A digital temperature control and curative effect management system for huolong moxibustion

Abstract

The application discloses a digital temperature control and curative effect management system for fire dragon moxibustion, and relates to the technical field of physiotherapy apparatuses, which comprises a temperature data acquisition unit, a temperature control signal generation and transmission unit, a score input and parameter acquisition unit, a curative effect simulation and strategy generation unit, and the like.

Description

Technical Field

[0001] This invention relates to the field of physiotherapy equipment technology, and more specifically, to a digital temperature control and efficacy management system for moxibustion. Background Technology

[0002] Fire Dragon Moxibustion, a classic external therapy in Traditional Chinese Medicine, utilizes the warm energy released by burning mugwort to penetrate key acupoints along the body's meridians, exerting its core effects of warming yang, dispelling cold, promoting blood circulation, and tonifying yang qi. It has shown significant efficacy in the clinical treatment of conditions such as deficiency-cold constitution, chronic musculoskeletal pain, and gynecological dysmenorrhea due to uterine cold. It is particularly suitable for patients with yang deficiency constitution and is highly recognized and favored by clinicians and patients alike.

[0003] However, traditional moxibustion relies entirely on the clinical experience of medical personnel, controlling temperature by manually adjusting the burning height, spacing, or thickness of the covering of the moxa sticks, lacking scientific quantitative standards and real-time monitoring methods. This makes it prone to temperature imbalances during treatment, leading to local skin overheating and burns, or affecting the stability of treatment effects due to insufficient heat penetration depth or insufficient temperature duration. Furthermore, the Du meridian and Bladder meridian on the back of the human body have a dense concentration of key acupoints with diverse functions, and the temperature tolerance thresholds and required heat stimulation intensity differ significantly between the main treatment acupoints and surrounding auxiliary acupoints. For example, the main acupoint requires continuous and stable heat penetration to activate the flow of Qi and blood in the meridians, while auxiliary acupoints require a gentle and appropriate temperature to avoid excessive heat damaging the body's vital energy. However, current technology cannot achieve simultaneous temperature acquisition and differentiated control at multiple targets, often resulting in uneven temperature distribution within the treatment area, with both local overheating and insufficient heat occurring simultaneously.

[0004] Furthermore, existing moxibustion devices mostly use uniform and fixed temperature settings, failing to fully consider individual patient differences, disease types, and the treatment needs of specific key acupoints. This results in a lack of personalized adaptability in treatment plans, making it difficult to accurately meet the pathological conditions and conditioning needs of different patients, thus limiting further improvements in efficacy.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] In response to the problems in related technologies, this invention proposes a digital temperature control and therapeutic effect management system for moxibustion, in order to overcome the aforementioned technical problems existing in the existing related technologies.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows:

[0008] A digital temperature control and therapeutic effect management system for moxibustion, the system comprising:

[0009] The temperature data acquisition unit is used to convert the analog temperature signal collected by the temperature measuring device into a digital signal using an analog-to-digital conversion channel, and to perform noise reduction and smoothing processing on the digital signal using a filtering algorithm to acquire temperature data.

[0010] The temperature control signal generation and transmission unit is used to draw a target temperature curve based on the preset fire dragon moxibustion physiotherapy data management library, compare the target temperature curve with the acquired temperature data, combine the patient's physiological feedback data, construct the controller architecture, and use the controller architecture to generate the corresponding temperature control signal and transmit it to the fire dragon moxibustion box.

[0011] The scoring and parameter acquisition unit is used to input the multidimensional efficacy scores of patients after each treatment through the software interface, combine temperature data and physiological feedback data to build an efficacy prediction model, and analyze the efficacy prediction results of the efficacy prediction model to obtain the optimal heating parameters.

[0012] The therapeutic effect simulation and strategy generation unit is used to analyze the patient's physiological feedback data, recommend candidate key acupoint combinations for the patient, and use a bio-thermal conduction simulation algorithm to simulate the therapeutic effect of the optimal heating parameter set on the candidate key acupoint combinations. Based on the simulation results, the optimal key acupoint combination is selected and a personalized optimal temperature control strategy is generated.

[0013] Furthermore, the temperature control signal generation and transmission unit includes:

[0014] The multidimensional error acquisition module is used to draw the target temperature curve of moxibustion treatment based on the preset moxibustion therapy data management library and the patient's personalized treatment needs, and compare the acquired temperature data with the target temperature curve to obtain the multidimensional error.

[0015] An adaptive control architecture construction module is used to collect patients' physiological feedback data in real time. The multidimensional error and physiological feedback data are used as the state space, and the height adjustment of the soft mesh inside the moxibustion box and the control of the moxa wool feeding amount are used as the action space. The corresponding reward function is designed and combined with the adaptive control algorithm to construct the adaptive control architecture.

[0016] The temperature control signal generation module is used to construct a time-series prediction model of temperature change trend using a long short-term memory network. Combined with the collected historical temperature data, the time-series prediction model is used to predict the temperature change trend in the future time period, and the prediction results are input into the adaptive control architecture to generate the corresponding temperature control signal.

[0017] The signal verification and transmission module is used to input the generated temperature control signal into the preset heat conduction physical model for simulation verification, perform safety verification on the simulation verification results in combination with the preset safety threshold, and transmit the temperature control signal that has passed the safety verification to the actuator of the Fire Dragon Moxibustion Box.

[0018] Furthermore, the scoring entry and parameter acquisition unit includes:

[0019] The efficacy prediction model building module is used to build efficacy prediction models based on the entered multidimensional efficacy scores and the acquired temperature data and physiological feedback data. The efficacy prediction models include short-term efficacy prediction models and long-term efficacy prediction models.

[0020] The efficacy prediction model selection module is used to dynamically select between short-term and long-term prediction models to predict efficacy based on the patient's individualized treatment goals and current treatment stage.

