Intelligent relieving system and method for magical pain after breast resection based on biological feedback

Abstract

The invention provides an intelligent relieving system and method for phantom pain after breast resection based on biofeedback, develops a non-drug and non-invasive intelligent system, recognizes the phantom pain state in real time through multi-dimensional biofeedback, dynamically adjusts an intervention strategy, and solves the problems of high dependency and poor universality of a traditional method. Specifically, through multi-modal signal fusion, sEMG (surface electromyography), heart rate variability (HRV) and skin conductance (EDA) data are integrated, and the accuracy of magic pain recognition is improved. Through an adaptive algorithm, feedback parameters (such as electrical stimulation intensity and virtual reality scenes) are dynamically optimized based on an AI analgesia intervention model (such as LSTM), and personalized intervention is realized. According to the invention, a dynamic adjustment mechanism is integrated for magic pain for the first time, and the limitation of a traditional single intervention mode is broken through based on a self-adaptive adjustment strategy of a multi-modal signal. The occurrence rate of postoperative chronic pain is reduced, and consumption of medical resources is reduced. An intelligent tool is provided for nursing science, and the development of'precise nursing 'is promoted.

Description

Technical Field

[0001] This invention relates to the field of clinical medical technology, and in particular to an intelligent pain relief system, application method, electronic device and computer-readable storage medium for postoperative phantom pain relief based on biofeedback. Background Technology

[0002] Phantom breast pain (PBP) is a common complication after breast cancer surgery. Patients typically experience persistent pain or paresthesia in the removed breast, with an incidence rate as high as 20%-68%. Its mechanism is related to nerve damage, central sensitization, and psychological factors. Traditional analgesia methods (such as oral medications and nerve blocks) have limitations such as strong drug dependence, high risk of invasive procedures, and lack of personalized intervention. Current technologies face the following technical bottlenecks in the clinical treatment of postoperative phantom pain in breast cancer patients: (1) Traditional biofeedback devices (such as EMG and sEMG) are mostly used for chronic pain or rehabilitation training, but lack personalized dynamic regulation mechanisms for phantom pain. (2) Existing systems (such as rehabilitation training equipment based on accelerometers) focus on motion monitoring and do not integrate multimodal biosignals (such as heart rate and electromyography) with intelligent algorithms, making it difficult to relieve phantom pain in real time; (3) Drug dependence: Opioids are prone to causing nausea, constipation and addiction, and their effects decrease with long-term use; (4) Insufficient psychological intervention: Patients’ anxiety, depression and other psychological states exacerbate pain perception, and existing methods do not integrate real-time psychological assessment modules; (5) Lack of intelligence: There is a lack of personalized regulation systems based on multimodal physiological signals (such as electromyography and heart rate variability). Recent studies have shown that combining biofeedback with neuroplasticity training can effectively alleviate neuropathic pain, while smart wearable devices and AI algorithms provide new pathways for real-time monitoring and dynamic intervention. Summary of the Invention

