculates the standardized Mahalanobis distance of the current fused feature vector relative to the baseline multivariate distribution of the healthy group. The larger the Z value, the higher the degree of deviation of the current local physiological state from the group's healthy baseline.

[0032] Furthermore, after generating physiological response intensity data, an evaluation step is also included: The physiological response intensity data is compared with a preset intensity-therapeutic effect correlation threshold; If the physiological response intensity data continuously exceeds the high response threshold, a positive therapeutic effect indication signal is generated; If the physiological response intensity data remains below the low response threshold, a parameter adjustment suggestion signal is generated, which includes adjusting the concentration of the traditional Chinese medicine external application formula or the application time.

[0033] Secondly, the present invention also provides a multimodal-based method for detecting the fruit efficacy of external Chinese medicine, used in any of the multimodal-based methods for detecting the fruit efficacy of external Chinese medicine, comprising: The signal acquisition module is used to acquire multi-dimensional physiological signals of the subject's herbal external application area, including infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. The preprocessing module is used to perform time synchronization, noise suppression and spatial registration on the multidimensional physiological signals to obtain preprocessed multidimensional physiological signals. The fusion processing module is used to input the preprocessed multi-dimensional physiological signals into the multimodal fusion model, and the multimodal fusion model outputs a fusion feature vector representing the local physiological state. The benchmark and comparison module is used to base the physiological response benchmark data on preset physiological response benchmark data, which includes first benchmark data and second benchmark data. The first benchmark data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second benchmark data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. The benchmark and comparison module is further configured to compare the fused feature vector with the first benchmark data to generate first comparison result data, and compare the fused feature vector with the second benchmark data to generate second comparison result data; The response intensity generation module is used to generate physiological response intensity data based on the first comparison result data and the second comparison result data; The trend prediction module is used to input physiological response intensity data generated within a continuous time period into the trend prediction model, and the trend prediction model outputs physiological response trend data within a future preset time window. The output module is used to output the physiological response intensity data and the physiological response trend data, so as to objectively record the dynamic changes in the local physiological state of the human body under the action of external application of traditional Chinese medicine.

[0034] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the method for detecting the efficacy of traditional Chinese medicine external treatments based on multimodality as described in any one of the claims.

[0035] This embodiment provides a method and system for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality, which has the following beneficial effects: This embodiment provides an objective, quantitative, and standardized method for evaluating the efficacy of external application of traditional Chinese medicine (TCM), effectively overcoming the limitations of traditional methods that rely on the subjective judgment of physicians. By simultaneously collecting infrared thermal imaging signals, laser speckle blood flow perfusion signals, skin conductivity signals, and body surface temperature signals from the TCM application area of ​​the subject, a comprehensive and complementary multi-dimensional physiological data foundation is constructed. After preprocessing such as time synchronization, noise suppression, and spatial registration, the data is input into a multimodal fusion model to generate a fusion feature vector representing the local physiological state. This vector is then compared with preset first baseline data (baseline of a healthy group) and second baseline data (individual initial state), generating first and second comparison result data, and based on these, physiological response intensity data and future trend prediction data are calculated. This technical process transforms the traditional qualitative and subjective evaluation method into a repeatable and quantifiable objective analysis process, significantly improving the consistency, reliability, and comparability of the evaluation results, and establishing a standardized evaluation paradigm for external application of TCM. It also significantly enhances the timeliness and accuracy of intervention decisions. By optimizing signal acquisition frequency, data preprocessing algorithms, and model inference efficiency, the system can complete a full analysis chain from multimodal signal acquisition to physiological response intensity and trend data output within seconds, enabling continuous and dynamic tracking of local physiological state changes under the influence of external application of traditional Chinese medicine. Based on the physiological response intensity data generated within a continuous time period, the trend prediction model can extrapolate the response evolution trend within a preset time window, allowing operators to proactively grasp the response development trend. This mechanism supports the shift from "retrospective summary after treatment" to "real-time feedback and adjustment during treatment," providing timely and reliable data support for personalized treatment parameter optimization. It lays a systematic data foundation for efficacy optimization and mechanism research. By simultaneously introducing primary baseline data representing the population's health baseline and secondary baseline data reflecting the individual's initial state, the system achieves a comprehensive assessment of physiological responses that considers both general reference standards and individual differences. The generated physiological response intensity data, trend data, and comparison results data can be stored in a structured manner to form a complete individual response profile, supporting longitudinal tracking and visualization analysis. After further aggregating multiple individual profiles, response pattern mining at the population level can be carried out to analyze the correlation between different Chinese medicine formulas, application duration, population constitution, and other factors and physiological responses. This provides solid methodological and data support for the scientific optimization of Chinese medicine external application programs, the in-depth explanation of the mechanism of action, and the development of evidence-based clinical application guidelines.

[0036] In the description of this specification, the references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0037] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodal methods, characterized in that, Includes the following steps: The subjects obtained multidimensional physiological signals from the area where the traditional Chinese medicine was applied externally. These multidimensional physiological signals included infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. The multidimensional physiological signals are subjected to time synchronization, noise suppression and spatial registration to obtain preprocessed multidimensional physiological signals. The preprocessed multidimensional physiological signal is input into a multimodal fusion model, and the multimodal fusion model outputs a fusion feature vector representing the local physiological state. According to the preset physiological response baseline data, the physiological response baseline data includes first baseline data and second baseline data, wherein the first baseline data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second baseline data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. The fused feature vector is compared with the first benchmark data to generate first comparison result data, which represents the degree of deviation of the current physiological state from the group baseline. The fused feature vector is compared with the second benchmark data to generate second comparison result data, which represents the magnitude of change in the current physiological state relative to the individual's initial state. Based on the first comparison result data and the second comparison result data, physiological response intensity data is generated; The physiological response intensity data generated within a continuous time period is input into the trend prediction model, and the trend prediction model outputs the physiological response trend data within a future preset time window.

2. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality according to claim 1, characterized in that, Before inputting the preprocessed multidimensional physiological signals into the multimodal fusion model, the following steps are also included: The drug diffusion area is identified based on the infrared thermal imaging signal, and drug diffusion area data is generated. Microcirculation response time-series patterns are extracted based on the laser speckle blood perfusion signal, and microcirculation response pattern data is generated; abnormal conduction areas corresponding to preset meridian paths or acupoint locations are identified based on the skin conductance signal, and meridian response feature data is generated. The drug diffusion area data, the microcirculation response pattern data, and the meridian response feature data are used as auxiliary inputs, and the preprocessed multi-dimensional physiological signals are input into the multimodal fusion model.

3. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodal analysis according to claim 1, characterized in that, The multimodal fusion model is configured based on preset drug penetration parameters, which include the component information, effective ingredient concentration, and application duration of the traditional Chinese medicine external application preparation. The multimodal fusion model generates a fusion feature vector based on the correlation between the drug penetration parameters and the preprocessed multidimensional physiological signals, which is used to represent the coupling characteristics between drug action and physiological response.

4. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality according to claim 1, characterized in that, The fused feature vector is compared with the first benchmark data to generate first comparison result data, including: Calculate the Mahalanobis distance between the fused feature vector and the first benchmark data, and use it as the first comparison result data; The step of comparing the fused feature vector with the second benchmark data to generate the second comparison result data includes: The cosine similarity between the fused feature vector and the second benchmark data is calculated and used as the second comparison result data.

5. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality according to claim 1, characterized in that, The step of generating physiological response intensity data based on the first comparison result data and the second comparison result data includes: The physiological response intensity data were calculated using a weighted fusion formula. : ； in, The Mahalanobis distance represented by the first comparison result data. The preset maximum allowable Mahalanobis distance threshold, The cosine similarity represented by the second comparison result data. The weighting coefficient has a range of values. Used to balance the relative importance of population baseline deviation and individual state changes.

6. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodal analysis according to claim 1, characterized in that, The process of generating the physiological response intensity data includes combining data from traditional Chinese medicine (TCM) diagnostic rules, which include: Discrimination rules for hot and cold attributes based on the spatial distribution characteristics of thermal imaging signals; The rules for judging the degree of blood and qi stagnation based on the temporal fluctuation characteristics of blood perfusion signals; The physiological response intensity data is weighted and adjusted based on the matching results between the fused feature vector and the TCM syndrome differentiation rules data.

7. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality according to claim 4, characterized in that, Also includes: The step of comparing the fused feature vector with the first benchmark data to generate the first comparison result data further includes: The fused feature vector is calculated as follows: ； Where Z represents the first comparison result data; the formula calculates the standardized Mahalanobis distance of the current fused feature vector relative to the baseline multivariate distribution of the healthy group. The larger the Z value, the higher the degree of deviation of the current local physiological state from the group's healthy baseline.

8. The method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality according to claim 7, characterized in that, Also includes: After generating physiological response intensity data, an evaluation step is also included: The physiological response intensity data is compared with a preset intensity-therapeutic effect correlation threshold; If the physiological response intensity data continuously exceeds the high response threshold, a positive therapeutic effect indication signal is generated; If the physiological response intensity data remains below the low response threshold, a parameter adjustment suggestion signal is generated, which includes adjusting the concentration of the traditional Chinese medicine external application formula or the application time.

9. A multimodal-based method for detecting the fruit efficacy of external Chinese medicine, used in the multimodal-based method for detecting the fruit efficacy of external Chinese medicine as described in any one of claims 1-8, characterized in that, include: The signal acquisition module is used to acquire multi-dimensional physiological signals of the subject's herbal external application area, including infrared thermal imaging signals, laser speckle blood perfusion signals, skin conductance signals, and body surface temperature signals. The preprocessing module is used to perform time synchronization, noise suppression and spatial registration on the multidimensional physiological signals to obtain preprocessed multidimensional physiological signals. The fusion processing module is used to input the preprocessed multi-dimensional physiological signals into the multimodal fusion model, and the multimodal fusion model outputs a fusion feature vector representing the local physiological state. The benchmark and comparison module is used to base the physiological response benchmark data on preset physiological response benchmark data, which includes first benchmark data and second benchmark data. The first benchmark data is the multimodal physiological baseline feature vector of the corresponding body surface area of ​​a healthy subject in the state without the application of traditional Chinese medicine external application, and the second benchmark data is the fusion feature vector of the same subject at the initial moment of the application of traditional Chinese medicine external application. The benchmark and comparison module is further configured to compare the fused feature vector with the first benchmark data to generate first comparison result data, and compare the fused feature vector with the second benchmark data to generate second comparison result data; The response intensity generation module is used to generate physiological response intensity data based on the first comparison result data and the second comparison result data; The trend prediction module is used to input physiological response intensity data generated within a continuous time period into the trend prediction model, and the trend prediction model outputs physiological response trend data within a future preset time window. The output module is used to output the physiological response intensity data and the physiological response trend data, so as to objectively record the dynamic changes in the local physiological state of the human body under the action of external application of traditional Chinese medicine.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for detecting the efficacy of external treatments of traditional Chinese medicine based on multimodality as described in any one of claims 1-8.

Images