[0021] The optimal heating parameter acquisition module is used to analyze the predicted results of therapeutic effects, establish a multi-objective optimization model, and use a multi-objective optimization algorithm to solve the multi-objective optimization model to obtain the optimal heating parameters.

[0022] Furthermore, the efficacy prediction model building module includes:

[0023] The feature extraction submodule is used to analyze the acquired temperature data and collected physiological feedback data using time-frequency analysis algorithms, and extract the corresponding time-series features and frequency domain features based on the analysis results.

[0024] The feature fusion submodule is used to extract semantic features from patients' multidimensional efficacy scores using natural language processing technology, and to perform cross-modal feature fusion of semantic features, temporal features and frequency domain features.

[0025] The short-term efficacy prediction model construction submodule is used to build a short-term efficacy prediction model based on dynamic graph structure by using graph neural networks and temporal attention mechanism based on the cross-modal feature fusion results.

[0026] The long-term efficacy prediction model construction submodule is used to build an initial long-term efficacy prediction model using deep learning algorithms. By combining transfer learning technology and patients' physiological feedback data, the initial long-term efficacy prediction model is optimized to obtain the long-term efficacy prediction model.

[0027] Furthermore, an initial long-term efficacy prediction model was constructed using deep learning algorithms. This model was then optimized by combining transfer learning techniques with patients' physiological feedback data, resulting in a long-term efficacy prediction model that includes:

[0028] Deep learning algorithms are used to process the cross-modal feature fusion results, and an initial long-term efficacy prediction model is constructed based on the processing results.

[0029] Collect historical efficacy data of patients with similar conditions, use transfer learning technology to learn relevant knowledge and experience from the historical efficacy data, and transfer the learning results to the initial long-term efficacy prediction model;

[0030] Based on the patient's physiological feedback data, the initial long-term efficacy prediction model after migration was optimized using the difference-in-differences algorithm to obtain the patient's long-term efficacy prediction model.

[0031] Furthermore, historical efficacy data of similar patients are collected, and transfer learning techniques are used to learn relevant knowledge and experience from the historical efficacy data. The learning results are then transferred to the initial long-term efficacy prediction model, including:

[0032] Based on the TCM efficacy evaluation criteria, the historical efficacy data of similar patients were analyzed to identify the evolutionary patterns in the treatment process. Based on these patterns, the treatment process of Fire Dragon Moxibustion was divided into several treatment stages.

[0033] The state combination and action chain mechanism are used to process each treatment stage, and a state space and corresponding action space are established based on the processing results.

[0034] By treating patients of the same type as intelligent agents, and combining the established state space with the corresponding action space, a reinforcement learning environment simulating the fire dragon moxibustion treatment process is constructed, and the state transition mechanism and reward signal of the learning environment are strengthened.

[0035] Based on the defined reinforcement learning environment, transfer learning techniques are used to learn relevant treatment knowledge and experience from historical efficacy data. Based on the learning results, the optimal treatment strategy is generated and transferred as prior knowledge to the initial long-term efficacy prediction model.

[0036] Furthermore, the optimal heating parameter acquisition module includes:

[0037] The heating parameter range acquisition submodule is used to analyze the predicted results of therapeutic efficacy using a preset causal forest model, identify the causal effects of different heating parameters on the treatment effect, assess the degree of influence of each heating parameter on the treatment risk, and obtain the heating parameter range.

[0038] The heating parameter set generation submodule is used to construct a multi-objective optimization model with the obtained heating parameter range as constraints, aiming to maximize the therapeutic effect and minimize the therapeutic risk. The multi-objective optimization algorithm is then used to solve the multi-objective optimization model to generate the optimal solution set as the optimal heating parameter set.

[0039] Furthermore, the efficacy simulation and strategy generation unit includes:

[0040] The heterogeneous graph construction module is used to extract physiological features from the collected patient physiological feedback data, and based on the attribute features of each key acupoint, analyze the correlation strength between the physiological features and the attribute features of the key acupoints, and construct a heterogeneous graph model with key acupoints as nodes and correlation strength as edges.

[0041] The correlation score module is used to train the heterogeneous graph model using a graph neural network. It learns the importance weights of each key acupoint to the treatment response through a multi-layer message passing mechanism, and obtains the correlation score between key acupoints and therapeutic efficacy.

[0042] The candidate key acupoint combination generation module is used to mine candidate key acupoint combinations with therapeutic synergy from the heterogeneous graph model based on the obtained correlation score using a greedy iterative algorithm and a collaborative filtering recommendation algorithm.

[0043] The simulation scoring module is used to construct a three-dimensional heat conduction model based on the theory of biological heat conduction, and use the optimal heating parameter set as boundary conditions. The three-dimensional heat conduction model is used to simulate the temperature distribution, heat penetration depth and tissue response of different candidate key acupoint combinations during the treatment process, and outputs a simulation score based on the simulation results.

[0044] The strategy generation module is used to select the optimal key acupoint combination from the candidate key acupoint combinations based on the output simulation score, combined with the physical constraints of the fire dragon moxibustion box and the preset simulation safety threshold, and combine the optimal key acupoint combination with the optimal heating parameter set to generate a personalized temperature control treatment strategy.

[0045] Furthermore, the candidate key acupoint combination generation module includes:

[0046] The collaborative similarity calculation submodule is used to treat each key acupoint in the heterogeneous graph model as an independent cluster and calculate the collaborative similarity between any two clusters based on the association strength.

[0047] The initial collaborative cluster formation submodule is used to use a greedy iterative algorithm to traverse all cluster pairs, select the two clusters with the highest collaborative similarity and merge them to form the initial collaborative cluster.

[0048] The collaborative cluster acquisition submodule is used to recalculate and update the collaborative similarity between the initial collaborative cluster and the remaining clusters until the collaborative similarity between any two clusters is lower than a preset threshold, at which point the iteration stops and several collaborative clusters are obtained.