[0003] To address the technical problems existing in the prior art, the present invention provides the following technical solution: On the one hand, a biofeedback-based intelligent pain relief system for post-mastectomy surgery is provided, the system comprising: Wearable device for collecting multimodal signals from mastectomy patients: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA) and uploading them to a cloud management platform; and, in response to an adaptive adjustment strategy issued by the cloud management platform, activating a biofeedback module and / or a psychological module to intervene in phantom pain in mastectomy patients. A cloud-based management platform is used to record the multimodal signals of patients after mastectomy, identify the multimodal signals and determine the patient's pain level and / or psychological load through a pre-deployed AI analgesia intervention model, and dynamically match corresponding adaptive adjustment strategies based on the pain level and / or psychological load and distribute them to the wearable device. The wearable device is communicatively connected to the cloud management platform. Preferably, the cloud management platform is further used for: Provides cloud storage services for patients who have undergone mastectomy. Preferably, the cloud management platform is further used for: Personalized rehabilitation suggestions for relieving phantom pain after mastectomy are generated based on the pain level and sent to the patient. Preferably, the system further includes: The patient's end is used to receive and respond to the personalized rehabilitation recommendations; The patient terminal is connected to the cloud management platform. Preferably, the wearable device includes: Biosensors are used to sense and acquire multimodal signals from patients after mastectomy: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA). A processor for preprocessing the multimodal signals; A communication module is used to upload the multimodal signal to the cloud management platform; and to receive adaptive adjustment strategies dynamically issued by the cloud management platform. The controller is used to respond to the adaptive adjustment strategy by controlling the biofeedback module and / or the psychological module to initiate phantom pain intervention; The biofeedback module is used to provide biostimulation interventions to patients. The psychological module is used to provide psychological guidance and intervention for patients. Power supply, used for supplying power; The biosensor is electrically connected to the processor; The processor, communication module, biofeedback module, psychological module, and power supply are all electrically connected to the controller. The wearable device communicates with the cloud management platform via a communication module. Preferably, the psychological module is a VR device used to play VR audio and video data in the adaptive adjustment strategy. Preferably, the generation method of the AI ​​analgesia intervention model includes: We collected intervention big data from several mastectomy patients, including their multimodal signals and historical intervention protocols. The patient's multimodal signals were subjected to feature quantification analysis to assess the patient's pain level and / or psychological burden; Label the patient's pain level with corresponding biostimulation intervention strategies, and / or label the patient's psychological load with corresponding VR audio and video data; Collect characteristic data for each patient: {pain level, psychological burden, intervention plan}, and form a feature set; Dynamically match corresponding adaptive adjustment strategies based on pain level and / or psychological load. The feature set is divided into a training set and a validation set according to the proportions. Input the training set into a pre-defined machine learning model to train and learn features; The trained AI analgesia intervention model is validated using a validation set. Once the validation is successful, the AI ​​analgesia intervention model is deployed and applied. On the other hand, an application method is provided, which is based on the above-described intelligent pain relief system for post-mastectomy phantom pain based on biofeedback. The method includes: Activate the wearable device and establish communication with the cloud management platform; The cloud management platform creates a post-mastectomy patient file and initiates an initial calibration procedure, collecting the patient's physiological parameters at rest as a benchmark for calibration. The cloud management platform issues sampling instructions to begin real-time monitoring and collection of multimodal signals from post-mastectomy patients via wearable devices: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA), and uploads them to the cloud management platform. The cloud management platform records the multimodal signals of patients after mastectomy into the patient's file, and The pre-deployed AI analgesia intervention model identifies the multimodal signals and determines the patient's pain level and / or psychological burden; The cloud management platform dynamically matches corresponding adaptive adjustment strategies based on pain level and / or psychological load and distributes them to the wearable device. The wearable device responds to the adaptive adjustment strategy issued by the cloud management platform and activates the biofeedback module and / or psychological module to intervene in phantom pain in patients after mastectomy. On the other hand, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions, which, when executed by the processor, implement the application method as described above. On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement the above-described application method. The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: This invention develops a non-drug, non-invasive intelligent system that identifies phantom pain in real time through multi-dimensional biofeedback and dynamically adjusts intervention strategies, addressing the problems of strong dependence and poor universality of traditional methods. Specifically, it improves the accuracy of phantom pain identification by integrating sEMG (surface electromyography), heart rate variability (HRV), and electrical conductance of the skin (EDA) data through multi-modal signal fusion. Adaptive algorithms dynamically optimize feedback parameters (such as electrical stimulation intensity and virtual reality scenarios) based on AI analgesia intervention models (such as LSTM) to achieve personalized intervention. This invention is the first to integrate a dynamic regulation mechanism for phantom pain, using an adaptive regulation strategy based on multi-modal signals, breaking through the limitations of traditional single intervention models. It reduces the incidence of postoperative chronic pain and decreases the consumption of medical resources. It provides intelligent tools for nursing and promotes the development of "precision nursing." Attached Figure Description