[0049] The synergy effect value acquisition submodule is used to calculate the multidimensional efficacy score of all key acupoints in each synergy cluster based on the obtained correlation score, and to perform differential processing on the multidimensional efficacy score to obtain the synergy effect value.

[0050] The collaborative recommendation submodule is used to mine and generate candidate key acupoint combinations with therapeutic synergistic effects from the collaborative cluster based on the obtained synergistic effect values ​​using a collaborative filtering recommendation algorithm.

[0051] Furthermore, based on the obtained synergy effect values, a collaborative filtering recommendation algorithm is used to mine and generate candidate key acupoint combinations with therapeutic synergy effects from the collaborative clusters, including:

[0052] Based on the synergistic cluster, the corresponding synergistic effect value, and the historical efficacy data of similar patients, a multidimensional feature matrix is ​​constructed, and a key acupoint is randomly selected from the multidimensional feature matrix as the target key acupoint.

[0053] Calculate the similarity index between the target key acupoint and the remaining key acupoints, use the important nearest neighbor algorithm to count the number of common nearest neighbors between the target key acupoint and the remaining key acupoints, and select key acupoints that meet the preset screening threshold from the remaining key acupoints as important nearest neighbor key acupoints of the target key acupoint.

[0054] The key acupoints in the collaborative cluster are designated as primary key acupoints, and the important nearest neighbor key acupoints are designated as secondary key acupoints. The collaborative filtering algorithm is used to filter the primary key acupoints to generate an initial candidate key acupoint combination.

[0055] The system uses a pre-defined category boosting tree model to predict and analyze the initial candidate key acupoint combinations. Combined with the patient's historical treatment data, it evaluates the efficacy of each initial candidate key acupoint combination and generates a generated candidate key acupoint combination with therapeutic synergy.

[0056] The beneficial effects of this invention are as follows:

[0057] 1. This invention converts the analog signals collected by the temperature measuring device into digital signals and uses a filtering algorithm for noise reduction and smoothing to ensure the accuracy of temperature data. This enables precise temperature control during treatment, avoiding burns caused by local overheating and insufficient temperature penetration, while also improving the safety and stability of treatment, making the treatment process more reliable.

[0058] 2. This invention generates a temperature control signal by combining patient physiological feedback data with the target temperature curve and transmits it to the Fire Dragon Moxibustion Box. The system can automatically adjust the temperature, accurately match personalized treatment plans, ensure that each patient receives the most suitable temperature stimulation during treatment, improve treatment effect, and avoid subjective deviation in temperature control.

[0059] 3. This invention uses a bio-thermal conduction simulation algorithm to simulate heating parameters, ensuring the scientific nature of the treatment plan. Combined with the patient's physiological feedback data and multi-dimensional efficacy scores, the system can optimize the treatment plan, select the most suitable key acupoint combination, and provide personalized temperature control strategies. This not only improves the patient's treatment experience but also reduces potential safety hazards during the treatment process. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 This is a schematic diagram of a digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention;

[0062] Figure 2 This is a flowchart of a digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention;

[0063] Figure 3 This is a principle block diagram of the scoring input and parameter acquisition unit in a digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention;

[0064] Figure 4 This is a principle block diagram of the therapeutic effect simulation and strategy generation unit in a digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention.

[0065] Figure 5 This is a pain digital score chart in a digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention.

[0066] In the picture:

[0067] 1. Temperature data acquisition unit; 2. Temperature control signal generation and transmission unit; 3. Scoring and parameter acquisition unit; 4. Therapeutic effect simulation and strategy generation unit. Detailed Implementation

[0068] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention.

[0069] According to an embodiment of the present invention, a digital temperature control and therapeutic effect management system for moxibustion is provided.

[0070] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figures 1-4 As shown, the digital temperature control and therapeutic effect management system for moxibustion according to an embodiment of the present invention includes:

[0071] Temperature data acquisition unit 1 is used to convert the analog temperature signal collected by the temperature measuring device into a digital signal using an analog-to-digital conversion channel, and to perform noise reduction and smoothing processing on the digital signal using a filtering algorithm to acquire temperature data.

[0072] It should be further explained that the analog-to-digital conversion channel converts analog signals into digital signals, typically using a high-precision analog-to-digital converter (such as a 12-bit or 16-bit ADC). During the conversion process, the ADC decomposes the analog signal into discrete digital data points according to the set sampling frequency; for example, if a 12-bit ADC is used, its quantization level is 4096 (2... 12 This can convert temperature signals into finer-grained digital values; in addition, common filtering methods include low-pass filters, weighted averages, Kalman filters, etc.; for example, the weighted average filtering method reduces the impact of sudden noise by weighting the current signal with the signals from the previous few moments.

[0073] The temperature control signal generation and transmission unit 2 is used to draw a target temperature curve based on the preset fire dragon moxibustion physiotherapy data management library, compare the target temperature curve with the acquired temperature data, combine the patient's physiological feedback data, construct a controller architecture, and use the controller architecture to generate corresponding temperature control signals to transmit to the fire dragon moxibustion box.

[0074] In this optional embodiment, the temperature control signal generation and transmission unit 2 includes:

[0075] The multidimensional error acquisition module is used to draw the target temperature curve of moxibustion treatment based on the preset moxibustion therapy data management library and the patient's personalized treatment needs, and compare the acquired temperature data with the target temperature curve to obtain the multidimensional error.

[0076] It should be further explained that the Fire Dragon Moxibustion Therapy Data Management Database is a comprehensive database integrating various information related to Fire Dragon Moxibustion treatment, covering data from multiple dimensions such as different treatment plans, temperature control parameters, patient constitution, and treatment response. The database records treatment cases for various common diseases, including specific parameters such as the temperature, heating time, and acupoint combinations used in each treatment process, as well as corresponding treatment effect assessments, such as symptom relief and patient subjective feelings. Simultaneously, the database also stores personalized treatment needs and efficacy feedback for different types of patients (such as age, gender, and constitution), thus providing doctors with precise treatment plan support. Through this data, doctors can draw the most suitable target temperature curve based on the patient's specific condition, achieving personalized treatment.

[0077] Personalized treatment needs refer to treatment plans tailored to each patient's specific condition, physical characteristics, and treatment goals. These needs consider factors such as the patient's age, gender, health status, medical history, treatment response, and lifestyle habits, and adjust treatment parameters based on this information, such as temperature, treatment time, and acupoint selection. Furthermore, personalized treatment needs include the patient's expectations for treatment outcomes, such as pain relief, functional improvement, and enhanced quality of life. Simultaneously, it is necessary to comprehensively consider factors such as comfort, safety, and risk control during the treatment process to ensure that the treatment is not only effective but also minimizes side effects or discomfort. Through this personalized data, doctors can develop precise treatment plans that are highly aligned with the patient's needs.

[0078] For example, the Roland-Morris Disability Questionnaire (as shown in Table 1) can be used to analyze the patient's physical characteristics, condition, and historical treatment data. Based on the treatment rules in the knowledge base and the efficacy data of similar patients, a dynamic target temperature curve can be plotted. In addition, during the treatment process, attention should be paid to multidimensional errors, such as temperature error, time error, spatial error, and positional error, to ensure the accuracy of the treatment effect and the stability of the treatment process.

[0079] Table 1: Roland-Morris Disability Questionnaire (RMDQ)

[0080] An adaptive control architecture construction module is used to collect patients' physiological feedback data in real time. The multidimensional error and physiological feedback data are used as the state space, and the height adjustment of the soft mesh inside the moxibustion box and the control of the moxa wool feed amount are used as the action space. The corresponding reward function is designed, and the adaptive control architecture is constructed in combination with the adaptive control algorithm.

[0081] It should be further explained that the physiological feedback data encompasses multi-dimensional information reflecting the patient's physical condition and response to treatment, including objective physiological indicators such as body temperature, pulse, blood pressure, respiratory rate, and skin conductance. This data is used to monitor the dynamic physiological changes of the patient in real time during the moxibustion treatment, thereby assessing the treatment effect and the body's tolerance. Furthermore, such as... Figure 5 As shown, this data system also integrates patients' subjective feedback, such as the pain level and comfort score assessed using the Numerical Rating Scale (NRS). By combining objective physiological indicators with subjective experience data, the adaptability and safety of the treatment process can be evaluated more comprehensively, providing a basis for dynamically optimizing temperature control strategies and acupoint selection during treatment, and improving the personalization and precision of treatment.

[0082] Based on this, a dual-loop adaptive control architecture integrating "positive optimization and negative feedback error correction" can be constructed, as follows:

[0083] When the state space monitoring shows that the treatment temperature is within the target range and physiological feedback indicators (such as skin impedance, heart rate variability, etc.) are good, the system finely adjusts the motion space parameters through a positive feedback mechanism. For example, it maintains the current moxa wool feeding rate and slightly adjusts the soft mesh height to stabilize the heat radiation intensity, thereby continuously enhancing the treatment effect and improving comfort. If an abnormal negative feedback signal is detected (such as a sudden change in skin impedance, local temperature exceeding the safety threshold, etc.), the system immediately triggers a reverse adjustment cycle: by increasing the soft mesh height to reduce heat radiation and simultaneously reducing the moxa wool feeding amount to reduce combustion heat generation, the temperature is quickly pulled back to the safe range to prevent tissue damage or adverse reactions. Conversely, if the system identifies that the temperature is too low (manifested as insufficient warm stimulation, no significant improvement in therapeutic indicators, etc.), a heating strategy is initiated: the soft mesh height is reduced to enhance heat conduction, and the moxa wool feeding amount is appropriately increased to enhance heat output, forming a closed-loop negative feedback mechanism of "temperature deviation recognition, reverse motion adjustment, and real-time state correction".

[0084] The temperature control signal generation module is used to construct a time-series prediction model of temperature change trend using a long short-term memory network. Combined with the collected historical temperature data, the time-series prediction model is used to predict the temperature change trend in the future time period, and the prediction results are input into the adaptive control architecture to generate the corresponding temperature control signal.

[0085] The signal verification and transmission module is used to input the generated temperature control signal into the preset heat conduction physical model for simulation verification, perform safety verification on the simulation verification results in combination with the preset safety threshold, and transmit the temperature control signal that has passed the safety verification to the actuator of the Fire Dragon Moxibustion Box.

[0086] The scoring input and parameter acquisition unit 3 is used to input the multidimensional efficacy scores of patients after each treatment through the software interface, combine temperature data and physiological feedback data to construct an efficacy prediction model, and analyze the efficacy prediction results of the efficacy prediction model to obtain the optimal heating parameters.

[0087] In this optional embodiment, the scoring input and parameter acquisition unit 3 includes:

[0088] The efficacy prediction model construction module 301 is used to construct an efficacy prediction model based on the entered multidimensional efficacy scores and the acquired temperature data and physiological feedback data. The efficacy prediction model includes a short-term efficacy prediction model and a long-term efficacy prediction model.

[0089] It should be noted that, as shown in Table 2, the multidimensional efficacy score includes the NRS score, the Yang deficiency syndrome score, and the RMDQ functional impairment score; among which, the Yang deficiency syndrome score is shown in Table 3; in addition, the physiological feedback data are shown in Tables 4-6.