[0004] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a schematic diagram of the topological structure of an intelligent pain relief system based on biofeedback after mastectomy, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the control system structure of a wearable device provided in an embodiment of the present invention; Figure 3 This is a flowchart of a model training method provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0005] The technical solution of the present invention will now be described with reference to the accompanying drawings. In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one. In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. In this embodiment of the invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same. To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments. This invention provides a biofeedback-based intelligent pain relief system for post-mastectomy surgery and its application. The method can be implemented using an electronic device, which can be a terminal or a server. Figure 1 The diagram shown is a topology of a biofeedback-based intelligent pain relief system for post-mastectomy phantom pain. The system includes: Wearable device for collecting multimodal signals from mastectomy patients: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA) and uploading them to a cloud management platform; and, in response to an adaptive adjustment strategy issued by the cloud management platform, activating a biofeedback module and / or a psychological module to intervene in phantom pain in mastectomy patients. A cloud-based management platform is used to record the multimodal signals of patients after mastectomy, identify the multimodal signals and determine the patient's pain level and / or psychological load through a pre-deployed AI analgesia intervention model, and dynamically match corresponding adaptive adjustment strategies based on the pain level and / or psychological load and distribute them to the wearable device. The wearable device is communicatively connected to the cloud management platform. This invention develops a non-drug, non-invasive intelligent system that identifies phantom pain in real time through multi-dimensional biofeedback and dynamically adjusts intervention strategies, addressing the problems of strong dependence and poor universality of traditional methods. Specifically, it improves the accuracy of phantom pain identification by integrating sEMG (surface electromyography), heart rate variability (HRV), and electrical conductance of the skin (EDA) data through multi-modal signal fusion. Adaptive algorithms dynamically optimize feedback parameters (such as electrical stimulation intensity and virtual reality scenarios) based on AI analgesia intervention models (such as LSTM) to achieve personalized intervention. This invention is the first to integrate a dynamic regulation mechanism for phantom pain, using an adaptive regulation strategy based on multi-modal signals, breaking through the limitations of traditional single intervention models. It reduces the incidence of postoperative chronic pain and decreases the consumption of medical resources. It provides intelligent tools for nursing and promotes the development of "precision nursing." The system architecture is as follows: Perception layer: integrates wearable sensors (the basal layer has positioning marks corresponding to the Tanzhong and Qimen acupoints on its inner surface (the positioning marks are determined by 3D human body scanning modeling), electromyography sensor, heart rate monitoring module, and skin conductivity meter) to collect physiological signals of the chest wall and residual limb area in real time. The analysis layer cloud management platform uses embedded AI chips to analyze signal characteristics through machine learning models (such as random forest and LSTM) to identify the threshold for phantom pain attacks and the level of psychological stress. Intervention layer: Biofeedback module: It regulates abnormal electromyographic activity and inhibits pain signal transmission through electrical stimulation or vibration feedback; Psychological intervention: Customized virtual scenarios (such as natural landscapes) guide patients to perform deep breathing and progressive muscle relaxation to reduce anxiety levels. The cloud-based management platform can also store patient data and support remote monitoring and parameter optimization by doctors. In this embodiment, wearable sensors, biofeedback modules, and psychological modules are integrated into the wearable device. The integration method can refer to the structure and principle of existing medical wearable devices, and this embodiment does not limit it. For example, integration on the device: The wearable device integrates a MyoWare electromyography sensor (sampling rate 1000Hz), a PPG optical heart rate module (accuracy ±2bpm), and a GSR skin conductance sensor. It accurately collects chest wall electromyographic activity and autonomic nerve responses through acupoint positioning markers (error <1mm). Wearable devices use the BLE 5.0 protocol to upload sEMG time-frequency characteristics (MFCC), HRV LF / HF ratio, and EDA rise slope to the cloud every 5 seconds. AI pain recognition engine: Deploy LSTM-Transformer hybrid model or other machine learning models in the cloud. The input layer accepts 128-dimensional feature vectors (including root mean square value of electromyography, NN50 heart rate variability index, etc.) and outputs pain level (VAS 0-10) and psychological burden index (PLI). Dynamic threshold algorithm: When the sEMG amplitude is >150μV and the EDA slope is >0.5μS / s, a phantom pain warning is triggered. Adaptive intervention mechanism: Biofeedback module: The parameters of transcutaneous electrical stimulation (TENS) are dynamically adjusted (frequency 4-100Hz, pulse width 50-200μs) to inhibit abnormal neural discharges through STDP pulse time-dependent plasticity. Psychology Module: Unity3D generates personalized VR scenes (such as forest meditation), and adjusts the ambient light intensity (200-1000 lux) and the guided speech rate (60-120 words / minute) according to the PLI value. Therefore, the system can improve the accuracy and response efficiency of phantom pain recognition, and dynamically adjust it online in real time. Preferably, the cloud management platform is further used for: Provides cloud storage services for patients who have undergone mastectomy. Regarding cloud storage technology, users can choose cloud storage services themselves. Preferably, the cloud management platform is further used for: Personalized rehabilitation suggestions for relieving phantom pain after mastectomy are generated based on the pain level and sent to the patient. The cloud management platform uses the AI ​​pain recognition engine "AI analgesia intervention model" to identify the multimodal signals and determine the patient's pain level and / or psychological load, enabling intelligent decision-making. For example, it can intelligently decide on corresponding biostimulation intervention strategies or VR audio and video data based on the pain level (the model identifies features and outputs annotation information, and the system responds based on the output results, such as accessing and playing VR audio and video based on the address, and sending the data stream to the patient for broadcast). Example as follows: def generate_plan(pain_level): if pain_level <= 3: return {"Physical Therapy": "Transcutaneous Electrical Stimulation 15min / session", "Medication": "Pregabalin 75mg qd"} elif 4<=pain_level<=6: return {"Nerve Block": "Intercostal Nerve Cryotherapy", "Psychological Intervention": "Mindfulness Training 20 min / day"} else: return {"Minimally Invasive Surgery": "Spinal Cord Stimulation Implantation", "Emergency Contact": "Pain Management Consultation"} Clinical application cases Case 1 (52-year-old female, NRS=2): Recommended treatment plan: Mirror therapy (3 times daily) + hot compress care; Results: Pain level dropped to 1 point after 2 weeks. Case 2 (38-year-old female, NRS=7): Trigger alert: Automatically schedule pain management team (MDT) consultations; Final treatment: NRS score dropped to 4 after pulsed radiofrequency therapy. Preferably, the system further includes: The patient's end is used to receive and respond to the personalized rehabilitation recommendations; The patient terminal is connected to the cloud management platform. The patient's end can be an app or a mini-program, which allows them to log in to the cloud management platform and receive personalized rehabilitation suggestions pushed by the platform. like Figure 2 As shown, preferably, the wearable device includes: Biosensors are used to sense and acquire multimodal signals from patients after mastectomy: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA). A processor for preprocessing the multimodal signals; A communication module is used to upload the multimodal signal to the cloud management platform; and to receive adaptive adjustment strategies dynamically issued by the cloud management platform. The controller is used to respond to the adaptive adjustment strategy by controlling the biofeedback module and / or the psychological module to initiate phantom pain intervention; The biofeedback module is used to provide biostimulation interventions to patients. The psychological module is used to provide psychological guidance and intervention for patients. Power supply, used for supplying power; The biosensor is electrically connected to the processor; The processor, communication module, biofeedback module, psychological module, and power supply are all electrically connected to the controller. The wearable device communicates with the cloud management platform via a communication module. For a list of specific types of biosensors, please refer to the preceding descriptions. Wearable devices can be configured as follows: 1. Biosensor Module sEMG acquisition: The Delsys Trigno Flex sensor (bandwidth 20-450Hz) was used to detect residual electromyographic signals of the pectoralis major muscle through surface electrodes. HRV monitoring: Integrated Maxim MAX86150 optical sensor (PPG+ECG dual-mode), sampling rate 256Hz; EDA detection: Empatica E4 dry electrode array (0.05-5μS resolution) was used. 2. Processing and Control Unit Main processor: Nordic nRF5340 dual-core MCU (Cortex-M33), responsible for signal filtering (FIR filter) and feature extraction. Communication module: Sierra Wireless HL7688 (LTE Cat-M1), supporting TLS 1.3 encrypted transmission. Controller: STM32U585AI (with hardware security module), responsible for execution strategy parsing and module scheduling. 3. Intervention Module Biofeedback: TENS neurostimulation unit (0-100mA adjustable pulse); Psychological intervention: Miniature bone conduction loudspeakers (frequency response 100Hz-10kHz) combined with audio from cognitive behavioral therapy. 4. Power system: 3.7V / 500mAh flexible battery (supports Qi wireless charging) Working principle and process: 1. Signal Acquisition Stage SEMG electrode detection of abnormal muscle discharge (characteristic frequency of phantom pain attack 8-13Hz), PPG / ECG composite algorithm to calculate RMSSD index of HRV, EDA electrode to capture skin conductance fluctuations caused by sympathetic nerve excitation, cloud interaction stage. The HL7688 module uploads encrypted data packets to the AWS IoTCore platform to receive the adjustment strategies (such as TENS intensity parameters and psychological intervention audio numbers). 2. Intervention Implementation Phase The controller initiates dual-mode intervention based on the strategy: Biofeedback: 0.5ms square wave pulse stimulation of intercostal nerves; Psychological module: Plays alpha wave-inducing voice guidance. Customized virtual scenes (such as natural landscapes) guide patients to perform deep breathing and progressive muscle relaxation to reduce anxiety levels. Before data collection, the system can initialize calibration: After the patient wears the device, the system automatically performs baseline calibration: collecting resting physiological parameters as a benchmark. Monitoring only begins after calibration. Preferably, the psychological module is a VR device used to play VR audio and video data from the adaptive adjustment strategy. The VR device can be integrated into a wearable device via hooks or similar means, enabling communication with the wearable device to synchronously transmit and play VR audio and video streams, thus facilitating patient use. like Figure 3 As shown, preferably, the generation method of the AI ​​analgesia intervention model includes: We collected intervention big data from several mastectomy patients, including their multimodal signals (sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA)) and historical intervention protocols. The patient's multimodal signals were subjected to feature quantification analysis to assess the patient's pain level and / or psychological burden; Label the patient's pain level with corresponding biostimulation intervention strategies, and / or label the patient's psychological load with corresponding VR audio and video data; Collect characteristic data for each patient: {pain level, psychological burden, intervention plan}, and form a feature set; Dynamically match corresponding adaptive adjustment strategies based on pain level and / or psychological load. The feature set is divided into a training set and a validation set according to the proportions. Input the training set into a pre-defined machine learning model to train and learn features; The trained AI analgesia intervention model is validated using a validation set. Once the validation is successful, the AI ​​analgesia intervention model is deployed and applied. This invention's embedded AI chip analyzes signal features using machine learning models (such as random forest and LSTM) to identify phantom pain attack thresholds and psychological stress levels. This embodiment uses an RF model to train and construct an AI-based analgesic intervention model. The steps are as follows: I. Model Training Process 1. Data collection and annotation Multimodal signal source: Physiological signals: Heart rate variability (HRV), electrical skin response (EDA), surface electromyography (sEMG), such as acquiring pectoralis major muscle EMG signals using a Noraxon Ultium device (sampling rate 2000Hz); HRV RMSSD and LF / HF ratio were calculated using a Polar H10 heart rate strap; EDA: peak frequency of skin conductance response was recorded using an Empatica E4 wristband. Behavioral data: frequency of analgesic use, number of times body position was adjusted; Psychological assessment: HADS scale score; Pain level assessment: NRS scale; Intervention strategy labeling (can be done by healthcare professionals based on patient assessment results): Biostimulation: Transcutaneous electrical nerve stimulation (TENS) parameters (frequency 2-100Hz, pulse width 50-200μs); VR content is categorized into three main types: natural scenes, social simulations, and cognitive training, with a total of 12 sub-types. Each sub-type is labeled with an audio / video file address ID, and the system will automatically read the file with that ID in subsequent outputs. 2. Feature Engineering # Feature extraction example (Python pseudocode) def extract_features(signals): time_features = [np.mean(HRV), np.std(EDA)] freq_features = [psd(sEMG, band=(8,13))] cross_features = pearsonr(HRV,EDA) return pd.DataFrame([time_features + freq_features + [cross_features]]) Construct patient characteristic groups: {pain level, psychological burden, intervention plan}. Construct feature sets from the characteristic groups of several patients, and partition the data: Training set: 80% (n=320 cases); Validation set: 20% (n=80 cases). 3. Model Building Infrastructure: A RF model consisting of 500 decision trees. from sklearn.ensemble import RandomForestClassifier model = RandomForestClassifier( n_estimators=200, max_depth=10, class_weight='balanced' ) model.fit(X_train, y_train) Key parameters: max_depth=8, min_samples_split=5 class_weight='balanced', oob_score=True. II. Validation Data Table (n=120 cases) III. Application Cases Case 1 A 45-year-old post-breast cancer surgery patient with an initial VAS score of 7.2 and a HADS anxiety subscale score of 14 (cutoff value): System Response Real-time monitoring: EDA surge + HRV decrease triggers early warning Strategy generation: Preferred intervention: TENS (80Hz, 150μs) + forest VR scene; Alternative option: Endorphin-stimulated music + guided breathing. Effect verification: After 30 minutes, the VAS score dropped to 4.1 and the anxiety score decreased to 9. Case 2 Postoperative acute pain intervention: Input data: sEMG amplitude 120μV, HRV-LF / HF=3.2, EDA peak value 0.8μS; Model prediction: Pain level 4 (NRS scale); Implementation strategy: TENS therapy (frequency 100Hz) + VR natural scene intervention; Effect: Pain level dropped to level 2 after 30 minutes. Case 3: Chronic Pain Management Input data: sEMG electromyography entropy value 1.8, HRV-RMSSD=22ms, EDA fluctuates slowly; Model output: High risk of psychological stress; Intervention plan: Customized VR meditation program + biofeedback training; Result: The HADS score decreased by 40% after 7 days. IV. Model Optimization Path Incremental learning: Update data with 10% new cases weekly; Multi-model fusion: Combining LSTM to process temporal features; Interpretability: SHAP value analysis showed that the HRV standard deviation contributed 37% to pain prediction. This system can reduce the amount of analgesics patients need and shorten the average length of hospital stay. The architecture of the RF model and its general model training and application principles can be found in existing descriptions. Random forest models are a powerful and widely used ensemble learning method, known for their high accuracy, resistance to overfitting, and ease of parallelization. I. RF Model Architecture Random forests are essentially a model based on the Bagging ensemble learning strategy, with numerous decision trees as its core component. Its "forest" consists of multiple decision trees. The key characteristic of its architecture is "randomness," which is mainly reflected in two aspects: Bagging - Sample randomness (row randomness) To construct each decision tree in the forest: Training set generation: A subset of samples (called the Bootstrap sample set or Bagging sample set) is randomly drawn with replacement from the original training dataset. Effect: Each sample set drawn is a random subset of the original dataset. Because sampling is done with replacement, some samples may be selected multiple times in a single draw (potentially accounting for 30-40% of the sample set), while other samples may not be selected at all (called out-of-bag samples). This sampling method allows each decision tree to be trained on a dataset that is slightly different but has a basically similar data distribution, increasing the diversity between trees. Feature subset randomness (column randomness) For each decision tree in the forest: When splitting each node in the tree: Feature selection: Instead of selecting the optimal split point from all features, M features (M is much smaller than the total number of features N) are randomly selected to form a subset of candidate features. Optimal split point calculation: Then, the optimal split point for each feature is calculated only in this randomly selected subset M of features (e.g., based on criteria such as Gini impurity, information gain, etc.), and the split point that performs best is selected for that node. M is typically set to sqrt(N) (for classification tasks) or N / 3 (for regression tasks). This value is a crucial hyperparameter. Effect: This further enhances the diversity between each tree. Even if some globally strong features exist, they may not be selected into the candidate feature subset when a node is split in a certain tree. This reduces the correlation between trees, allowing the model to focus more on different aspects of the data. This improves the robustness of the model and enhances its ability to resist noise and redundant features. Core architecture summary: Parallel training: T decision trees are trained independently and in parallel. Double randomness: Each tree is trained based on its unique bootstrap sample (sample random) and a unique subset of features when nodes split (feature random). Simple base classifiers: A single tree itself serves as a base classifier or base regressor. These are typically decision trees that are unpruned or lightly pruned (e.g., limiting the maximum depth or the minimum number of samples per leaf node), allowing them to have high variance but low bias (i.e., it is okay to overfit a single tree first). II. Training Application Principles Training process Set hyperparameters: ntrees: The number of decision trees T in the forest (usually tens to hundreds, or even thousands). mtry / max_features: The number of features M randomly selected when a node splits. Parameters for a single tree (optional): such as the maximum depth of the tree (max_depth), the minimum number of samples per leaf node (min_samples_leaf), and the minimum number of samples per split node (min_samples_split). RF typically does not impose as strict a limit on the depth of a single tree as it does on a single pruned tree. Bootstrap: Whether to use sampling with replacement (usually True). Build each tree independently: For t in 1 to T: Create a Bootstrap sample set D_t: randomly draw |D| samples with replacement from the original dataset D. Create the root node of the current tree. Recursive training tree Tree_t: Starting from the root node: Stopping Criteria: Check if the current node meets the predefined stopping criteria (such as reaching the maximum depth, the number of samples in the node being less than min_samples_split or min_samples_leaf, the node purity reaching the threshold, or no features that can be distinguished). If the criteria are met, the node becomes a leaf node, and its value is set to the mode (classification) or mean (regression) of the samples in the node. Division process: If the stopping condition is not met: Candidate feature set: M features are randomly selected from all N features to form a candidate feature subset. Finding the optimal split point: On this candidate set of M features, calculate the split criteria score for each feature and all its possible split points (Gini impurity or information gain is commonly used for classification; mean squared error or MAE is commonly used for regression). Apply optimal splitting: Select the feature with the best score and the splitting point combination to split the current node into child nodes. Recursively construct subtrees: Recursively call step 2 for each child node until the stopping condition is met. (Optional) Estimate OOB error: For the samples that the tree did not use during training (out-of-bag samples OOB_t), record the tree's predictions for them. OOB samples from all trees can be aggregated into a large pseudo-validation set, which can be used to evaluate the overall generalization error of the forest (out-of-bag estimation) without a dedicated validation set, and can also be used to calculate feature importance. Prediction / Application Process Input: A new input sample x. Single-tree prediction: When a sample x is fed into each decision tree (Tree_1, Tree_2, ..., Tree_T) in the forest, each tree will independently output a prediction result: Classification problem: Output a class label y_i for each tree. Regression problem: Output a numerical value y_i for each tree. Ensemble prediction: Classification problem: A majority voting rule is used. The class labels output by all trees T are counted, and the prediction result is the class that received the most votes. Regression problem: An averaging rule is used. The average of all tree T output values ​​is calculated as the final predicted value. (Optional probability estimation): In classification, the frequency (vote rate) of each category predicted by the tree can also be used as a probability estimate of whether the sample belongs to that category. Core advantages and their connection to principles High accuracy: "Three cobblers are better than one Zhuge Liang." The majority voting / averaging mechanism integrates the results of numerous independent and diverse decision trees, effectively reducing the inherent variance (overfitting) of a single decision tree. Simultaneously, due to the high complexity (relatively low