[0090] Table 2: Multidimensional Therapeutic Efficacy Scoring Table

[0091] Table 3: Yang Deficiency Syndrome Scoring Table

[0092] Table 4: Physiological Feedback Data Table

[0093] Table 5: Physiological Feedback Data Table

[0094] Table 6: Physiological Feedback Data Table

[0095] In this optional embodiment, the efficacy prediction model construction module 301 includes:

[0096] The feature extraction submodule is used to analyze the acquired temperature data and collected physiological feedback data using time-frequency analysis algorithms, and extract the corresponding time-series features and frequency domain features based on the analysis results.

[0097] The feature fusion submodule is used to extract semantic features from patients' multidimensional efficacy scores using natural language processing technology, and to perform cross-modal feature fusion of semantic features, temporal features and frequency domain features.

[0098] The short-term efficacy prediction model construction submodule is used to establish a short-term efficacy prediction model based on a dynamic graph structure by utilizing graph neural networks and temporal attention mechanisms based on the cross-modal feature fusion results.

[0099] The long-term efficacy prediction model construction submodule is used to build an initial long-term efficacy prediction model using deep learning algorithms. By combining transfer learning technology and patients' physiological feedback data, the initial long-term efficacy prediction model is optimized to obtain the long-term efficacy prediction model.

[0100] In this optional embodiment, an initial long-term efficacy prediction model is constructed using a deep learning algorithm. This model is then optimized by combining transfer learning techniques with the patient's physiological feedback data, resulting in a long-term efficacy prediction model that includes:

[0101] Deep learning algorithms are used to process the cross-modal feature fusion results, and an initial long-term efficacy prediction model is constructed based on the processing results.

[0102] Historical efficacy data of similar patients were collected, and transfer learning technology was used to learn relevant knowledge and experience from the historical efficacy data. The learning results were then transferred to the initial long-term efficacy prediction model.

[0103] It should be further noted that the historical efficacy data of similar patients covers multiple dimensions, including treatment records, efficacy assessment results, treatment parameters (such as temperature, time, key acupoint combinations, etc.), physiological feedback data during treatment, and individual characteristics of patients (such as age, gender, body type, etc.). In addition, it also includes efficacy scores after treatment and adverse reactions or discomfort that occurred during treatment. By analyzing the historical efficacy data, it is possible to analyze the impact of different treatment methods and parameter combinations on the efficacy of patients, providing a knowledge base for transfer learning.

[0104] In this optional embodiment, historical efficacy data of similar patients are collected, transfer learning techniques are used to learn relevant knowledge and experience from the historical efficacy data, and the learning results are transferred to the initial long-term efficacy prediction model, including:

[0105] Based on the TCM efficacy evaluation standards, the historical efficacy data of similar patients were analyzed to identify the evolutionary patterns in the treatment process. Based on these patterns, the treatment process of Fire Dragon Moxibustion was divided into several treatment stages.

[0106] It should be further explained that the TCM efficacy evaluation criteria include both qualitative and quantitative assessments of treatment effects, mainly including symptom improvement, post-treatment clinical response, changes in physical signs, patient subjective feelings, and improvement in quality of life. Common evaluation criteria include disease remission rate, symptom disappearance rate, physical recovery, degree of pain relief, and functional improvement. These criteria help to comprehensively evaluate the effectiveness of moxibustion therapy and provide a basis for adjusting subsequent treatment plans. Furthermore, the treatment stages typically include an initial adjustment stage, an improvement stage, and a consolidation stage. The initial adjustment stage mainly aims to relieve acute symptoms and adjust the patient's constitution, usually using gentle treatment methods. The improvement stage focuses on the gradual reduction of symptoms and the recovery of function, at which point the intensity and frequency of treatment may gradually increase. The consolidation stage aims to maintain efficacy and prevent recurrence, with the frequency of treatment potentially decreasing, focusing on maintaining the patient's health. These stages are determined based on the patient's treatment response and efficacy progress, helping to optimize treatment plans and control the sustainability of treatment effects.

[0107] The state combination and action chain mechanism is used to process each treatment stage, and a state space and corresponding action space are established based on the processing results.

[0108] It should be further explained that state combination refers to combining various states during the treatment process (such as therapeutic effects at different time points, patient physiological feedback, etc.) to form a complete state space. Each state can reflect the patient's specific situation at a certain treatment stage, such as physiological indicators such as pain reduction, body temperature changes, and pulse stability. Action chain mechanism, on the other hand, refers to optimizing and controlling each operation during the treatment process (such as adjusting temperature, changing acupoints, adjusting the amount of moxa wool administered) as a continuous action chain under specific states. For example, in the initial adjustment stage, the focus is on temperature regulation and acupoint selection, while in the improvement stage, the emphasis is on the gradual increase of temperature and the adjustment of treatment frequency.

[0109] By treating patients of the same type as intelligent agents, and combining the established state space with the corresponding action space, a reinforcement learning environment simulating the fire dragon moxibustion treatment process is constructed, and the state transition mechanism and reward signal of the learning environment are strengthened.

[0110] Based on the defined reinforcement learning environment, transfer learning techniques are used to learn relevant treatment knowledge and experience from historical efficacy data. Based on the learning results, the optimal treatment strategy is generated and transferred as prior knowledge to the initial long-term efficacy prediction model.

[0111] Based on the patient's physiological feedback data, the initial long-term efficacy prediction model after migration was optimized using the difference-in-differences algorithm to obtain the patient's long-term efficacy prediction model.

[0112] The efficacy prediction model selection module 302 is used to dynamically select between short-term and long-term prediction models to predict efficacy based on the patient's individualized treatment goals and current treatment stage.

[0113] It's important to clarify that for patients requiring rapid assessment of treatment effectiveness, short-term predictive models offer immediate feedback at the initial stage or a specific phase of treatment. For patients concerned with treatment sustainability and long-term outcomes, long-term predictive models help assess the long-term impact of treatment. For example, when a patient is in the early stages of treatment and the goal is rapid symptom relief or to determine treatment effectiveness, a short-term predictive model is preferable. This model focuses on predicting short-term efficacy (e.g., within 1 or 2 weeks), helping to adjust the treatment plan promptly. However, for patients in the middle or later stages of treatment, where the goals involve the sustainability of efficacy and long-term disease control, long-term predictive models are more suitable. These models analyze the patient's long-term physiological feedback data (e.g., temperature fluctuations, symptom persistence, quality of life) and historical efficacy data to predict the long-term efficacy of treatment, such as the effect after 3 or 6 months.