bias) of the base classifier (single tree), the overall model has high accuracy. Double randomness ensures diversity. Overfitting resistance (robustness): Its core stems from Bagging and feature randomness. Bagging: Bootstrap sampling and data perturbation reduce how sensitive the model is to small changes in the training data. Feature subsets: By forcing each tree to see only a subset of features, the global dominance of strong or noisy features is reduced, preventing the model from getting stuck in specific patterns of the training data. Ensembling unpruned trees actually avoids the difficulty of precisely pruning individual decision trees. Parallelized nonparametric models: make no strong assumptions about the distribution of the data. Handling high-dimensional / imbalanced data: Feature selection is naturally performed during the process and can work without special dimensionality reduction; imbalance can be handled by adjusting the Bootstrap sample weights or class weights. Output importance assessment: During training, the relative importance of features can be naturally estimated based on calculations (such as the average reduction in OOB prediction error of features) or permutation importance. No explicit validation set evaluation is required: out-of-bag (OOB) samples provide an inherent mechanism for estimating generalization ability (OOB error). Natural parallelization: The training process for all trees is independent and identical, making it ideal for accelerating training using multi-core CPUs or distributed systems. Random forests are an architecture that combines numerous decision trees based on a Bagging ensemble strategy. They generate diverse training sets through bootstrap random sampling with replacement and enforce diversity in the base learners by randomly selecting a subset of features at each decision tree node split. In application, the predictions from all base learners (decision trees) are aggregated using either majority voting (classification) or averaging (regression). This doubly random architecture and ensemble mechanism are the core principles behind the high accuracy, robustness, and wide applicability of random forests. They are a powerful tool for solving classification and regression problems. On the other hand, an application method is provided, which is based on the above-described intelligent pain relief system for post-mastectomy phantom pain based on biofeedback. The method includes: Activate the wearable device and establish communication with the cloud management platform; The cloud management platform creates a post-mastectomy patient file and initiates an initial calibration procedure, collecting the patient's physiological parameters at rest as a benchmark for calibration. The cloud management platform issues sampling instructions to begin real-time monitoring and collection of multimodal signals from post-mastectomy patients via wearable devices: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA), and uploads them to the cloud management platform. The cloud management platform records the multimodal signals of patients after mastectomy into the patient's file, and The pre-deployed AI analgesia intervention model identifies the multimodal signals and determines the patient's pain level and / or psychological burden; The cloud management platform dynamically matches corresponding adaptive adjustment strategies based on pain level and / or psychological load and distributes them to the wearable device. The wearable device responds to the adaptive adjustment strategy issued by the cloud management platform and activates the biofeedback module and / or psychological module to intervene in phantom pain in patients after mastectomy. The principles and implementation process of the method interaction can be understood in conjunction with the previous system. Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 4 As shown, electronic device 410 may include a first processor 2001. Optionally, the electronic device 410 may also include a memory 2002 and a transceiver 2003. The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus. The following is combined with Figure 4 A detailed description of each component of electronic device 410 is provided below: The first processor 2001 is the control center of the electronic device 410. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs). Optionally, the first processor 2001 can perform various functions of the electronic device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002. In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 4 CPU0 and CPU1 are shown in the diagram. In a specific implementation, as one example, the electronic device 410 may also include multiple processors, for example... Figure 4 The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions). The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here. Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and be connected via the interface circuit of the electronic device 410. Figure 4 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this. The transceiver 2003 is used to communicate with network devices or with terminal devices. Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 4 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function. Optionally, the transceiver 2003 can be integrated with the first processor 2001, or it can exist independently and be connected via the interface circuit of the electronic device 410. Figure 4 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this. It should be noted that, Figure 4 The structure of the electronic device 410 shown does not constitute a limitation on the router. Actual knowledge structure identification devices may include more or fewer components than shown, or combine certain components, or have different component arrangements. Furthermore, the technical effects of the electronic device 410 can be referred to the technical effects of the intelligent pain relief system based on biofeedback after mastectomy described in the above method embodiments, and will not be repeated here. It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM). The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable methods. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive. It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding. In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items. It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, methods, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the embodiments provided by this invention, it should be understood that the disclosed devices, methods, and approaches can be implemented in other ways. For example, the method embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, methods, or units, and may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. In addition, the functional units in the various embodiments of the present invention 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. If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a 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 several 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 described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A biofeedback-based intelligent pain relief system for post-mastectomy surgery, characterized in that, The system includes: Wearable device for collecting multimodal signals from mastectomy patients: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA) and uploading them to a cloud management platform; and, in response to an adaptive adjustment strategy issued by the cloud management platform, activating a biofeedback module and / or a psychological module to intervene in phantom pain in mastectomy patients. A cloud-based management platform is used to record the multimodal signals of patients after mastectomy, identify the multimodal signals and determine the patient's pain level and / or psychological load through a pre-deployed AI analgesia intervention model, and dynamically match corresponding adaptive adjustment strategies based on the pain level and / or psychological load and distribute them to the wearable device. The wearable device is communicatively connected to the cloud management platform.

2. The intelligent pain relief system based on biofeedback after mastectomy according to claim 1, characterized in that, The cloud management platform is also used for: Provides cloud storage services for patients who have undergone mastectomy.

3. The intelligent pain relief system based on biofeedback after mastectomy as described in claim 1, characterized in that, The cloud management platform is also used for: Personalized rehabilitation suggestions for relieving phantom pain after mastectomy are generated based on the pain level and sent to the patient.

4. The intelligent pain relief system based on biofeedback after mastectomy as described in claim 3, characterized in that, The system also includes: The patient's end is used to receive and respond to the personalized rehabilitation recommendations; The patient terminal is connected to the cloud management platform.

5. The intelligent pain relief system based on biofeedback after mastectomy as described in claim 1, characterized in that, The wearable device includes: Biosensors are used to sense and acquire multimodal signals from patients after mastectomy: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA). A processor for preprocessing the multimodal signals; A communication module is used to upload the multimodal signal to the cloud management platform; and to receive adaptive adjustment strategies dynamically issued by the cloud management platform. The controller is used to respond to the adaptive adjustment strategy by controlling the biofeedback module and / or the psychological module to initiate phantom pain intervention; The biofeedback module is used to provide biostimulation interventions to patients. The psychological module is used to provide psychological guidance and intervention for patients. Power supply, used for supplying power; The biosensor is electrically connected to the processor; The processor, communication module, biofeedback module, psychological module, and power supply are all electrically connected to the controller. The wearable device communicates with the cloud management platform via a communication module.