[0114] The optimal heating parameter acquisition module 303 is used to analyze the predicted results of therapeutic effects, establish a multi-objective optimization model, and use a multi-objective optimization algorithm to solve the multi-objective optimization model to obtain the optimal heating parameters.

[0115] In this optional embodiment, the optimal heating parameter acquisition module 303 includes:

[0116] The heating parameter range acquisition submodule is used to analyze the predicted results of therapeutic efficacy using a preset causal forest model, identify the causal effects of different heating parameters on the treatment effect, assess the degree of influence of each heating parameter on the treatment risk, and obtain the heating parameter range.

[0117] It should be further explained that the causal forest model is a machine learning method used to identify causal relationships between variables. In the context of traditional Chinese medicine physiotherapy, this model can analyze the causal effects between different heating parameters and treatment efficacy and risks. For example, the model can identify that the treatment effect is most significant within a certain temperature range, while excessively high or low temperatures may lead to decreased efficacy or increased risk of adverse reactions. Based on the degree of causal influence of each heating parameter on efficacy and safety, the model can further determine its optimal range of action. For example, if the analysis results show that a temperature of 40°C to 45°C can significantly improve the treatment effect while maintaining a low risk, then this range is determined to be the effective range of the heating parameter, and the heating parameter range is obtained.

[0118] The heating parameter set generation submodule is used to construct a multi-objective optimization model with the obtained heating parameter range as constraints, aiming to maximize the therapeutic effect and minimize the therapeutic risk. The multi-objective optimization algorithm is then used to solve the multi-objective optimization model to generate the optimal solution set as the optimal heating parameter set.

[0119] It should be noted that the optimal heating parameter set includes parameters such as the optimal temperature, heating time, and moxa wool feeding rate.

[0120] The therapeutic effect simulation and strategy generation unit 4 is used to analyze the patient's physiological feedback data, recommend candidate key acupoint combinations for the patient, and use the bio-thermal conduction simulation algorithm to simulate the therapeutic effect of the optimal heating parameter set on the candidate key acupoint combinations. Based on the simulation results, the optimal key acupoint combination is selected and a personalized optimal temperature control strategy is generated.

[0121] In this optional embodiment, the therapeutic effect simulation and strategy generation unit 4 includes:

[0122] The heterogeneous graph construction module 401 is used to extract physiological features from the collected patient physiological feedback data, and based on the attribute features of each key acupoint, analyze the correlation strength between the physiological features and the attribute features of the key acupoints, and construct a heterogeneous graph model with key acupoints as nodes and correlation strength as edges.

[0123] It should be further explained that physiological characteristics include body temperature, heart rate, blood pressure, respiratory rate, skin conductance, blood oxygen saturation, and electromyographic signals, reflecting the patient's current physiological state and treatment response; the attribute characteristics of key acupoints refer to the functional positioning of each key acupoint in traditional Chinese medicine theory, such as its meridian affiliation, therapeutic effects, indication types, depth of action, and thermal sensitivity; the correlation strength is the degree of matching or correlation between physiological characteristics and the attribute characteristics of key acupoints, which can be quantified based on statistical correlation (e.g., Pearson correlation coefficient), information gain, mutual information, or model-based weighting indicators. The larger the value, the stronger the response or the higher the regulatory capacity of the key acupoint to a certain physiological characteristic.

[0124] The correlation score module 402 is used to train the heterogeneous graph model using a graph neural network. It learns the importance weights of each key acupoint to the treatment response through a multi-layer message passing mechanism, and obtains the correlation score between key acupoints and therapeutic efficacy.

[0125] The candidate key acupoint combination generation module 403 is used to mine candidate key acupoint combinations with therapeutic synergy from the heterogeneous graph model based on the obtained correlation score using a greedy iterative algorithm and a collaborative filtering recommendation algorithm.

[0126] In this optional embodiment, the candidate key acupoint combination generation module 403 includes:

[0127] The collaborative similarity calculation submodule is used to treat each key acupoint in the heterogeneous graph model as an independent cluster and calculate the collaborative similarity between any two clusters based on the association strength.

[0128] The initial collaborative cluster formation submodule is used to use a greedy iterative algorithm to traverse all cluster pairs, select the two clusters with the highest collaborative similarity and merge them to form the initial collaborative cluster.

[0129] The collaborative cluster acquisition submodule is used to recalculate and update the collaborative similarity between the initial collaborative cluster and the remaining clusters until the collaborative similarity between any two clusters is lower than a preset threshold, at which point the iteration stops and several collaborative clusters are obtained.

[0130] The synergy effect value acquisition submodule is used to calculate the multidimensional efficacy score of all key acupoints in each synergy cluster based on the obtained correlation score, and to perform differential processing on the multidimensional efficacy score to obtain the synergy effect value.

[0131] The collaborative recommendation submodule is used to mine and generate candidate key acupoint combinations with therapeutic synergistic effects from the collaborative cluster based on the obtained synergistic effect values ​​using a collaborative filtering recommendation algorithm.

[0132] In this optional embodiment, based on the obtained synergistic effect value, the collaborative filtering recommendation algorithm is used to mine and generate candidate key acupoint combinations with therapeutic synergistic effects from the collaborative cluster, including:

[0133] Based on the synergistic cluster, the corresponding synergistic effect value, and the historical efficacy data of similar patients, a multidimensional feature matrix is ​​constructed, and a key acupoint is randomly selected from the multidimensional feature matrix as the target key acupoint.