6. The intelligent pain relief system based on biofeedback after mastectomy for postoperative phantom pain as described in claim 5, characterized in that, The psychological module is a VR device used to play VR audio and video data in the adaptive adjustment strategy.

7. The intelligent pain relief system based on biofeedback after mastectomy as described in claim 1, characterized in that, The generation method of the AI ​​analgesia intervention model includes: We collected intervention big data from several mastectomy patients, including their multimodal signals and historical intervention protocols. The patient's multimodal signals were subjected to feature quantification analysis to assess the patient's pain level and / or psychological burden; Label the patient's pain level with corresponding biostimulation intervention strategies, and / or label the patient's psychological load with corresponding VR audio and video data; Collect characteristic data for each patient: {pain level, psychological burden, intervention plan}, and form a feature set; Dynamically match corresponding adaptive adjustment strategies based on pain level and / or psychological load. The feature set is divided into a training set and a validation set according to the proportions. Input the training set into a pre-defined machine learning model to train and learn features; The trained AI analgesia intervention model is validated using a validation set. Once the validation is successful, the AI ​​analgesia intervention model is deployed and applied.

8. An application method, said application method being implemented based on the intelligent pain relief system for post-mastectomy phantom pain based on biofeedback as described in any one of claims 1-7, characterized in that, The method includes: Activate the wearable device and establish communication with the cloud management platform; The cloud management platform creates a post-mastectomy patient file and initiates an initial calibration procedure, collecting the patient's physiological parameters at rest as a benchmark for calibration. The cloud management platform issues sampling instructions to begin real-time monitoring and collection of multimodal signals from post-mastectomy patients via wearable devices: sEMG (surface electromyography), heart rate variability (HRV), and skin conductance (EDA), and uploads them to the cloud management platform. The cloud management platform records the multimodal signals of patients after mastectomy into the patient's file, and The pre-deployed AI analgesia intervention model identifies the multimodal signals and determines the patient's pain level and / or psychological burden; The cloud management platform dynamically matches corresponding adaptive adjustment strategies based on pain level and / or psychological load and distributes them to the wearable device. The wearable device responds to the adaptive adjustment strategy issued by the cloud management platform and activates the biofeedback module and / or psychological module to intervene in phantom pain in patients after mastectomy.

9. An electronic device, characterized in that, The electronic device includes: processor; A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in claim 8.

Images