[0134] Calculate the similarity index between the target key acupoint and the remaining key acupoints, use the important nearest neighbor algorithm to count the number of common nearest neighbors between the target key acupoint and the remaining key acupoints, and select key acupoints that meet the preset screening threshold from the remaining key acupoints as important nearest neighbor key acupoints of the target key acupoint.

[0135] The key acupoints in the collaborative cluster are designated as primary key acupoints, and the important nearest neighbor key acupoints are designated as secondary key acupoints. The collaborative filtering algorithm is used to filter the primary key acupoints to generate an initial candidate key acupoint combination.

[0136] The system uses a pre-defined category boosting tree model to predict and analyze the initial candidate key acupoint combinations. Combined with the patient's historical treatment data, it evaluates the efficacy of each initial candidate key acupoint combination and generates a generated candidate key acupoint combination with therapeutic synergy.

[0137] The simulation scoring module 404 is used to construct a three-dimensional heat conduction model based on the theory of biological heat conduction, and use the optimal heating parameter set as boundary conditions. The three-dimensional heat conduction model is used to simulate the temperature distribution, heat penetration depth and tissue response of different candidate key acupoint combinations during the treatment process, and outputs a simulation score based on the simulation results.

[0138] It should be noted that three-dimensional heat conduction models are usually based on the Pennes biological heat conduction equation, which takes into account factors such as tissue thermal conductivity, blood perfusion rate, and metabolic heat generation, and can realistically simulate the temperature change process of human tissue under thermal stimulation.

[0139] The strategy generation module 405 is used to select the optimal key acupoint combination from the candidate key acupoint combinations based on the output simulation score, combined with the physical constraints of the fire dragon moxibustion box and the preset simulation safety threshold, and combine the optimal key acupoint combination with the optimal heating parameter set to generate a personalized temperature control treatment strategy.

[0140] It should be further explained that, based on the output simulation score, combined with the physical constraints of the Fire Dragon Moxibustion Box and the preset simulation safety threshold, the optimal key acupoint combination is selected from the candidate key acupoint combinations. This optimal key acupoint combination is then combined with the optimal heating parameter set to generate a personalized temperature control treatment strategy, specifically including:

[0141] Based on the output simulation scores, all candidate key acupoint combinations are ranked and screened. A higher score indicates a better expected therapeutic effect under the current heating parameters. Further optimization and screening are performed considering the actual physical constraints of the moxibustion box, including technical indicators such as the device's maximum heating power, adjustable temperature range, and temperature control stability. For example, if the maximum allowable temperature of the moxibustion box is 50°C and the temperature control accuracy is ±2°C, solutions that cannot operate stably within this accuracy or exceed the temperature limit will be eliminated. Furthermore, a preset simulation safety threshold is introduced for safety verification. This safety threshold covers the safety of key parameters such as local skin temperature, heat penetration depth, and duration of heat stimulation. Range; for example, to avoid tissue burns, the local temperature is usually set not to exceed 45°C. Therefore, in the simulation, if a certain scheme causes the local temperature to exceed 44°C or the heat penetration depth to exceed the physiological tolerance range, the scheme will be judged as having a safety risk and will be eliminated. Combining the optimal heating parameter set (e.g., temperature 44°C, heating time 15 minutes, moxa wool feeding rate 0.8g / min) with the optimal combination of key acupoints, a treatment plan is constructed. This plan can be dynamically adjusted according to the patient's specific characteristics (e.g., body type, severity of condition, and stage of treatment) to form a precise temperature control treatment strategy, ensuring that the treatment process is optimized throughout the entire process while maximizing efficacy and safety.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A digital temperature control and therapeutic effect management system for moxibustion, characterized in that, The system includes: The temperature data acquisition unit is used to convert the analog temperature signal collected by the temperature measuring device into a digital signal using an analog-to-digital conversion channel, and to perform noise reduction and smoothing processing on the digital signal using a filtering algorithm to acquire temperature data. The temperature control signal generation and transmission unit is used to draw a target temperature curve based on the preset fire dragon moxibustion physiotherapy data management library, compare the target temperature curve with the acquired temperature data, combine the patient's physiological feedback data, construct the controller architecture, and use the controller architecture to generate the corresponding temperature control signal and transmit it to the fire dragon moxibustion box. The scoring and parameter acquisition unit is used to input the multidimensional efficacy scores of patients after each treatment through the software interface, combine temperature data and physiological feedback data to build an efficacy prediction model, and analyze the efficacy prediction results of the efficacy prediction model to obtain the optimal heating parameters. The therapeutic effect simulation and strategy generation unit is used to analyze the patient's physiological feedback data, recommend candidate key acupoint combinations for the patient, and use a bio-thermal conduction simulation algorithm to simulate the therapeutic effect of the optimal heating parameter set on the candidate key acupoint combinations. Based on the simulation results, the optimal key acupoint combination is selected and a personalized optimal temperature control strategy is generated.

2. The digital temperature control and therapeutic effect management system for moxibustion according to claim 1, characterized in that, The temperature control signal generation and transmission unit includes: The multidimensional error acquisition module is used to draw the target temperature curve of moxibustion treatment based on the preset moxibustion therapy data management library and the patient's personalized treatment needs, and compare the acquired temperature data with the target temperature curve to obtain the multidimensional error. An adaptive control architecture construction module is used to collect patients' physiological feedback data in real time. The multidimensional error and physiological feedback data are used as the state space, and the height adjustment of the soft mesh inside the moxibustion box and the control of the moxa wool feeding amount are used as the action space. The corresponding reward function is designed and combined with the adaptive control algorithm to construct the adaptive control architecture. The temperature control signal generation module is used to construct a time-series prediction model of temperature change trend using a long short-term memory network. Combined with the collected historical temperature data, the time-series prediction model is used to predict the temperature change trend in the future time period, and the prediction results are input into the adaptive control architecture to generate the corresponding temperature control signal. The signal verification and transmission module is used to input the generated temperature control signal into the preset heat conduction physical model for simulation verification, perform safety verification on the simulation verification results in combination with the preset safety threshold, and transmit the temperature control signal that has passed the safety verification to the actuator of the Fire Dragon Moxibustion Box.

3. The digital temperature control and therapeutic effect management system for moxibustion according to claim 1, characterized in that, The scoring input and parameter acquisition unit includes: The efficacy prediction model construction module is used to construct an efficacy prediction model based on the entered multidimensional efficacy score and the acquired temperature data and physiological feedback data. The efficacy prediction model includes a short-term efficacy prediction model and a long-term efficacy prediction model. The efficacy prediction model selection module is used to dynamically select between short-term and long-term prediction models to predict efficacy based on the patient's individualized treatment goals and current treatment stage. The optimal heating parameter acquisition module is used to analyze the predicted results of therapeutic effects, establish a multi-objective optimization model, and use a multi-objective optimization algorithm to solve the multi-objective optimization model to obtain the optimal heating parameters.

4. The digital temperature control and therapeutic effect management system for moxibustion according to claim 3, characterized in that, The optimal heating parameter acquisition module includes: The heating parameter range acquisition submodule is used to analyze the predicted results of therapeutic efficacy using a preset causal forest model, identify the causal effects of different heating parameters on the treatment effect, assess the degree of influence of each heating parameter on the treatment risk, and obtain the heating parameter range. The heating parameter set generation submodule is used to construct a multi-objective optimization model with the obtained heating parameter range as constraints, aiming to maximize the therapeutic effect and minimize the therapeutic risk. The multi-objective optimization algorithm is then used to solve the multi-objective optimization model to generate the optimal solution set as the optimal heating parameter set.

5. The digital temperature control and therapeutic effect management system for moxibustion according to claim 1, characterized in that, The therapeutic effect simulation and strategy generation unit includes: The heterogeneous graph construction module is used to extract physiological features from the collected patient physiological feedback data, and based on the attribute features of each key acupoint, analyze the correlation strength between the physiological features and the attribute features of the key acupoints, and construct a heterogeneous graph model with key acupoints as nodes and correlation strength as edges. The correlation score module is used to train the heterogeneous graph model using a graph neural network. It learns the importance weights of each key acupoint to the treatment response through a multi-layer message passing mechanism, and obtains the correlation score between key acupoints and therapeutic efficacy. The candidate key acupoint combination generation module is used to mine candidate key acupoint combinations with therapeutic synergy from the heterogeneous graph model based on the obtained correlation score using a greedy iterative algorithm and a collaborative filtering recommendation algorithm. The simulation scoring module is used to construct a three-dimensional heat conduction model based on the theory of biological heat conduction, and use the optimal heating parameter set as boundary conditions. The three-dimensional heat conduction model is used to simulate the temperature distribution, heat penetration depth and tissue response of different candidate key acupoint combinations during the treatment process, and outputs a simulation score based on the simulation results. The strategy generation module is used to select the optimal key acupoint combination from the candidate key acupoint combinations based on the output simulation score, combined with the physical constraints of the fire dragon moxibustion box and the preset simulation safety threshold, and combine the optimal key acupoint combination with the optimal heating parameter set to generate a personalized temperature control treatment strategy.

6. The digital temperature control and therapeutic effect management system for moxibustion according to claim 5, characterized in that, The candidate key acupoint combination generation module includes: The collaborative similarity calculation submodule is used to treat each key acupoint in the heterogeneous graph model as an independent cluster and calculate the collaborative similarity between any two clusters based on the association strength. The initial collaborative cluster formation submodule is used to use a greedy iterative algorithm to traverse all cluster pairs, select the two clusters with the highest collaborative similarity and merge them to form the initial collaborative cluster. The collaborative cluster acquisition submodule is used to recalculate and update the collaborative similarity between the initial collaborative cluster and the remaining clusters until the collaborative similarity between any two clusters is lower than a preset threshold, at which point the iteration stops and several collaborative clusters are obtained. The synergy effect value acquisition submodule is used to calculate the multidimensional efficacy score of all key acupoints in each synergy cluster based on the obtained correlation score, and to perform differential processing on the multidimensional efficacy score to obtain the synergy effect value. The collaborative recommendation submodule is used to mine and generate candidate key acupoint combinations with therapeutic synergistic effects from the collaborative cluster based on the obtained synergistic effect values ​​using a collaborative filtering recommendation algorithm.

7. The digital temperature control and therapeutic effect management system for moxibustion according to claim 6, characterized in that, Based on the obtained synergistic effect values, the collaborative filtering recommendation algorithm is used to mine and generate candidate key acupoint combinations with therapeutic synergistic effects from the collaborative cluster, including: Based on the synergistic cluster, the corresponding synergistic effect value, and the historical efficacy data of similar patients, a multidimensional feature matrix is ​​constructed, and a key acupoint is randomly selected from the multidimensional feature matrix as the target key acupoint. Calculate the similarity index between the target key acupoint and the remaining key acupoints, use the nearest neighbor algorithm to count the number of common nearest neighbors between the target key acupoint and the remaining key acupoints, and select key acupoints that meet the preset screening threshold from the remaining key acupoints as important nearest neighbor key acupoints of the target key acupoint. The key acupoints in the collaborative cluster are designated as primary key acupoints, and the important nearest neighbor key acupoints are designated as secondary key acupoints. The collaborative filtering algorithm is used to filter the primary key acupoints to generate an initial candidate key acupoint combination. The system uses a pre-defined category boosting tree model to predict and analyze the initial candidate key acupoint combinations. Combined with the patient's historical treatment data, it evaluates the efficacy of each initial candidate key acupoint combination and generates a generated candidate key acupoint combination with therapeutic synergy.

Images